JP3575334B2 - FMCW radar equipment - Google Patents

Description

また、スペクトルピークが複素ベクトルで表現される場合には、スペクトルピークの振幅が複素ベクトルの絶対値（長さ）で示され、位相が回転角で示されるので、この複素ベクトルから評価ベクトルを求めてその絶対値を算出してもよい。
【００３６】
（８）請求項８の発明は、真の周波数シフト量を決定する手法を例示したものである。
まず、例えば下記式（Ｇ）を用いて、各スペクトルピークに対する評価ベクトルの絶対値の近傍和を求める。ここで、近傍和を用いるのは、各スペクトルピークの一点だけで評価するよりもその近傍の周波数を含めて評価した方が、評価の精度が高くなるからである。
【００３７】

Ｐ；（何番目のピークかを示す）ピーク周波数番号、ｎ；近傍の幅
次に、前記各ピーク近傍に関する近傍和Ｓｕｍ２を求めた後に、例えば下記式（Ｈ）の様に、各近傍和Ｓｕｍ２を合計して、各周波数シフト量毎のスペクトル全体和Ｓｕｍ１を求める。
【００３８】
スペクトル全体和Ｓｕｍ１＝Σ近傍和Ｓｕｍ２ …（Ｈ）
そして、例えば、この評価値の各スペクトル全体和Ｓｕｍ１（｜Ｖｐ｜）のうち、その値が最も小さい周波数シフト量を、真の周波数シフト量ＴＳｎとして決定することができる。つまり、スペクトル全体和Ｓｕｍ１が小さいほど、各評価ベクトルの絶対値が全体として小さく、よって、各スペクトルピークの一致度が大きいと考えられるからである。
【００３９】
（９）請求項９の発明は、真の周波数シフト量が決定した後に、真の周波数シフト量に対応した上りスペクトル及び下りスペクトルにおいて、その各一対のスペクトルピーク毎に、どの様に物標の静止物判定を行うかを例示したものである。
【００４０】
ここでは、ある評価ベクトルの絶対値の近傍和を所定の閾値と比較し、その近傍和が閾値以下の場合には、振幅評価値が小さく且つ位相差評価値が小さく、同じ静止物によるスペクトルピークの一致であると考えられるので、物標が静止物であると判断している。
【００４１】
（１０）請求項１０の発明では、ビームの向きを考慮する。
つまり、レーダのビームが車両の進行方向以外を向く場合には、静止物の移動方向とドップラ効果によって相対速度を検出可能なビーム方向とのズレが生じ、静止物の判定に誤差が生じる。
【００４２】
そこで、本発明では、ビームの向きを考慮して周波数シフト量を設定しているので、より正確に静止物の判定を行うことができる。
例えばビームの方向が車両の進行方向からθだけ側方にズレていた場合には、このビームによって測定した物標の相対速度にｃｏｓθかけた値を、真の物標の相対速度と見なすことができる。
【００４３】
（１１）請求項１１の発明では、移動物を考慮する。
ＦＭＣＷレーダを車載レーダとして用いた場合、ガードレール等の路側物のような静止物や前方を走行中の車両である移動物が混在してスペクトルに表れ、自車と静止物の関係によっては、静止物と移動物のスペクトルピークが合成され、静止物判定を正確に行えない場合がある。
【００４４】
そこで、本発明では、既に移動物と認識されている物標に対して、今回の移動位置を予測して移動物予測フラグを設定し、今回の判定対象のスペクトルピークに対して移動物予測フラグがセットされている場合には、合成されたスペクトルピークであると見なして、そのスペクトルピークを静止物とは判定しないようにしている。
【００４５】
これにより、路側物が多く存在するような市街地を走行中も、移動物ピークと静止物ピークが合成された合成ピークを静止物と認識することがなく、より正確な静止物判定が可能となる。
（１２）請求項１２の発明は、ＦＭＣＷレーダ装置の前提となる基本構成に関しては、前記請求項１の発明と同様であり、ここでは、特に、前記請求項１０の発明の様に、ビームの向きを考慮して周波数シフト量を設定している。これによって、より正確に静止物判定を行うことができる。
【００４６】
つまり、図５に示す様に、車両の進行方向と静止物（路側物）との間にズレがあると、車両と静止物との間の正確な相対速度にズレが生じ、それが周波数シフト量の算出に影響を及ぼすので、ここでは、ビームの方向を加味することにより、その影響を低減するものである。
【００４７】
（１３）請求項１３の発明は、請求項４の発明と同様な作用効果を奏する。
（１４）請求項１４の発明は、請求項５の発明と同様な作用効果を奏する。
（１５）請求項１５の発明は、請求項６の発明と同様な作用効果を奏する。
（１６）請求項１６の発明は、請求項７の発明と同様な作用効果を奏する。
【００４８】
（１７）請求項１７の発明は、請求項９の発明と同様な作用効果を奏する。
【００６１】
【発明の実施の形態】
以下に、本発明のＦＭＣＷレーダ装置の実施の形態の例（実施例）を図面と共に説明する。
（実施例１）
ａ）図１は、本発明が適用された実施例の障害物検出用のＦＭＣＷレーダ装置（以下単にレーダ装置と記す）の全体構成を表すブロック図である。尚、本実施例のレーダ装置は、いわゆる位相差モノパルスレーダ装置である。
【００６２】
図１に示すように、本実施例のレーダ装置２は、変調信号Ｓｍに応じて所定の周波数に変調されたレーダ波を送信する送信器１２、送信器１２から放射され、障害物に反射されたレーダ波を受信する一対の受信器１４，１６からなる送受信部１０と、送信器１２に変調信号Ｓｍを供給すると共に、受信器１４，１６から出力される中間周波のビート信号Ｂ１，Ｂ２に基づき、障害物を検出し且つ静止物を判定するための処理を実行する信号処理部２０とにより構成されている。
【００６３】
ここで、送信器１２が本発明の送信手段、受信器１４，１６が受信手段、信号処理部２０が他の各手段（スペクトル作成手段及び検出手段等）に相当する。
そして、本実施例では、当該レーダ装置２により自動車前方の障害物を検出するために、送受信部１０が自動車の前面に取り付けられ、信号処理部２０が、車室内又は車室近傍の所定位置に取り付けられている。
【００６４】
ここで、まず送信器１２は、送信信号として、ミリ波帯の高周波信号を生成する電圧制御発振器（ＶＣＯ）１２ｂと、変調信号Ｓｍを電圧制御発振器１２ｂの調整レベルに変換して電圧制御発振器１２ｂに供給する変調器（ＭＯＤ）１２ａと、電圧制御発振器１２ｂからの送信信号を電力分配して各受信器１４，１６に供給されるローカル信号を生成する電力分配器（ＣＯＵＰ）１２ｃ，１２ｄと、送信信号に応じてレーダ波を放射する送信アンテナ１２ｅとにより構成されている。
【００６５】
また、受信器１４は、レーダ波を受信する受信アンテナ１４ａと、受信アンテナ１４ａからの受信信号に電力分配器１２ｄからのローカル信号を混合するミキサ１４ｂと、ミキサ１４ｂの出力を増幅する前置増幅器１４ｃと、前置増幅器１４ｃの出力から不要な高周波成分を除去し、送信信号及び受信信号の周波数の差成分であるビート信号Ｂ１を抽出するローパスフィルタ１４ｄと、ビート信号Ｂ１を必要な信号レベルに増幅する後置増幅器１４ｅと、により構成されている。
【００６６】
なお、受信器１６は、受信器１４と全く同様の構成（１４ａ〜１４ｅが１６ａ〜１６ｅに対応）をしており、電力分配器１２ｃからローカル信号の供給を受け、ビート信号Ｂ２を出力する。そして、受信器１４を受信チャネルＣＨ１、受信器１６を受信チャネルＣＨ２と呼ぶ。
【００６７】
一方、信号処理部２０は、起動信号Ｃ１により起動され、三角波状の変調信号Ｓｍを発生する三角波発生器２２と、起動信号Ｃ２により起動され、受信器１４，１６からのビート信号Ｂ１，Ｂ２をデジタルデータＤ１，Ｄ２に変換するＡ／Ｄ変換器２４ａ，２４ｂと、ＣＰＵ２６ａ，ＲＯＭ２６ｂ，ＲＡＭ２６ｃを中心に構成され、起動信号Ｃ１，Ｃ２を送出して三角波発生器２２及びＡ／Ｄ変換器２４ａ，２４ｂを動作させる。それと共に、Ａ／Ｄ変換器２４ａ，２４ｂを介して得られるデジタルデータＤ１，Ｄ２に基づき、障害物との距離、相対速度、及び障害物の方位の検出を行い且つ静止物の判定を行う障害物検出処理（後述する）を実行する周知のマイクロコンピュータ２６と、マイクロコンピュータ２６の指令に基づき高速フーリエ変換（ＦＦＴ）の演算を実行する演算処理装置２８と、により構成されている。
【００６８】
なお、Ａ／Ｄ変換器２４ａ，２４ｂは、起動信号Ｃ２により動作を開始すると、所定時間間隔毎にビート信号Ｂ１，Ｂ２をＡ／Ｄ変換して、ＲＡＭ２６ｃの所定領域に書き込むと共に、所定回数のＡ／Ｄ変換を終了すると、ＲＡＭ２６ｃ上に設定された終了フラグ（図示せず）をセットして、動作を停止するように構成されている。
【００６９】
そして、起動信号Ｃ１により、三角波発生器２２が起動され、変調器１２ａを介して電圧制御発振器１２ｂに変調信号Ｓｍが入力されると、電圧制御発振器１２ｂは、変調信号Ｓｍの三角波状の波形の上り勾配に応じて所定の割合で周波数が増大（以後、この区間を上昇部と呼ぶ）し、それに引き続く下り勾配に応じて周波数が減少（以後、この区間を下降部と呼ぶ）するように変調された送信信号を出力する。
【００７０】
図２は、送信信号の変調状態を表す説明図である。図２に示すように、変調信号Ｓｍにより、送信信号の周波数は、１／ｆｍの期間に△Ｆだけ増減するように変調され、その変化の中心周波数はｆ０である。なお、１００ｍｓ間隔で周波数が変調されているのは、後述する障害物検出処理が１００ｍｓ周期で実行され、その処理の中で起動信号Ｃ１が生成されるからである。
【００７１】
この送信信号に応じたレーダ波が送信器１２から送出され、障害物に反射したレーダ波が、受信器１４，１６にて受信される。そして、受信器１４，１６では、受信アンテナ１４ａ，１６ａから出力される受信信号と、送信器１２からの送信信号とが混合されることにより、ビート信号Ｂ１，Ｂ２が生成される。なお、受信信号は、レーダ波が障害物まで間を往復する時間だけ送信信号に対して遅延し、且つ、障害物との間に相対速度がある場合には、これに応じてドップラシフトを受ける。このため、ビート信号Ｂ１，Ｂ２は、この遅延成分ｆｒとドップラ成分ｆｄとを含んだもの（図１７参照）となる。
【００７２】
そして、図３に示すように、Ａ／Ｄ変換器２４ａによりビート信号Ｂ１をＡ／Ｄ変換してなるデジタルデータＤ１は、ＲＡＭ２６ｃ上のデータブロックＤＢ１，ＤＢ２に順次格納され、一方、Ａ／Ｄ変換器２４ｂによりビート信号Ｂ２をＡ／Ｄ変換してなるデジタルデータＤ２は、同様に、データブロックＤＢ３，ＤＢ４に格納される。ところで、Ａ／Ｄ変換器２４ａ，２４ｂは、三角波発生器２２の起動と共に起動され、変調信号Ｓｍが出力されている間に、所定回数のＡ／Ｄ変換を行うようにされているため、前半数のデータが格納されるデータブロックＤＢ１，ＤＢ３には、送信信号の上昇部に対応した上昇部データが格納され、後半数のデータが格納されるデータブロックＤＢ２，ＤＢ４には、送信信号の下降部に対応した下降部データが格納されることになる。
【００７３】
このようにして各データブロックＤＢ１〜ＤＢ４に格納されたデータは、マイクロコンピュータ２６及び演算処理装置２８にて処理され、障害物及び静止物の検出のために使用される。
ｂ）次に、マイクロコンピュータ２６にて実行される障害物検出処理を、図４のフローチャートを参照して説明する。なお、この障害物検出処理は、１００ｍｓ周期で起動される。
【００７４】
図４に示すように、本処理が起動されると、まず、ステップ１１０にて、起動信号Ｃ１を出力して三角波発生器２２を起動し、続くステップ１２０にて、ＲＡＭ２６ｃ上の終了フラグをクリアすると共に、起動信号Ｃ２を出力してＡ／Ｄ変換器２４ａ，２４ｂを起動する。
【００７５】
これにより、三角波発生器２２からの変調信号Ｓｍを受けた送信器１２により、周波数変調されたレーダ波が送信されると共に、障害物により反射したレーダ波を受信することにより受信器１４，１６から出力されるビート信号Ｂ１，Ｂ２が、Ａ／Ｄ変換器２４ａ，２４ｂを介してデジタルデータＤ１，Ｄ２に変換されＲＡＭ２６ｃに書き込まれる。
【００７６】
続くステップ１３０では、ＲＡＭ２６ｃ上の終了フラグを調べることにより、Ａ／Ｄ変換が終了したか否かを判断する。そして、終了フラグがセットされていなければ、Ａ／Ｄ変換は終了していないものとして、同ステップ１３０を繰り返し実行することで待機し、一方、終了フラグがセットされていれば、Ａ／Ｄ変換は終了したものとしてステップ１４０に移行する。
【００７７】
ステップ１４０では、ＲＡＭ２６ｃ上のデータブロックＤＢ１〜ＤＢ４のいずれか一つを順次選択し、そのデータブロックＤＢｉ（ｉ＝１〜４）のデータを演算処理装置２８に入力してＦＦＴの演算を実行させる。なお、演算処理装置２８に入力されるデータは、ＦＦＴの演算により表れるサイドローブを抑制するために、ハニング窓や三角窓等を用いた周知のウィンドウ処理が施される。そして、この演算結果として、各周波数毎の複素ベクトルが得られる。
【００７８】
ステップ１５０では、複素ベクトルの絶対値、即ちその複素ベクトルが示す周波数成分の振幅に基づき、周波数スペクトル上でピーク（スペクトルピーク）の頂点となる全ての周波数成分（以下ピーク周波数成分と呼ぶ）を検出して、その周波数をピーク周波数として特定し、ステップ１６０に進む。なお、スペクトルピークの頂点の検出方法としては、例えば、周波数に対する振幅の変化量を順次求め、その前後にて変化量の符号が反転する周波数にスペクトルピークの頂点があるものとして、その周波数を特定すればよい。
【００７９】
ステップ１６０では、ステップ１５０にて特定されたピーク周波数成分の位相を算出する。この位相は、複素ベクトルが実数軸となす角度に等しく、複素ベクトルから簡単に求められる。
続くステップ１７０では、未処理のデータブロックＤＢｉがあるか否かを判断し、未処理のものがあれば、ステップ１４０に戻って、その未処理のデータブロックＤＢｉについて、ステップ１４０〜１６０の処理を実行し、一方、未処理のものがなければ、ステップ１７５に移行する。
【００８０】
ステップ１７５では、後に詳述する様に、検出した障害物が静止物かどうかを判定する静止物判定処理を行う。
続くステップ１８０では、ピーク周波数成分の振幅、即ちパワーを夫々比較することにより、上昇部と下降部とで同じパワーを有するものを、同一障害物からの反射波に基づくピーク周波数成分のペアとして特定するペアリング処理を実行する。
【００８１】
但し、ここでは、前記ステップ１７５の静止物判定処理の際に、明らかにペアでないと判定されたピーク同士は、ペアリング処理を行わない。
尚、このペアリング処理は、例えば特願平８−１７９２２７号の図７及びその説明等に示す処理と同様であるので、その詳しい説明は省略する。
【００８２】
続くステップ１９０では、ステップ１８０にてペアリングされたピーク周波数成分を用いて、障害物との距離，相対速度、及び障害物の方位を算出する距離・速度方位算出処理を実行して本処理を終了する。
例えば上昇部及び下降部毎に、各受信チャンネルＣＨ１，ＣＨ２間で位相差を算出し、その位相差の符号が等しくない場合には、下記式（１），（２）を用いて障害物との距離Ｄ及び相対速度Ｖを算出する。
【００８３】
Ｖ＝（Ｃ／（４＊ｆ０））＊（ｆｂ２−ｆｂ１） …（１）
Ｄ＝（Ｃ／（８＊△Ｆ＊ｆｍ））＊（ｆｂ１＋ｆｂ２） …（２）
但し、△Ｆは送信信号の周波数変位幅（周波数変位幅）、ｆ０は送信信号の中心周波数、１／ｆｍは１周期の変調に要する時間（即ちｆｍは三角波の繰り返し周波数）、Ｃは光速、ｆｂ１は上昇部のビート周波数（上りビート周波数）、ｆｂ２は下降部のビート周波数（下りビート周波数）、Ｃは光速を表す。
【００８４】
尚、この距離・速度・方位算出処理は、例えば特願平８−１７９２２７号の図５及びその説明等に示す処理と同様であるので、その詳しい説明は省略する。
ｃ）次に、前記ステップ１７５にて行われる静止物判定処理の基本原理について説明する。
【００８５】
ここでは、静止物の判定には、まず、車速センサ等の誤差を考慮して複数の周波数シフト量（Ｓｎ−１、Ｓｎ、Ｓｎ＋１）を設定し、評価関数を用いて、その中から真の周波数シフト量ＴＳｎを求める。そして、真の周波数シフト量ＴＳｎに対応した上り及び下りスペクトルを用い、そのピーク周波数成分に対応した物標が、移動物か静止物かを判定する。以下詳細に説明する。
【００８６】
（ｉ）まず、周波数シフト量（以下単にシフト量とも記す）の算出の手順▲１▼〜▲３▼を説明する。
▲１▼まず、自車速度ＶＢ等を用い、基本周波数シフト量（基本シフト量）を算出する基本周波数シフト量演算式を設定する。
【００８７】
つまり、前記式（１）を変形して、静止物の判定のために、スペクトルをどれだけずらすかを決めるための基本量（基本シフト量＝（ｆｂ２−ｆｂ１））を算出する式（３）を設定する。
基本シフト量＝（ｆｂ２−ｆｂ１）＝（４＊ＶＢ＊ｆ０）／Ｃ …（３）
但し、ｆｂ１は上りビート周波数、ｆｂ２は下りビート周波数、ＶＢは自車速度、ｆ０は送信信号の中心周波数、Ｃは光速を表す。
【００８８】
このとき、車速センサの誤差が既知、或は学習済みの場合は、補正係数、マップ演算等によって補正した自車速度ＶＢを利用する。
▲２▼次に、レーザ装置１２のビーム角度に応じた補正を行う。
図６に示す様に、ビームステア、スキャンセンサの場合のセンサ正面方向、あるいは進行方向からのビーム角度をθとすると、ドップラ効果を利用して相対速度を検出するレーダの場合、検出可能な速度成分は、ビーム方向に等しい速度成分である（−ＶＢ＊ＣＯＳθ）であるために、補正を行う必要がある。この成分は、ビームをステア、スキャンニングする角度が大きくなればなるほど小さくなり、真の移動速度（接近する速度；−ＶＢ）とのずれが大きくなり、正確な周波数シフトが行えなくなる。
【００８９】
従って、ここでは、基本シフト量に角度補正係数（ＣＯＳθ）を加味した第１補正のための下記式（４）を設定する。これにより、θが大きくなるほど、基本シフト量は小さく補正される。
第１補正後シフト量＝（４＊ＣＯＳ（θ）＊ＶＢ＊ｆ０）／Ｃ …（４）
▲３▼次に、車速センサの応答遅れを加味した補正を行う。
【００９０】
車速センサは、一般的に、駆動系、車輪系からのパルス信号の時間間隔を測定し、その値より実際の車速を検出する。しかし、実際には、安定性、ノイズ等により、時間的にフィルタリングされているため、実際の車速との間には応答遅れにがある。例えば時速１００ｋｍで走行するような状況では、その遅れは気にならないが、加速、減速時には、時速数キロの時間遅れを生ずる。そこで、本実施例では、その遅れを考慮して、許容値を持った処理を行う。
【００９１】
具体的には、実際の車両の種類によって時間遅れ幅は異なるので、予め車両毎に基本遅れ幅を持たせることが可能である。これは、実車速、車速センサのフィルタの時定数等によって異なるが、ここでは、速度遅れ値Ｄｖとして設定する。従って、この値を用いて第２補正のための下記式（５）を設定する。
【００９２】

ここで、速度遅れ値Ｄｖの値としては、車速センサの分解能を考慮した値とする。例えば車速センサが±５ｋｍ／ｈの誤差が生じる場合には、Ｄｖ＝（−５，０，＋５）の様に例えば３通りに設定することが可能である。
【００９３】
つまり、実際には速度遅れ値Ｄｖの幅だけ車速センサの時間遅れが生じる可能性があるので、前記式（５）により、基本シフト量にある幅を持たせることができる。即ち、後に詳述する様に、周波数シフト量にある幅を持たせて複数の周波シフト量を設定し、その中で一番マッチした周波数シフト量を選択することにより、真の車速に基づいた周波数シフト量を設定することができる。
【００９４】
尚、処理を簡素化するために、前記▲２▼の角度成分の影響を吸収するような値にＤｖを設定することも可能である。
（ｉｉ）次に、評価関数による評価方法について説明する。
ここでは、前記式（５）で求めた複数の周波数シフト量を用いて、各々下降部のスペクトルをシフトし、上昇部のスペクトルとの一致度の比較を行う。
【００９５】
▲１▼従来では、一意に決定された周波数シフト量を用いて、上り及び下りスペクトルの対応するスペクトルピークのピーク周波数成分の減算のみを行っていたが、本実施例では、複数の周波数シフト量（従ってシフトする際のシフト幅）の中から最適な周波数シフト量を求めるために、下記の評価関数を用いる。
【００９６】
この評価関数は、下記の式（６），（７）に示す様に、スペクトルピークの振幅だけでなく、位相差モノパルスレーダで得た方位情報を示す位相差も用いる。尚、位相差は、２系統の受信系を持つモノパルスレーダで受信したそれぞれの位相情報を減算した値であり、この位相差を利用して対象物の方位を求める方式が位相差モノパルスレーダである。
【００９７】

尚、位相差モノパルスレーダの場合、その構成上、上昇部、下降部では符号が逆転するので、その和が０ならば、一致していることになる。
【００９８】
そして、図６に示す様に、前記振幅評価値Ｙと位相差評価値Ｘとを持つ評価ベクトルＶｐの長さ｜Ｖｐ｜を評価値とする。
次に、この評価値｜Ｖｐ｜を各ピーク周波数成分毎に求めてその合計を求めるのであるが、その場合には、下記式（８）に示す様に、目的とするピーク周波数成分だけでなく、その近傍の周波数に関しても、同様に評価値｜Ｖｐ｜を求めて、それらの和（近傍和Ｓｕｍ２）を求める。尚、近傍和Ｓｕｍ２を求める範囲は、図８に示す様に、一点鎖線を中心にして左右の破線で挟まれた帯状の範囲であり、ＦＦＴの分解能により変化する。
【００９９】
ここで、近傍和Ｓｕｍ２を求めるのは、単一のピーク周波数成分を用いる場合に比べて、その精度が高いからである。

但し、Ｐ；（評価するピークの順番を示す）ピーク周波数番号、
ｎ；近傍の幅（近傍をｎ個に区分した場合）
また、この場合、全てのピーク周波数成分に関してその近傍和を求めるのではなく、静止物及び移動物の判定を行うピーク周波数成分のみに対してその近傍和Ｓｕｍ２を求める。これは、全てに対して処理を行うと、ノイズ、クラッタ等のピークにより、正しい結果が検出されない場合があるばかりか、演算時間が大量に必要となるからである。
【０１００】
ここで、例えば図７に示す様に、上昇部（上りスペクトル）に存在するスペクトルピークを、Ｐｕ１、Ｐｕ２、Ｐｕ３、Ｐｕ４、下降部（下りスペクトル）に存在するスペクトルピークを、Ｐｄ１、Ｐｄ２、Ｐｄ３、Ｐｄ４（但し、１，２，３は静止物、４は移動物のスペクトルピークとする）とした場合を考える。
【０１０１】
この場合、下りスペクトルを複数の周波数シフト量だけシフトさせると、図８に示す様になる。この図８では、基本シフト量Ｓｎに対して、Ｓｎ±１の３種の周波数シフト量とした。尚、周波数シフト量Ｓｎに関して±１の値の意味は、シフト量を演算する際の車速センサの応答遅れ、誤差等を考慮した幅であり、説明の便宜上、±１の幅を用いた。
【０１０２】
そして、各々の周波数シフト量に対応した上り及び下りスペクトルに対して、即ち上りと下りの両スペクトルの間で対応する一対のスペクトルピークに対して、各々各ピーク周波数成分の近傍において近傍和Ｓｕｍ２を算出するのである。▲２▼次に、前記各スペクトルピークに関する近傍和Ｓｕｍ２を求めた後に、下記式（９）の様に、各近傍和Ｓｕｍ２を合計して、各周波数シフト量毎のスペクトル全体和Ｓｕｍ１を求める。
【０１０３】
スペクトル全体和Ｓｕｍ１＝Σ近傍和Ｓｕｍ２ …（９）
そして、この各スペクトル全体和Ｓｕｍ１（｜Ｖｐ｜）が最も小さい周波数シフト量を、真の周波数シフト量ＴＳｎとする。従って、例えば図８では、中央のスペクトルに対応した周波数シフト量Ｓｎが、真の周波数シフト量ＴＳｎとして選択される。
【０１０４】
この様に、スペクトル全体和Ｓｕｍ１（｜Ｖｐ｜）が最小のものを真の周波数シフト量ＴＳｎとする理由は、スペクトル全体和Ｓｕｍ１が小さいほど、各評価ベクトルの絶対値が全体として小さく、よって、各スペクトルピークの一致度が大きいと考えられるからである。
【０１０５】
尚、ここでは、シフト幅として３種を例に挙げたが、実際には、第２補正後シフト量を示す前記式（３）で求められる幅に基づいて、真の周波数シフト量ＴＳｎが決定される。
また、一旦真の周波数シフト量ＴＳｎが求められた後は、次回の演算の際に、この真の周波数シフト量を加味して周波数シフト量の演算を行ってもよい。
【０１０６】
（ｉｉｉ）次に、真の周波数シフト量ＴＳｎを用いた移動物、静止物の分離方法について説明する。
先の（ｉ）、（ｉｉ）の演算において、車速センサの遅れ、誤差、ビーム向き等の影響を除外し、真の周波数シフト量ＴＳｎを求めた。ここでは、この周波数シフト量だけシフトしたスペクトルの一致度によって、移動物、静止物を判定する。
【０１０７】
従来は、スペクトルの減算を行うのみであったが、本実施例では、それに伴う偶然にピークレベルが等しいスペクトルピークが存在する可能性を考えて、即ち、移動物と静止物とのピークが合成されている場合を考慮して処理を行う。
具体的には、前記図８に示す様に、移動物と静止物との分離は、先の評価値の近傍和Ｓｕｍ２を用いて行う。
【０１０８】
つまり、図８に示す様に、あるスペクトルピークの近傍和Ｓｕｍ２が閾値Ｔｈｐ以下の場合には、スペクトルピークの振幅（ピーク振幅）及びビームの方位（位相差）に関して、上昇部及び下降部のスペクトルピークが一致していると見なして、一致したスペクトルピークが静止物であるとの静止物判定を行う。一方、あるスペクトルピークの近傍和Ｓｕｍ２が閾値Ｔｈｐを上回る場合には、移動物と静止物のスペクトルピークの組み合せ、又はノイズ等によるスペクトルピークの組み合せと判断して、静止物判定を行わない。尚、閾値Ｔｈｐに関しては、車両の走行状態、天候等により変更してもよく、固定値に限定されない。
【０１０９】
（ｉｖ）次に、前記移動物と静止物の合成ピーク判定について説明する。
判定対象のピークが、移動物と静止物のスペクトルピークが合成されたものであるかどうかの判定に関しては、移動物予測フラグを用いることによって行う。つまり、図９に示す様に、前回静止物と判定されなかったスペクトルピークで、別の手段により移動物と判定されたスペクトルピークに関しては、その移動物の運動状態によって、△ｔ後に出現するであろうスペクトルピークの位置（ピーク位置）が予測される。この予測位置を示しフラグを移動物予測フラグと呼ぶ。従って、仮に静止物判定された場合でも、そのピーク位置に移動物予測フラグがたっていた場合には、移動物と静止物の合成ピークであると判定し、静止物判定は行わない。尚、同様にして、順次移動物予測フラグをセットしてゆく。
【０１１０】
ｄ）次に、前記原理に基づいて行われる前記ステップ１７５の静止物判定処理について、図１０のフローチャートに基づいて説明する。
本処理は、上述した原理に基づいて、レーダ装置２により認識された障害物（物標）のスペクトルピークが、静止物に該当するものであるか否かを判定するための処理である。
【０１１１】
まず、図１０のステップ２００にて、前記▲１▼〜▲３▼の手順にて設定した前記式（５）に基づいて、基本シフト量を補正した第２補正後シフト量、即ち周波数シフト量の幅を決定する。
続くステップ２１０では、（まだ周波数シフトを実行していない）例えばシフト幅Ｓｎ−１から周波数シフトを行う。例えば下りスペクトル全体をシフトする。従って、２回目に本処理を通過した場合には、次のシフト幅の周波数シフトを行う。
【０１１２】
続くステップ２２０では、前記ステップ２１０にて周波数シフトしたスペクトルにおいて、その評価を行うべき所定のスペクトルピークに対して、前記式（６），（７）に基づいて、そのスペクトルピークの頂点のピーク周波数成分の近傍における評価値｜Ｖｐ｜を順次算出する。
【０１１３】
続くステップ２３０では、前記式（８）に基づいて、前記ステップ２２０にて算出した所定のスペクトルピークの頂点の近傍の評価値｜Ｖｐ｜を合計して、そのピークの近傍和Ｓｕｍ２を算出する。
続くステップ２４０では、判定を希望するピークの数だけ近傍和Ｓｕｍ２の算出の処理が終了したか否かを判定する。例えば図８に示す様に、例えばシフト幅Ｓｎ−１において、４つのスペクトルピークに関して、各々の近傍和Ｓｕｍ２を全て算出したか否かを判定する。ここで肯定判断されるとステップ２５０に進み、一方否定判断されると前記ステップ２２０以降の処理に戻り、他のスペクトルピークの評価値｜Ｖｐ｜及び近傍和Ｓｕｍ２の算出を行う。
【０１１４】
ステップ２５０では、前記ステップ２２０〜２４０にて、全てのスペクトルピークの近傍和Ｓｕｍ２の算出が終了したので、前記式（９）に基づいて、それらを合計して、スペクトル全体和Ｓｕｍ１の算出を行う。
続くステップ２６０では、シフト幅回数シフトしたか否かを判定する。例えば周波数シフト量が、Ｓｎ−１、Ｓｎ、Ｓｎ＋１の３通りある場合には、各々の周波数シフト量において、上述したスペクトル全体和Ｓｕｍ１等の演算が行われたか否かを判定する。ここで肯定判断されるとステップ２７０に進み、一方否定判断されるとステップ２１０以降の処理戻り、前記と同様にして他のスペクトル全体和Ｓｕｍ１の算出の処理を行う。
【０１１５】
ステップ２７０では、全て（例えば図８では３通り）のスペクトル全体和Ｓｕｍ１の値を比較し、その最も小さな値に対応する周波数シフト量を、真の周波数シフト量ＴＳｎとする。
続くステップ２８０では、真の周波数シフト量にて周波数シフトしたスペクトルに関し（図８では中央のシフト量Ｓｎのスペクトル）、所定のスペクトルピーク（詳しくは上りと下りのスペクトルで対応した一対のスペクトルピーク）の近傍和Ｓｕｍ２が、閾値Ｔｈｐ以下か否かを判定する。ここで肯定判断されるとステップ２９０に進み、一方否定判断されるとステップ３２０に進む。
【０１１６】
ステップ３２０では、スペクトルピークの近傍和Ｓｕｍ２が閾値Ｔｈｐより大きいので、即ち、静止物と判定するには評価の一致度が低いので、そのスペクトルピークに該当する物標識が移動物であるとして、移動物であることを示す移動物フラグをセットし、ステップ３３０に進む。
【０１１７】
一方、ステップ２９０では、そのピーク位置に移動物予測フラグがセットされているかどうかを判定する。ここで肯定判断されるとステップ３１０に進み、一方、否定判断されるとステップ３００に進む。
ステップ３１０では、移動物予測フラグがセットされているので、そのスペクトルピークは静止物と移動物との合成ピークであると判断して、合成ピークを示す合成ピークフラグをセットし、ステップ３３０に進む。
【０１１８】
一方、ステップ３００では、移動物予測フラグがセットされていないので、即ち近傍和Ｓｕｍ２が閾値Ｔｈｐ以下で且つ移動物予測フラグがセットされていなので、そのスペクトルピークは静止物のスペクトルピークであると判断して、静止物を示す静止物フラグをセットし、ステップ３３０に進む。
【０１１９】
ステップ３３０では、判定を希望するスペクトルピークの数だけ、前記ステップ２８０〜３１０における処理、即ちそのスペクトルピークが何を意味するのかの判定処理が終了したか否かを判定する。ここで否定判断されると前記ステップ２８０以降の処理を繰り返し、一方肯定判断されると一旦本処理を終了する。
【０１２０】
以上説明したように、本実施例のレーダ装置２においては、静止物の判定の際に、車速センサ等の誤差を考慮して複数の周波数シフト量（Ｓｎ−１、Ｓｎ、Ｓｎ＋１）を設定し、前記式（６）〜（９）からなる評価関数を用いて、その中から真の周波数シフト量ＴＳｎを求める。そして、真の周波数シフト量ＴＳｎに対応した上り及び下りスペクトルを用いて、その両スペクトルのスペクトルピークに対する評価を行って、移動物と静止物とを分離している。これにより、移動物と静止物とを正確に区別することができる。
【０１２１】
つまり、本実施例では、ビームの向きを考慮した補正を行い、更に、車速センサ等による誤差を考慮して、従来の様に、単一な周波数シフト量ではなく、複数の周波数シフト量を設定するとともに、評価値｜Ｖｐ｜から求めた近傍和Ｓｕｍ２及びスペクトル全体和Ｓｕｍ１を用いて周波数シフト量の評価を行うことにより、誤差等を排除した真の周波数シフト量ＴＳｎを決定することができる。
【０１２２】
しかも、この真の周波数シフト量ＴＳｎに対応した上り及び下りスペクトルを用い、両スペクトルにおいてその各一対のスペクトルピークの近傍和Ｓｕｍ２と閾値Ｔｈｐとを比較することにより、移動物を排除することができるので、個々の物標に対して、それが移動物か静止物かを正確に区別することができる。
【０１２３】
その上、移動物予測フラグを用いることにより、移動物と静止物との合成フラグを排除することができる。それにより、静止物のみを確実に認識することができ、その点からも静止物判定の精度が向上するという利点がある。
（実施例２）
次に、実施例２について説明するが、前記実施例１と同様な箇所の説明は省略する。
【０１２４】
前記実施例１では、スペクトルピーク（パワースペクトル）の波形に対して、シフト等の処理を行い、静止物判定を行った。しかし、上昇部及び下降部のスペクトル波形を記憶するためには、レーダ装置に波形形状を記憶するための大容量メモリが必要になる。また、周波数解析を認識処理とは別のプロセッサで行っている場合には、通信速度の関係で、波形データを全て送信することはできないことがある。
【０１２５】
この対策として、本実施例では、スペクトル波形のピーク情報、即ちスペクトルピークの頂点に関するピーク情報（頂点を示すピーク周波数、頂点における振幅（ピークレベル）、対応するスペクトルピーク同士の頂点を示す位相の差（位相差））のみを用いて、静止物判定を行った。以下、詳細に説明する。
【０１２６】
ａ）まず、本実施例の原理について説明する。
本実施例では、前記実施例１の式（５）で求めた複数の周波数シフト量を用いて、下りスペクトルをシフトし、上りスペクトルと下りスペクトルの対応するスペクトルピークの一致度の比較を行うが、ここでは、各スペクトルピークの所定幅の波形情報ではなく、図１１の太線で示す様に、各スペクトルピークの頂点に対応する値であるピーク情報のみを用いてスペクトルマッチ度の評価を行う。
【０１２７】
▲１▼具体的には、まず、前記実施例１と同様に、前記振幅評価値Ｙと位相差評価値Ｘとを持つ評価ベクトルＶｐの長さ｜Ｖｐ｜を評価値とする。
ここで、図１１に示す様に、上りスペクトルに存在するスペクトルピークを、Ｐｕ１、Ｐｕ２、Ｐｕ３、Ｐｕ４、下りスペクトルに存在するスペクトルピークを、Ｐｄ１、Ｐｄ２、Ｐｄ３、Ｐｄ４（但し、１，２，３は静止物、４は移動物のスペクトルピーク）とした場合を考える。
【０１２８】
この場合、下りスペクトルを複数の周波数シフト量だけシフトさせると、図１２に示す様になる。この図１２では、基本シフト量Ｓｎに対して、Ｓｎ±１の３種の周波数シフト量とした。
そして、各々の周波数シフト量に対応した上り及び下りスペクトルに対して、即ち上りと下りの両スペクトルの間で対応する一対のスペクトルピークに対して、各スペクトルピークの頂点に対応するピーク情報の評価値｜Ｖｐ｜を算出する。
【０１２９】
▲２▼次に、前記各評価値｜Ｖｐ｜を求めた後に、前記実施例１の様に近傍和Ｓｕｍ２を求めるのではなく、各評価値｜Ｖｐ｜を合計して、各周波数シフト量毎の評価値和ＳｕｍＶｐを求める。この評価値和ＳｕｍＶｐとは、前記実施例１のスペクトル全体和Ｓｕｍ１に対応するものである。
【０１３０】
つまり、本実施例では、近傍和Ｓｕｍ２が不要であるので、この評価値和ＳｕｍＶｐが、前記実施例１のスペクトル全体和に相当するものとなる。
そして、この各評価値和ＳｕｍＶｐが最も小さい周波数シフト量を、真の周波数シフト量ＴＳｎとする。従って、例えば図１２では、中央のスペクトルに対応した周波数シフト量Ｓｎが、真の周波数シフト量ＴＳｎとして選択される。
【０１３１】
▲３▼また、真の周波数シフト量ＴＳｎを用いた移動物、静止物の分離は、下記の様にして行う。
図１２に示す様に、あるスペクトルピークの評価値｜Ｖｐ｜が閾値ＴＨｐ１以下の場合には、ピークレベル及びビームの方位（位相差）に関して、上り及び下りスペクトルのスペクトルピークが一致していると見なして、一致したスペクトルピークが静止物であるとの静止物判定を行う。一方、あるスペクトルピークの評価値｜Ｖｐ｜が閾値ＴＨｐ１を上回る場合には、移動物と静止物のピークの組み合せ、又はノイズ等によるピークの組み合せと判断して、静止物判定を行わない。
【０１３２】
ｂ）次に、前記原理に基づいて行われる静止物判定処理について、図１３のフローチャートに基づいて説明する。
本処理は、上述した原理に基づいて、レーダ装置２により認識された障害物（物標）のスペクトルピークが、静止物に該当するものであるか否かを判定するための処理である。
【０１３３】
まず、図１３のステップ４００にて、前記実施例１と同様に、基本シフト量を補正した第２補正後シフト量、即ち周波数シフト量の幅を決定する。
続くステップ４１０では、例えばシフト幅Ｓｎ−１から順次周波数シフトする。続くステップ４２０では、前記ステップ４１０にて周波数シフトしたスペクトルにおいて、その評価を行うべき所定のスペクトルピークに対して、そのスペクトルピークの頂点のピーク情報のみを用いて、前記評価値｜Ｖｐ｜を順次算出する。
【０１３４】
続くステップ４４０では、判定を希望するスペクトルピークの数だけ評価値｜Ｖｐ｜の算出の処理が終了したか否かを判定する。ここで肯定判断されるとステップ４５０に進み、一方否定判断されると前記ステップ４２０に戻る。
続くステップ４５０では、前記ステップ４２０にて算出した所定のスペクトルピークの評価値｜Ｖｐ｜を合計して、その評価値和ＳｕｍＶｐを算出する。
【０１３５】
続くステップ４６０では、シフト幅回数シフトしたか否かを判定する。ここで肯定判断されるとステップ４７０に進み、一方否定判断されるとステップ４１０に戻る。
ステップ４７０では、全ての評価値和ＳｕｍＶｐの値を比較し、その最も小さな値に対応する周波数シフト量を、真の周波数シフト量ＴＳｎとする。
【０１３６】
続くステップ４８０では、真の周波数シフト量ＴＳｎにて周波数シフトしたスペクトルに関し（図１２では中央のシフト量Ｓｎのスペクトル）、所定のピークの評価値｜Ｖｐ｜が、閾値ＴＨｐ１以下か否かを判定する。ここで肯定判断されるとステップ４９０に進み、一方否定判断されるとステップ５２０に進む。
【０１３７】
ステップ５２０では、スペクトルピークの評価値｜Ｖｐ｜が閾値ＴＨｐ１より大きいので、移動物であることを示す移動物フラグをセットし、ステップ５３０に進む。
一方、ステップ４９０では、そのスペクトルピークの位置に移動物予測フラグがセットされているかどうかを判定する。ここで肯定判断されるとステップ５１０に進み、一方、否定判断されるとステップ５００に進む。
【０１３８】
ステップ５１０では、移動物予測フラグがセットされているので、合成ピークを示す合成ピークフラグをセットし、ステップ５３０に進む。
一方、ステップ５００では、移動物予測フラグがセットされていないので、静止物を示す静止物フラグをセットし、ステップ５３０に進む。
【０１３９】
ステップ５３０では、判定を希望するスペクトルピーク全ての処理が終了したか否かを判定する。ここで否定判断されると前記ステップ４８０に戻り、一方肯定判断されると一旦本処理を終了する。
この様に、本実施例では、前記実施例１の様に、（スペクトルピークの頂点近傍等の）波形情報を用いるのではなく、頂点のピーク情報のみを利用して静止物判定を行っている。
【０１４０】
そのため、演算速度が速いという効果がある。また、大容量メモリが不要であり、コストを低減することができる。更に、周波数解析を認識処理とは別のプロセッサで行っている場合でも、ピーク情報のみを送信すればよく、通信速度の影響を受け難いという利点がある。
（実施例３）
次に、実施例３について説明するが、前記実施例１，２と同様な箇所の説明は省略する。
【０１４１】
前記実施例２では、スペクトルピークの頂点のピーク情報のみを用いて静止物判定を実施したが、例えば図１４に示す様に、連続したガードレールからのスペクトルのように、ピーク形状が急峻でなく、なだらかで広い幅を持つような形状のものに関しては、スペクトルピークの頂点が微妙に変動し上昇部と下降部でスペクトルピークの頂点が一致しない場合（▲１▼と▲２▼の不一致）が存在する。
【０１４２】
この対策として、本実施例では、周波数シフトした後に、基準となる上りスペクトルのスペクトルピークの頂点のピーク周波数を中心として、所定の幅Ｐｗ以内に存在する下りスペクトルのスペクトルピークの頂点を、評価値｜Ｖｐ｜を算出する対象とすることにより、静止物判定を行った。以下、詳細に説明する。
【０１４３】
ａ）まず、本実施例の原理について説明する。
具体的には、まず、前記実施例１と同様に、前記振幅評価値Ｙと位相差評価値Ｘとを持つ評価ベクトルＶｐの長さ｜Ｖｐ｜を評価値とする。
ここで、図１４に示す様に、上りスペクトルに存在するスペクトルピークを、Ｐｕ１、Ｐｕ２、Ｐｕ３、Ｐｕ４、下りスペクトルに存在するスペクトルピークを、Ｐｄ１、Ｐｄ２、Ｐｄ３、Ｐｄ４（但し、１，２，３は静止物、４は移動物のスペクトルピーク）とした場合を考える。
【０１４４】
この場合、下りスペクトルを基本シフト量Ｓｎだけシフトさせると、図１５（ａ）に示す様になる。
そして、上りと下りの両スペクトルの間で対応する一対のスペクトルピークに対して、各スペクトルピークの頂点に対応するピーク情報の評価値｜Ｖｐ｜を算出するのであるが、ここでは、基準となる上りスペクトルの各スペクトルピークのピーク周波数を中心として、所定の幅Ｐｗ以内に、下りスペクトルの対応するスペクトルピークの頂点が存在するかどうかのチェックを行った。
具体的には、上りスペクトルの所定のスペクトルピークの頂点のピーク周波数に着目し、下りスペクトルの対応するスペクトルピークにおいて、前記ピーク周波数から±Ｐｗ／２の範囲にある振幅値を、その幅Ｐｗを小さく区切って順次調べる。この場合、頂点のピーク情報であるピークレベルは記憶されているが、それ以外の位置の波形の振幅値は「０」である。
【０１４５】
従って、例えばＡピークの様に、幅Ｐｗ以内に、上昇部のＡピークの頂点に対応する下降部のＡピークの頂点が存在する場合、その頂点同士の評価値｜Ｖｐ｜は、頂点同士でない場合の評価値｜Ｖｐ｜よりも小さい。よって、ここでは、評価値｜Ｖｐ｜を算出して、その最小値を求める。即ち、本実施例の評価値｜Ｖｐ｜の最小値が、前記実施例２の（理想状態における）評価値｜Ｖｐ｜に対応したものである。
【０１４６】
尚、移動物、静止物の分離は、前記静止物判定を行わないもの以外は、前記実施例２と同様である。
ｂ）次に、前記原理に基づいて行われる静止物判定処理について、図１６のフローチャートに基づいて説明する。
【０１４７】
まず、図１６のステップ６００にて、前記実施例１，２と同様に、基本シフト量を補正した第２補正後シフト量、即ち周波数シフト量の幅を決定する。
続くステップ６１０では、基本シフト量Ｓｎだけ周波数シフトする。
続くステップ６２０では、前記ステップ６１０にて周波数シフトしたスペクトルにおいて、その評価を行うべき所定のスペクトルピークに対して、その頂点のピーク情報を用いて前記評価値｜Ｖｐ｜を算出するのであるが、このとき、上述した様に、所定幅Ｐｗにおける評価値｜Ｖｐ｜の最小値を求める。
【０１４８】
続くステップ６３０では、判定を希望するスペクトルピークの数だけ評価値｜Ｖｐ｜の最小値の算出の処理が終了したか否かを判定する。ここで肯定判断されるとステップ６４０に進み、一方否定判断されると前記ステップ６２０に戻る。ステップ６４０では、スペクトルピークの評価値｜Ｖｐ｜の最小値が、閾値ＴＨｐ２以下か否かを判定する。ここで肯定判断されるとステップ６５０に進み、一方否定判断されるとステップ６８０に進む。
【０１４９】
ステップ６８０では、スペクトルピークの評価値｜Ｖｐ｜の最小値が閾値ＴＨｐ２より大きいので、移動物であることを示す移動物フラグをセットし、ステップ６９０に進む。
一方、ステップ６５０では、そのスペクトルピークの位置に移動物予測フラグがセットされているかどうかを判定する。ここで肯定判断されるとステップ６７０に進み、一方、否定判断されるとステップ６６０に進む。
【０１５０】
ステップ６７０では、移動物予測フラグがセットされているので、合成ピークを示す合成ピークフラグをセットし、ステップ６９０に進む。
一方、ステップ６６０では、移動物予測フラグがセットされていないので、静止物を示す静止物フラグをセットし、ステップ６９０に進む。
【０１５１】
ステップ６９０では、判定を希望するスペクトルピーク全ての処理が終了したか否かを判定する。ここで否定判断されると前記ステップ６４０に戻り、一方肯定判断されると一旦本処理を終了する。
この様に、本実施例では、スペクトルピークの頂点のピーク情報のみを利用して静止物判定を行っているので、前記実施例２と同様な効果を奏する。
【０１５２】
更に、本実施例では、上昇部のスペクトルピークの頂点に対して、その所定幅Ｐｗにおける下降部のスペクトルピークの頂点があるか否かをチェックし、評価値｜Ｖｐ｜の最小値を求め、評価値｜Ｖｐ｜の最小値を用いて静止物判定を行っている。
【０１５３】
そのため、例えばガードレールの様に、スペクトルピークの頂点が変動する可能性がある場合でも、より確実に静止物判定を行うことができるという利点がある。
尚、静止物判定を実施しないものに関しては、別の手段（例えばペアシフトによる処理）を行って、別途静止物判定を行うことができる。
（実施例４）
次に、実施例４について説明するが、前記実施例１〜３と同様な箇所の説明は省略する。
【０１５４】
本実施例は、基本的には、前記実施例３と同様な処理を行うが、評価値｜Ｖｐ｜の最小値を求める手法が異なるので、異なる点のみを説明する。
つまり、本実施例でも、上りと下りの両スペクトルの間で対応する一対のスペクトルピークに対して、各スペクトルピークの頂点に対応するピーク情報の評価値｜Ｖｐ｜を算出するのであるが、ここでは、下りスペクトル全体を一括してシフトするのではなく、下りスペクトルの個々のスペクトルピークを、周波数シフト量（例えばＳｎ）±Ｐｗ／２の範囲でシフトして、スペクトルマッチ度を評価する。
【０１５５】
具体的には、図１５（ｂ）に示す様に、下降部の例えばＣピークを、Ｓｎ−Ｐｗ／２〜Ｓｎ＋Ｐｗ／２の範囲でずらし、その範囲内でずらした場合において、前記実施例３と同様にして、所定幅Ｐｗにおける評価値｜Ｖｐ｜の最小値を求める処理を行う。尚、他のスペクトルピークも同様にずらして、スペクトルマッチ度の評価を行う。
【０１５６】
本実施例においても、前記実施例３と同様な効果を奏する。
尚、本発明は前記実施例に何ら限定されることなく、本発明の技術的範囲を逸脱しない限り、種々の態様で実施できることはいうまでもない。
また、前記実施例では、ＦＭＣＷレーダ装置について述べたが、この装置による制御を実行させる手段を記憶している記録媒体にも、本発明は適用できる。
【０１５７】
例えば記録媒体としては、マイクロコンピュータとして構成される電子制御装置、マイクロチップ、フロッピィディスク、ハードディスク、光ディスク等の各種の記録媒体が挙げられる。
つまり、上述したＦＭＣＷレーダー装置の制御を実行させることができる例えばプログラム等の手段を記憶したものであれば、特に限定はない。
【図面の簡単な説明】
【図１】実施例１のレーダ装置の全体構成を表すブロック図である。
【図２】送信信号の周波数の変化を表すグラフである。
【図３】ＲＡＭに格納されるデータを表す説明図である。
【図４】障害物検出処理を表すフローチャートである。
【図５】レーザレーダのビームの向きによる補正を示す説明図である。
【図６】評価値を示す説明図である。
【図７】上昇部と下降部のスペクトルを示す説明図である。
【図８】３種のシフト幅にてシフトした場合のスペクトルを示す説明図である。
【図９】移動物予測フラグのセット方法を示す説明図である。
【図１０】実施例１の静止物判定処理を示すフローチャートである。
【図１１】実施例２の上昇部と下降部のスペクトルを示す説明図である。
【図１２】３種のシフト幅にてシフトした場合のスペクトルを示す説明図である。
【図１３】実施例２の静止物判定処理を示すフローチャートである。
【図１４】実施例３の上昇部と下降部のスペクトルを示す説明図である。
【図１５】基本シフト幅にてシフトした場合を示し、（ａ）は実施例３におけるスペクトルを示す説明図、（ｂ）は実施例４におけるスペクトルを示す説明図である。
【図１６】実施例３の静止物判定処理を示すフローチャートである。
【図１７】ＦＭＣＷレーダの原理を表す説明図である。
【図１８】ＦＭＣＷレーダによるビート信号スペクトルを示す説明図である。
【符号の説明】
２…レーダ装置 １０…送受信部
１２…送信器 １２ａ…変調器
１２ｂ…電圧制御発振器 １２ｃ，１２ｄ…電力分配器
１２ｅ…送信アンテナ １４，１６…受信器
１４ａ，１６ａ…受信アンテナ １４ｂ，１６ｂ…ミキサ
１４ｃ，１６ｃ…前置増幅器 １４ｄ，１６ｄ…ローパスフィルタ
１４ｅ，１６ｅ…後置増幅器 ２０…信号処理部
２２…三角波発生器 ２４ａ，２４ｂ…Ａ／Ｄ変換器
２６…マイクロコンピュータ ２８…演算処理装置[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention is an FMCW radar device which is used for preventing collision of a moving body, following a certain distance, and detecting a relative speed and a distance to a target existing outside the moving body by transmitting / receiving radar waves.In placeRelated.
[0002]
[Prior art]
a) Conventionally, in an FMCW radar device, a transmission signal whose frequency is gradually increased and decreased by a triangular modulation signal is transmitted as a radar wave, a radar wave reflected by a target is received, and a reception signal is received. A beat signal is generated by mixing with a transmission signal. Then, the frequency of the beat signal (beat frequency) is specified for each section of an ascending portion where the frequency of the transmission signal increases and a descending portion where the frequency decreases using a signal processor or the like. The following formulas (A) and (B) are used to calculate the distance D to the target and the relative speed V based on the beat frequency (up beat frequency) fb1 and the down beat frequency fb2 (down beat frequency). are doing.
[0003]
V = (C / (4 * f0)) * (fb2-fb1) (A)
D = (C / (8 * △ F * fm)) * (fb1 + fb2) (B)
Where ΔF is the frequency displacement width (frequency displacement width) of the transmission signal, f0 is the center frequency of the transmission signal, 1 / fm is the time required for one period of modulation (that is, fm is the repetition frequency of a triangular wave), and C is the speed of light. Represent.
[0004]
Therefore, when measurement is performed using the FMCW radar device, for example, a change in the frequency of the transmission signal T and the frequency of the reception signal R shown in FIG. 17 is obtained according to the relationship between the FMCW radar device and the target.
Specifically, as shown in FIG. 17 (a), when the moving speed of the moving object to which the radar device is attached is equal to the moving speed of the target reflecting the radar wave (relative speed V = 0), the reflected light is reflected on the target. The received radar wave is delayed by the time required for reciprocation with the target, and the graph of the received signal R is obtained by shifting the graph of the transmitted signal T along the time axis. It is equal to the frequency fb2 (fb1 = fb2).
[0005]
On the other hand, as shown in FIG. 17B, when the moving speed with respect to the target is different (relative speed V ≠ 0), the radar wave reflected by the target further has a Doppler according to the relative speed V with respect to the target. Due to the shift, the graph of the received signal R is obtained by shifting the graph of the transmitted signal T along the frequency axis by the Doppler shift due to the relative speed V, and is different from the up beat frequency fb1 and the down beat frequency fb2. (Fb1 ≠ fb2).
[0006]
Therefore, the distance D to the target and the relative speed V can be calculated based on the upbeat frequency fb1 and the downbeat frequency fb2.
b) In recent years, as a technique for distinguishing a moving object from a stationary object by using the above-mentioned FMCW radar device, a technique described in the following JP-A-7-98375 or JP-A-7-191133 has been proposed. ing.
[0007]
This technique uses the physical principle that a stationary object approaches at -VB if the vehicle runs at speed VB.
Specifically, for example, when the direction approaching the FMCW radar device is positive, and when the own vehicle speed is represented by -VB, the speed of the stationary object viewed from the FMCW radar device is VB, the upbeat beat frequency fb1 and The difference from the down beat frequency fb2 is represented by the following equation (C).
[0008]
(Fb2-fb1) = (4 * VB * f0) / C (C)
When the spectrum is displayed by performing a frequency analysis by a well-known Fourier transform, the spectrum of the rising beat signal including the rising beat frequency fb1 (the rising beat signal spectrum) and the spectrum of the falling beat signal including the falling beat frequency fb2 are obtained. (Downward beat signal spectrum) is as shown in FIG.
[0009]
At this time, if the own vehicle speed VB is known, the down beat signal spectrum is shifted by the frequency of (fb2−fb1), as shown in FIG. 18B.
Here, if the target to be measured is a stationary object, that is, if the target is approaching at the own vehicle speed VB, the spectrum frequencies of the target and the target are approaching each other. Whether or not the target is a stationary object can be determined based on the state of coincidence between the spectra of the up beat signal and the down beat signal shown in (b).
[0010]
[Problems to be solved by the invention]
However, according to the above-described technique, the spectrum shift amount (frequency shift amount) is calculated based on the principle that a stationary object approaches −VB only when the vehicle speed is VB. As described in (1) to (4), there is a problem that accuracy in distinguishing a moving object from a stationary object is lacking.
[0011]
(1) An accurate frequency shift amount cannot be calculated due to a response delay or an error of the vehicle speed sensor.
That is, the frequency shift amount can be basically obtained from the own vehicle speed. However, when the own vehicle speed is calculated by another on-vehicle computer, the actual shift of the own vehicle due to communication delay, filtering effect, and the like. There is a delay from the speed, and there is also an error in the sensor itself. Therefore, if the frequency shift amount is simply calculated from the own vehicle speed, the frequency shift amount is not accurate, and as a result, an incorrect determination is made.
[0012]
{Circle around (2)} Since the direction of the radar beam is not taken into account, an accurate frequency shift amount cannot be determined.
In a radar, a beam steer, or a scan beam sensor in which the beam is directed in a direction other than the traveling direction of the vehicle, a deviation occurs between the (apparent) moving direction of the stationary object and the beam direction in which the relative velocity can be detected by the Doppler effect. This effect becomes more pronounced as the radar detects a wide range (large beam steering angle).
[0013]
{Circle around (3)} In the spectrum comparison after the shift, since only the amplitude information is to be evaluated, an incorrect coincidence determination is performed.
Conventionally, when performing spectrum comparison, only amplitude information such as the peak level and shape of a spectrum peak (which is a large power spectrum having a peak corresponding to a target) is used. When a peak exists, a problem occurs. That is, since the spectrum peak from the moving object is recognized as that from the stationary object and separated and removed, the recognition of the moving object, which should be the original purpose, is not performed correctly.
[0014]
{Circle around (4)} When spectral peaks from a stationary object and a moving object overlap, the spectral peaks of the moving object may be erased.
For example, when the FMCW radar is used as a vehicle-mounted radar, a stationary object such as a roadside object such as a guardrail and a moving object which is a vehicle traveling ahead appear in the spectrum. In such a case, depending on the relationship between the vehicle speed and the distance to the stationary object, the distance to the moving object, the relative speed, and the like, the spectral peaks of the stationary object and the moving object may still be synthesized.
[0015]
In particular, in a stationary object having a large reflection level such as a tunnel entrance, the peak level is large, and the frequency spread near the peak is large, so that the spectral peak of the moving object may be buried or overlapped. May be observed as if it were a single spectral peak.
[0016]
In this case, if the spectrum is simply subtracted as in the above-described known technique, the spectrum peak is also subtracted from the moving object, and the radar cannot function normally.
The present invention has been made to solve the above problems, and has as its object to provide an FMCW radar device that can accurately recognize a moving object and a stationary object.
[0017]
[Means for Solving the Problems]
(1) According to the first aspect of the present invention, the transmission means generates a transmission signal whose frequency is gradually increased and decreased by, for example, a triangular wave modulation signal and transmits the radar signal as a radar wave, and the reception means is reflected by the target. The received radar wave is received to generate a received signal, and the received signal is mixed with a transmission signal to generate a beat signal.
[0018]
Then, as shown in FIG. 7, for example, as shown in FIG. 7, the spectrum creating means performs spectrum analysis corresponding to a plurality of targets by using a well-known Fourier transform to analyze the frequency of the transmission signal. In addition to creating an uplink spectrum having (power spectrum), a downlink spectrum having a plurality of spectrum peaks is similarly created from a downlink beat signal at the time of downlink modulation in which the frequency of the transmission signal decreases.
[0019]
Next, the detecting means shifts at least one of the spectrum peak of the upstream spectrum and the spectrum peak of the downstream spectrum by a predetermined frequency, and for example, so that the downstream spectrum coincides with the upstream spectrum, for example, to the left of FIG. The moving state of the target is detected by moving each frequency peak and comparing each spectrum peak.
[0020]
In particular, in the present invention, when setting the frequency shift amount based on the speed of the vehicle equipped with the FMCW radar device, the multiple shift amount setting means sets the plurality of frequency shift amounts in consideration of measurement errors. .
For example, in the related art, the frequency shift amount (fb2−fb1) is set based on the following equation (C). However, in the present invention, not only a single frequency shift amount but also a basic frequency shift amount is used. Another frequency shift amount is set within a predetermined range on both sides of the amount.
[0021]
(Fb2-fb1) = (4 * VB * f0) / C (C)
Next, the evaluation means evaluates a spectrum matching degree indicating a matching degree between the two spectra for each of the uplink spectrum and the downlink spectrum corresponding to each frequency shift amount. Here, when there are many spectral peaks, the degree of coincidence between the spectral peaks is examined, and the overall evaluation is performed to evaluate the degree of spectral match between the two spectra. If there is only one pair of spectrum peaks, the degree of matching between the two spectra may be evaluated based on the degree of matching between the pair of spectrum peaks.
[0022]
For example, the downlink spectrum is shifted by a certain frequency shift amount, and in this state, how much each spectrum peak of the uplink spectrum matches each spectrum peak of the downlink spectrum, for example, in amplitude and phase (spectral matching degree) is determined. For example, the evaluation can be performed by using the total spectrum sum Sum1 obtained by summing the sum Sum2 of the evaluation values | Vp | described later.
[0023]
Then, based on this evaluation result, the determining means determines that the frequency shift amount having the highest degree of spectrum matching is a true frequency shift amount that is less affected by errors in the vehicle speed sensor and the like. In other words, it is considered that the larger the difference between each spectrum peak of the upstream spectrum and each spectrum peak of the downstream spectrum is, the greater the influence of various errors is. Therefore, here, the frequency shift amount (for example, spectrum (The sum Sum1 is the smallest) is selected because the influence of the error is small.
[0024]
Next, with respect to the up spectrum and the down spectrum corresponding to the true frequency shift amount, whether the target corresponding to each spectrum peak is a stationary object is determined by the stillness determination unit (for example, using the neighborhood sum Sum2 described later). A stationary object is detected by determining whether or not the stationary object is present (for example, when the neighborhood sum Sum2 is equal to or less than a predetermined value Thp, the stationary object is determined).
[0025]
As described above, in the present invention, a plurality of frequency shift amounts are set, a true frequency shift amount is obtained from the plurality of frequency shift amounts, and the stationary state of the target is determined using a spectrum corresponding to the true frequency shift amount. .
In other words, when the vehicle speed is calculated by another vehicle-mounted computer, a delay from the actual vehicle speed occurs due to the effect of communication delay and filtering, and the vehicle speed sensor itself has an error. As described above, the frequency shift amount is given a range in advance, and the optimal frequency shift amount is selected and used, thereby eliminating the error of the vehicle speed sensor etc. and accurately stopping the vehicle from many targets. Things can be determined.
[0026]
(2) The invention of claim 2 exemplifies the invention of claim 1, in which a basic frequency shift amount and a frequency shift amount deviated therefrom by a predetermined amount are set.
For example, as shown in the following formula (D), in the present invention, not only a single frequency shift amount but also another frequency shift amount within a predetermined range (± △ Dv) on both sides of the basic frequency shift amount is set. ing.
[0027]
(Fb2−fb1) = (4 * (VB ± Dv) * f0) / C (D)
Therefore, since it can be considered that a true frequency shift exists in these frequency shifts, the true frequency shift can be determined by the above-described evaluation.
[0028]
(3) According to the third aspect of the present invention, the degree of spectral matching of each spectrum peak is evaluated by, for example, a nearby sum Sum2 described later, and each frequency is calculated by, for example, a later-described overall sum Sum1 of the sum of the neighboring sums Sum2 of each spectral peak. The degree of spectrum matching for each shift amount can be evaluated.
[0029]
Therefore, for example, the frequency shift amount in which the total spectrum sum Sum1 is the smallest can be set as the true frequency shift amount.
(4) According to the fourth aspect of the present invention, the degree of spectrum matching is evaluated for each spectral peak by, for example, the neighborhood sum Sum2, and based on the evaluation, "for example, when the neighborhood sum Sum2 is lower than a predetermined threshold value, It is possible to determine the stationary object of the target, such as "determine as an object".
[0030]
In addition, for example, as shown by Sn in FIG. 8, when the spectrum matching degree of each spectrum peak exceeds a predetermined threshold, it can be determined that the object is a moving object. Therefore, for example, it is possible to determine a stationary object from targets having a spectral matching degree equal to or less than the threshold. When the possibility of a synthesized peak is low, a target whose spectral match degree is equal to or less than a threshold can be determined as a stationary object.
[0031]
(5) In the invention of claim 5, when the degree of spectrum matching is evaluated, the evaluation is performed based on information in a frequency band having a predetermined width of a spectrum peak. As a result, more accurate evaluation can be performed, and as a result, an accurate determination of the true frequency shift amount and a stationary object determination can be performed.
[0032]
(6) In the invention of claim 6, since the spectrum matching degree is evaluated based not only on the amplitude of the power spectrum but also on the azimuth information of the target, more accurate evaluation can be performed.
For example, when there is a spectrum peak of the same level by chance in both the upstream and downstream spectra, accurate stationary object determination may not be performed.
[0033]
However, for example, when a phase difference monopulse radar having two reception antennas is used, the azimuth information from the target is indicated by the phase difference between the two systems, and the phase difference is signified by the upstream part and the downstream part. Have the same value but different, and this can be used to evaluate the degree of spectrum matching.
[0034]
(7) The invention of claim 7 exemplifies a spectrum evaluation method. For example, as shown in FIG. 6, the amplitude evaluation value based on the amplitude of the spectrum peak is Y, and the phase based on the azimuth information of the target is Since the absolute value (for example, | Vp |) of the evaluation vector where the evaluation value is X is calculated, and the spectrum matching degree is evaluated using the absolute value, the still object can be more accurately determined.
[0035]
The amplitude evaluation value can be obtained from the following equation (E), and the phase difference evaluation value can be obtained from the following equation (F).

When the spectrum peak is represented by a complex vector, the amplitude of the spectrum peak is represented by the absolute value (length) of the complex vector, and the phase is represented by the rotation angle. Alternatively, the absolute value may be calculated.
[0036]
(8) The invention of claim 8 exemplifies a method of determining a true frequency shift amount.
First, for example, using the following equation (G), the neighborhood sum of the absolute value of the evaluation vector for each spectrum peak is obtained. Here, the reason why the neighborhood sum is used is that the evaluation accuracy is higher when the evaluation is performed including the frequency in the vicinity thereof than when the evaluation is performed only at one point of each spectrum peak.
[0037]

P; peak frequency number (indicating the number of the peak), n; width in the vicinity
Next, after calculating the sum of the neighbors Sum2 regarding the vicinity of each peak, the sum of the neighbors Sum2 is summed up, for example, as in the following equation (H) to obtain the total sum of the spectrum Sum1 for each frequency shift amount.
[0038]
Total sum of the spectrum Sum1 = ΣNearest sum Sum2 (H)
Then, for example, of the spectrum sum Sum1 (| Vp |) of the evaluation value, the frequency shift amount having the smallest value can be determined as the true frequency shift amount TSn. That is, it is considered that the smaller the total spectrum sum Sum1, the smaller the absolute value of each evaluation vector as a whole, and therefore the higher the degree of coincidence of each spectrum peak.
[0039]
(9) According to the ninth aspect of the present invention, after the true frequency shift amount is determined, in the up spectrum and the down spectrum corresponding to the true frequency shift amount, for each pair of spectral peaks, This is an example of whether to perform stationary object determination.
[0040]
Here, the neighborhood sum of the absolute value of a certain evaluation vector is compared with a predetermined threshold value, and when the neighborhood sum is equal to or smaller than the threshold value, the amplitude evaluation value is small and the phase difference evaluation value is small, Therefore, the target is determined to be a stationary object.
[0041]
(10) According to the tenth aspect, the direction of the beam is considered.
That is, when the beam of the radar is directed in a direction other than the traveling direction of the vehicle, a deviation occurs between the moving direction of the stationary object and the beam direction in which the relative speed can be detected by the Doppler effect, and an error occurs in the determination of the stationary object.
[0042]
Therefore, in the present invention, since the frequency shift amount is set in consideration of the direction of the beam, it is possible to more accurately determine the stationary object.
For example, if the direction of the beam is shifted laterally by θ from the traveling direction of the vehicle, the value obtained by multiplying the relative speed of the target measured by this beam by cos θ may be regarded as the relative speed of the true target. it can.
[0043]
(11) In the invention of claim 11, a moving object is considered.
When the FMCW radar is used as an on-vehicle radar, a stationary object such as a roadside object such as a guardrail or a moving object that is a vehicle traveling ahead appears in the spectrum in a mixed manner. There are cases where the spectral peaks of an object and a moving object are synthesized, and the stationary object cannot be accurately determined.
[0044]
Therefore, in the present invention, a moving object prediction flag is set by predicting the current moving position for a target that has already been recognized as a moving object, and the moving object prediction flag is set for the current determination target spectrum peak. Is set, it is regarded as a synthesized spectrum peak, and the spectrum peak is not determined to be a stationary object.
[0045]
Thus, even when traveling in an urban area where there are many roadside objects, a combined peak obtained by combining a moving object peak and a stationary object peak is not recognized as a stationary object, and more accurate stationary object determination can be performed. .
(12) The twelfth aspect of the invention is the same as the first aspect of the invention regarding the basic configuration on which the FMCW radar apparatus is based. The frequency shift amount is set in consideration of the direction. As a result, the stationary object can be more accurately determined.
[0046]
That is, as shown in FIG. 5, when there is a deviation between the traveling direction of the vehicle and the stationary object (roadside object), a deviation occurs in the precise relative speed between the vehicle and the stationary object, and this is caused by the frequency shift. Since the calculation of the quantity is affected, the influence is reduced here by taking the direction of the beam into consideration.
[0047]
(13) The invention of claim 13 has the same operation and effect as the invention of claim 4.
(14) The invention of claim 14 has the same effect as the invention of claim 5.
(15) The invention of claim 15 has the same operation and effect as the invention of claim 6.
(16) The invention of claim 16 has the same effect as the invention of claim 7.
[0048]
(17) The invention of claim 17 has the same operation and effect as the invention of claim 9.
[0061]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an example (example) of an embodiment of the FMCW radar device of the present invention will be described with reference to the drawings.
(Example 1)
a) FIG. 1 is a block diagram showing an entire configuration of an FMCW radar apparatus for obstacle detection (hereinafter simply referred to as a radar apparatus) according to an embodiment to which the present invention is applied. The radar device according to the present embodiment is a so-called phase difference monopulse radar device.
[0062]
As shown in FIG. 1, a radar apparatus 2 according to the present embodiment transmits a radar wave modulated to a predetermined frequency in accordance with a modulation signal Sm, and is radiated from the transmitter 12 and reflected by an obstacle. The transmitter / receiver 10 comprising a pair of receivers 14 and 16 for receiving the radar wave and the transmitter 12 to supply the modulation signal Sm to the intermediate frequency beat signals B1 and B2 output from the receivers 14 and 16. And a signal processing unit 20 for executing a process for detecting an obstacle and determining a stationary object.
[0063]
Here, the transmitter 12 corresponds to the transmitting unit of the present invention, the receivers 14 and 16 correspond to the receiving unit, and the signal processing unit 20 corresponds to other units (such as the spectrum creating unit and the detecting unit).
In this embodiment, in order to detect an obstacle ahead of the vehicle by the radar device 2, the transmission / reception unit 10 is attached to the front of the vehicle, and the signal processing unit 20 is located at a predetermined position in the vehicle compartment or in the vicinity of the vehicle compartment. Installed.
[0064]
Here, the transmitter 12 first converts a voltage-controlled oscillator (VCO) 12b that generates a high-frequency signal in the millimeter wave band as a transmission signal, and converts the modulation signal Sm into an adjustment level of the voltage-controlled oscillator 12b, (MOD) 12a that supplies power to the receivers, and power dividers (COUPs) 12c and 12d that distribute the power of the transmission signal from the voltage controlled oscillator 12b to generate local signals that are supplied to the receivers 14 and 16, A transmission antenna 12e that radiates a radar wave according to a transmission signal.
[0065]
Further, the receiver 14 includes a receiving antenna 14a for receiving a radar wave, a mixer 14b for mixing a received signal from the receiving antenna 14a with a local signal from the power distributor 12d, and a preamplifier for amplifying an output of the mixer 14b. 14c, a low-pass filter 14d that removes unnecessary high-frequency components from the output of the preamplifier 14c and extracts a beat signal B1, which is a difference component between the frequencies of the transmission signal and the reception signal, and sets the beat signal B1 to a required signal level. And a post-amplifier 14e for amplification.
[0066]
The receiver 16 has exactly the same configuration as the receiver 14 (14a to 14e correspond to 16a to 16e), receives a local signal from the power distributor 12c, and outputs a beat signal B2. The receiver 14 is called a receiving channel CH1, and the receiver 16 is called a receiving channel CH2.
[0067]
On the other hand, the signal processing unit 20 is activated by the activation signal C1 and generates the triangular wave-shaped modulation signal Sm, and is activated by the activation signal C2 and converts the beat signals B1 and B2 from the receivers 14 and 16 into the triangular wave. A / D converters 24a and 24b for converting digital data D1 and D2, and CPUs 26a, ROM 26b and RAM 26c are mainly provided. 24b is operated. At the same time, based on the digital data D1 and D2 obtained through the A / D converters 24a and 24b, the obstacle to detect the distance to the obstacle, the relative speed, and the direction of the obstacle, and to determine the stationary object The microcomputer 26 includes a well-known microcomputer 26 that executes an object detection process (to be described later), and an arithmetic processing device 28 that executes an operation of a fast Fourier transform (FFT) based on an instruction from the microcomputer 26.
[0068]
When the A / D converters 24a and 24b start operating in response to the activation signal C2, the A / D converters A / D convert the beat signals B1 and B2 at predetermined time intervals, write the beat signals B1 and B2 into a predetermined area of the RAM 26c, and write a predetermined number of times. When the A / D conversion ends, an end flag (not shown) set on the RAM 26c is set, and the operation is stopped.
[0069]
Then, the triangular wave generator 22 is activated by the activation signal C1, and when the modulation signal Sm is input to the voltage control oscillator 12b via the modulator 12a, the voltage control oscillator 12b generates the triangular waveform of the modulation signal Sm. Modulation is performed such that the frequency increases at a predetermined rate according to the ascending gradient (hereinafter, this section is referred to as a rising section), and the frequency decreases according to the following descending gradient (hereinafter, this section is referred to as a descending section). And outputs the transmitted signal.
[0070]
FIG. 2 is an explanatory diagram illustrating a modulation state of a transmission signal. As shown in FIG. 2, the frequency of the transmission signal is modulated by the modulation signal Sm so as to increase or decrease by ΔF during the period of 1 / fm, and the center frequency of the change is f0. The reason why the frequency is modulated at intervals of 100 ms is that an obstacle detection process described later is executed at a cycle of 100 ms, and the activation signal C1 is generated during the process.
[0071]
A radar wave corresponding to the transmission signal is transmitted from the transmitter 12, and the radar waves reflected on the obstacle are received by the receivers 14 and 16. Then, in the receivers 14 and 16, beat signals B1 and B2 are generated by mixing the reception signals output from the reception antennas 14a and 16a and the transmission signals from the transmitter 12. Note that the received signal is delayed with respect to the transmitted signal by the time that the radar wave travels back and forth to the obstacle, and if there is a relative speed between the radar signal and the obstacle, the received signal is subjected to a Doppler shift according to this. . Therefore, the beat signals B1 and B2 include the delay component fr and the Doppler component fd (see FIG. 17).
[0072]
Then, as shown in FIG. 3, digital data D1 obtained by A / D conversion of the beat signal B1 by the A / D converter 24a is sequentially stored in data blocks DB1 and DB2 on the RAM 26c. Digital data D2 obtained by A / D-converting the beat signal B2 by the converter 24b is similarly stored in the data blocks DB3 and DB4. The A / D converters 24a and 24b are activated at the same time as the activation of the triangular wave generator 22, and perform a predetermined number of A / D conversions while the modulation signal Sm is being output. The data blocks DB1 and DB3 in which the number of data are stored store the rising part data corresponding to the rising part of the transmission signal, and the data blocks DB2 and DB4 in which the latter half of the data are stored store the falling part of the transmission signal. The descending part data corresponding to the part is stored.
[0073]
The data stored in each of the data blocks DB1 to DB4 in this manner is processed by the microcomputer 26 and the arithmetic processing unit 28, and is used for detecting obstacles and stationary objects.
b) Next, the obstacle detection processing executed by the microcomputer 26 will be described with reference to the flowchart of FIG. Note that this obstacle detection processing is started at a period of 100 ms.
[0074]
As shown in FIG. 4, when the present process is started, first, at step 110, a start signal C1 is output to start the triangular wave generator 22, and at step 120, the end flag on the RAM 26c is cleared. At the same time, the activation signal C2 is output to activate the A / D converters 24a and 24b.
[0075]
As a result, the transmitter 12 that has received the modulation signal Sm from the triangular wave generator 22 transmits the frequency-modulated radar wave, and receives the radar wave reflected by the obstacle, thereby transmitting the radar wave from the receivers 14 and 16. The output beat signals B1 and B2 are converted into digital data D1 and D2 via A / D converters 24a and 24b and written to the RAM 26c.
[0076]
In the following step 130, it is determined whether or not the A / D conversion has been completed by checking the end flag on the RAM 26c. If the end flag has not been set, it is determined that the A / D conversion has not been completed, and the process repeats step 130 to wait. On the other hand, if the end flag has been set, the A / D conversion has not been completed. Is determined to have been completed, and the process proceeds to step 140.
[0077]
In step 140, one of the data blocks DB1 to DB4 on the RAM 26c is sequentially selected, and the data of the data block DBi (i = 1 to 4) is input to the arithmetic processing unit 28 to execute the FFT operation. . The data input to the arithmetic processing device 28 is subjected to well-known window processing using a Hanning window, a triangular window, or the like in order to suppress a side lobe appearing by the FFT operation. Then, a complex vector for each frequency is obtained as the calculation result.
[0078]
In step 150, based on the absolute value of the complex vector, that is, based on the amplitude of the frequency component indicated by the complex vector, all frequency components (hereinafter, referred to as peak frequency components) at the peak of the peak (spectral peak) on the frequency spectrum are detected. Then, the frequency is specified as the peak frequency, and the process proceeds to step 160. As a method of detecting the peak of the spectrum peak, for example, the amount of change in the amplitude with respect to the frequency is sequentially obtained, and the frequency at which the sign of the amount of change is reversed before and after that is identified as the peak of the spectrum peak. do it.
[0079]
In step 160, the phase of the peak frequency component specified in step 150 is calculated. This phase is equal to the angle between the complex vector and the real axis, and can be easily obtained from the complex vector.
In the following step 170, it is determined whether or not there is an unprocessed data block DBi. If there is an unprocessed data block DBi, the process returns to step 140, and the processing of steps 140 to 160 is performed on the unprocessed data block DBi. Execute, if there is no unprocessed one, go to step 175.
[0080]
In step 175, as will be described in detail later, a stationary object determination process is performed to determine whether the detected obstacle is a stationary object.
In the following step 180, by comparing the amplitudes of the peak frequency components, that is, the powers, those having the same power in the rising part and the falling part are specified as a pair of the peak frequency components based on the reflected waves from the same obstacle. Execute the pairing process.
[0081]
However, in this case, in the stationary object determination process in the step 175, the pairing process is not performed for peaks that are clearly determined to be not a pair.
This pairing process is the same as, for example, the process shown in FIG. 7 of Japanese Patent Application No. 8-179227 and its description, and a detailed description thereof will be omitted.
[0082]
In the following step 190, a distance / speed azimuth calculation process for calculating the distance to the obstacle, the relative speed, and the azimuth of the obstacle is executed by using the peak frequency components paired in step 180, and this process is executed. finish.
For example, a phase difference is calculated between each of the receiving channels CH1 and CH2 for each of the ascending portion and the descending portion, and if the signs of the phase differences are not equal, an obstacle is determined using the following equations (1) and (2). Is calculated.
[0083]
V = (C / (4 * f0)) * (fb2-fb1) (1)
D = (C / (8 * △ F * fm)) * (fb1 + fb2) (2)
Here, ΔF is the frequency displacement width (frequency displacement width) of the transmission signal, f0 is the center frequency of the transmission signal, 1 / fm is the time required for one cycle of modulation (that is, fm is the repetition frequency of a triangular wave), C is the speed of light, fb1 represents the rising beat frequency (upbeat frequency), fb2 represents the falling beat frequency (downbeat frequency), and C represents the speed of light.
[0084]
This distance / velocity / azimuth calculation processing is the same as the processing shown in, for example, FIG. 5 of Japanese Patent Application No. 8-179227 and its description, and therefore, detailed description thereof will be omitted.
c) Next, the basic principle of the stationary object determination process performed in step 175 will be described.
[0085]
Here, in the determination of a stationary object, first, a plurality of frequency shift amounts (Sn-1, Sn, Sn + 1) are set in consideration of an error of a vehicle speed sensor or the like, and a true value is determined from among them by using an evaluation function. The frequency shift amount TSn is obtained. Then, using the up and down spectrums corresponding to the true frequency shift amount TSn, it is determined whether the target corresponding to the peak frequency component is a moving object or a stationary object. This will be described in detail below.
[0086]
(I) First, procedures (1) to (3) for calculating the frequency shift amount (hereinafter, also simply referred to as shift amount) will be described.
(1) First, a basic frequency shift amount calculation formula for calculating a basic frequency shift amount (basic shift amount) is set using the vehicle speed VB or the like.
[0087]
That is, the formula (1) is modified to calculate a basic amount (basic shift amount = (fb2−fb1)) for determining how much the spectrum is shifted for the determination of a stationary object. Set.
Basic shift amount = (fb2−fb1) = (4 * VB * f0) / C (3)
Here, fb1 is the up beat frequency, fb2 is the down beat frequency, VB is the vehicle speed, f0 is the center frequency of the transmission signal, and C is the speed of light.
[0088]
At this time, if the error of the vehicle speed sensor is known or has been learned, the own vehicle speed VB corrected by a correction coefficient, map calculation, or the like is used.
{Circle around (2)} Next, correction is performed according to the beam angle of the laser device 12.
As shown in FIG. 6, assuming that the beam angle from the sensor front direction or the traveling direction in the case of the beam steering and the scan sensor is θ, in the case of a radar that detects the relative velocity using the Doppler effect, the detectable velocity Since the component is a velocity component equal to the beam direction (−VB * COSθ), it is necessary to perform correction. This component becomes smaller as the beam steer and scan angles increase, the deviation from the true moving speed (approaching speed; -VB) increases, and accurate frequency shift cannot be performed.
[0089]
Therefore, here, the following equation (4) for the first correction in which the angle correction coefficient (COSθ) is added to the basic shift amount is set. Thereby, the basic shift amount is corrected to be smaller as θ becomes larger.
Shift amount after first correction = (4 * COS (θ) * VB * f0) / C (4)
{Circle around (3)} Next, a correction is made in consideration of the response delay of the vehicle speed sensor.
[0090]
The vehicle speed sensor generally measures a time interval between pulse signals from a drive system and a wheel system, and detects an actual vehicle speed from the measured value. However, in actuality, since it is temporally filtered due to stability, noise, or the like, there is a response delay from the actual vehicle speed. For example, when traveling at a speed of 100 km / h, the delay is not significant, but when accelerating or decelerating, a time delay of several kilometers per hour occurs. Therefore, in the present embodiment, a process having an allowable value is performed in consideration of the delay.
[0091]
Specifically, the time delay varies depending on the type of the actual vehicle, and therefore, it is possible to provide a basic delay for each vehicle in advance. This depends on the actual vehicle speed, the time constant of the filter of the vehicle speed sensor, and the like, but is set here as the speed delay value Dv. Therefore, the following equation (5) for the second correction is set using this value.
[0092]

Here, the value of the speed delay value Dv is a value in consideration of the resolution of the vehicle speed sensor. For example, when an error of ± 5 km / h occurs in the vehicle speed sensor, it is possible to set three types, for example, as Dv = (− 5, 0, +5).
[0093]
In other words, there is a possibility that the time delay of the vehicle speed sensor may actually occur by the width of the speed delay value Dv, so that the basic shift amount can be given a certain width by the above equation (5). That is, as will be described in detail later, a plurality of frequency shift amounts are set with a certain width in the frequency shift amount, and the most matched frequency shift amount is selected from among the plurality of frequency shift amounts. The frequency shift amount can be set.
[0094]
In order to simplify the processing, it is possible to set Dv to a value that absorbs the influence of the angle component (2).
(Ii) Next, an evaluation method using an evaluation function will be described.
Here, the spectrum of the descending portion is shifted using the plurality of frequency shift amounts obtained by the above equation (5), and the degree of coincidence with the spectrum of the ascending portion is compared.
[0095]
{Circle around (1)} Conventionally, only the peak frequency components of the corresponding spectrum peaks of the uplink and downlink spectra are subtracted using the uniquely determined frequency shift amount. However, in this embodiment, a plurality of frequency shift amounts are used. The following evaluation function is used to determine the optimal frequency shift amount from among (the shift width at the time of shifting).
[0096]
This evaluation function uses not only the amplitude of the spectrum peak but also the phase difference indicating the azimuth information obtained by the phase difference monopulse radar as shown in the following equations (6) and (7). The phase difference is a value obtained by subtracting the respective phase information received by a monopulse radar having two receiving systems, and a phase difference monopulse radar uses the phase difference to determine the azimuth of the object. .
[0097]

In the case of a phase-difference monopulse radar, the signs are inverted in the rising part and the falling part because of the configuration, so that if the sum is 0, they match.
[0098]
Then, as shown in FIG. 6, the length | Vp | of the evaluation vector Vp having the amplitude evaluation value Y and the phase difference evaluation value X is used as the evaluation value.
Next, this evaluation value | Vp | is obtained for each peak frequency component, and the sum is obtained. In this case, as shown in the following equation (8), not only the target peak frequency component but also Similarly, the evaluation value | Vp | is obtained for the frequencies in the vicinity thereof, and the sum thereof (the sum Sum2) is obtained. The range in which the neighborhood sum Sum2 is obtained is, as shown in FIG. 8, a band-shaped range sandwiched between left and right broken lines with the dashed line as the center, and varies depending on the resolution of the FFT.
[0099]
Here, the reason why the neighborhood sum Sum2 is obtained is that the accuracy is higher than in the case where a single peak frequency component is used.

Where P; peak frequency number (indicating the order of peaks to be evaluated),
n: Width of neighborhood (when neighborhood is divided into n)
In this case, instead of calculating the sum of the vicinity of all the peak frequency components, the sum Sum2 of the vicinity of only the peak frequency components for determining the stationary object and the moving object is calculated. This is because if processing is performed on all the data, correct results may not be detected due to peaks such as noise and clutter, and a large amount of calculation time is required.
[0100]
Here, as shown in FIG. 7, for example, spectral peaks present in the rising part (upward spectrum) are denoted by Pu1, Pu2, Pu3, and Pu4, and spectral peaks present in the falling part (downward spectrum) are denoted by Pd1, Pd2, and Pd3. , Pd4 (where 1, 2, 3 are stationary objects and 4 is a spectrum peak of a moving object).
[0101]
In this case, when the downlink spectrum is shifted by a plurality of frequency shift amounts, the result is as shown in FIG. In FIG. 8, three kinds of frequency shift amounts of Sn ± 1 are set for the basic shift amount Sn. The value of ± 1 with respect to the frequency shift amount Sn means a width in consideration of a response delay, an error, and the like of the vehicle speed sensor when calculating the shift amount, and the ± 1 width is used for convenience of description.
[0102]
Then, for the uplink and downlink spectrums corresponding to the respective frequency shift amounts, that is, for a pair of spectral peaks corresponding to both the uplink and downlink spectra, the neighborhood sum Sum2 is calculated in the vicinity of each peak frequency component. It is calculated. {Circle around (2)} Next, after calculating the sum Sum2 for each of the spectrum peaks, the sum Sum2 is summed up as in the following equation (9) to obtain the sum Sum1 of the entire spectrum for each frequency shift amount.
[0103]
Total sum Sum1 = Σneighborhood sum2 (9)
Then, the frequency shift amount with the smallest sum Sum1 (| Vp |) of each spectrum is defined as a true frequency shift amount TSn. Therefore, in FIG. 8, for example, the frequency shift amount Sn corresponding to the center spectrum is selected as the true frequency shift amount TSn.
[0104]
As described above, the reason that the total sum Sum1 (| Vp |) is the true frequency shift amount TSn is that the smaller the total sum Sum1, the smaller the absolute value of each evaluation vector as a whole. This is because the degree of coincidence of each spectrum peak is considered to be large.
[0105]
Here, three types of shift widths have been described as examples, but actually, the true frequency shift amount TSn is determined based on the width obtained by the above equation (3) indicating the second corrected shift amount. Is done.
Further, once the true frequency shift amount TSn is obtained, the calculation of the frequency shift amount may be performed in consideration of the true frequency shift amount in the next calculation.
[0106]
(Iii) Next, a method of separating a moving object and a stationary object using the true frequency shift amount TSn will be described.
In the calculations (i) and (ii) above, the true frequency shift amount TSn was obtained by excluding the influence of the vehicle speed sensor delay, error, beam direction, and the like. Here, a moving object and a stationary object are determined based on the degree of coincidence of the spectrum shifted by the frequency shift amount.
[0107]
Conventionally, only the subtraction of the spectrum was performed, but in the present embodiment, considering the possibility that a spectrum peak having the same peak level may accidentally exist, that is, the peak of the moving object and the peak of the stationary object are combined. The processing is performed in consideration of the case where it is performed.
Specifically, as shown in FIG. 8, the separation between the moving object and the stationary object is performed using the sum Sum2 of the evaluation values.
[0108]
That is, as shown in FIG. 8, when the sum Sum2 of a certain spectral peak is equal to or smaller than the threshold Thp, the spectrum of the rising part and the falling part with respect to the amplitude of the spectrum peak (peak amplitude) and the beam azimuth (phase difference) are determined. Assuming that the peaks match, a stationary object determination that the coincident spectral peak is a stationary object is performed. On the other hand, if the sum Sum2 of a certain spectral peak exceeds the threshold value Thp, it is determined that the combination of the spectral peaks of the moving object and the stationary object, or the combination of the spectral peaks due to noise or the like, and the stationary object is not determined. Note that the threshold value Thp may be changed depending on the running state of the vehicle, weather, and the like, and is not limited to a fixed value.
[0109]
(Iv) Next, the determination of the combined peak of the moving object and the stationary object will be described.
The determination as to whether or not the peak to be determined is a combination of the spectral peaks of the moving object and the stationary object is performed by using the moving object prediction flag. That is, as shown in FIG. 9, a spectrum peak which was not determined as a stationary object in the previous time and which is determined as a moving object by another means appears after Δt depending on the motion state of the moving object. The position of the likely spectral peak (peak position) is predicted. The flag indicating the predicted position is called a moving object prediction flag. Therefore, even if the stationary object is determined, if the moving object prediction flag is set at the peak position, the moving object and the stationary object are determined to be a combined peak, and the stationary object determination is not performed. In the same manner, the moving object prediction flag is sequentially set.
[0110]
d) Next, the stationary object determination process in step 175 performed based on the principle will be described with reference to the flowchart in FIG.
This process is a process for determining whether or not the spectrum peak of the obstacle (target) recognized by the radar device 2 corresponds to a stationary object based on the above-described principle.
[0111]
First, in step 200 of FIG. 10, the second corrected shift amount obtained by correcting the basic shift amount based on the equation (5) set in the procedures (1) to (3), that is, the frequency shift amount Determine the width of
In the following step 210, the frequency shift is performed from, for example, the shift width Sn-1 (frequency shift has not been executed). For example, the entire downstream spectrum is shifted. Therefore, when the present processing is passed for the second time, the frequency shift of the next shift width is performed.
[0112]
In the following step 220, the peak frequency of the peak of the spectrum peak which is to be evaluated is determined based on the above-mentioned equations (6) and (7) in the spectrum frequency-shifted in the above-mentioned step 210, based on the above formulas (6) and (7). Evaluation values | Vp | near the components are sequentially calculated.
[0113]
In the following step 230, the evaluation values | Vp | near the vertexes of the predetermined spectrum peak calculated in step 220 are summed based on the equation (8) to calculate the sum Sum2 of the vicinity of the peak.
In the following step 240, it is determined whether or not the calculation of the sum Sum2 has been completed for the number of peaks desired to be determined. For example, as shown in FIG. 8, it is determined whether or not all the sums Sum2 of the respective neighbors have been calculated for the four spectral peaks at the shift width Sn-1. If an affirmative determination is made here, the process proceeds to step 250, while if a negative determination is made, the process returns to step 220 and thereafter, and the evaluation values | Vp | of other spectral peaks and the neighborhood sum Sum2 are calculated.
[0114]
In Step 250, since the calculation of the sum Sum2 of all the spectral peaks is completed in Steps 220 to 240, the sum is calculated based on the equation (9) to calculate the total sum Sum1 of the spectrum. .
In the following step 260, it is determined whether or not the shift width has been shifted the number of times. For example, when there are three types of frequency shift amounts of Sn-1, Sn, and Sn + 1, it is determined whether or not the above-described calculation of the total spectrum sum Sum1 or the like has been performed for each frequency shift amount. If an affirmative determination is made here, the process proceeds to step 270. On the other hand, if a negative determination is made, the process returns to step 210 and thereafter, and a process of calculating another total sum Sum1 is performed in the same manner as described above.
[0115]
In step 270, the values of all (for example, three in FIG. 8) total spectrum sums Sum1 are compared, and the frequency shift amount corresponding to the smallest value is set as the true frequency shift amount TSn.
In the following step 280, a predetermined spectrum peak (specifically, a pair of spectrum peaks corresponding to the up spectrum and the down spectrum) with respect to the spectrum frequency-shifted by the true frequency shift amount (in FIG. 8, the spectrum of the center shift amount Sn). It is determined whether or not the neighborhood sum Sum2 is equal to or smaller than the threshold value Thp. If an affirmative determination is made here, the process proceeds to step 290, while if a negative determination is made, the process proceeds to step 320.
[0116]
In step 320, since the sum Sum2 of the spectrum peaks is larger than the threshold Thp, that is, since the degree of coincidence of evaluation is low to determine a stationary object, it is determined that the object marker corresponding to the spectrum peak is a moving object. A moving object flag indicating that the object is an object is set, and the process proceeds to step 330.
[0117]
On the other hand, in step 290, it is determined whether or not the moving object prediction flag is set at the peak position. If the determination is affirmative, the process proceeds to step 310, whereas if the determination is negative, the process proceeds to step 300.
In step 310, since the moving object prediction flag is set, it is determined that the spectrum peak is a combined peak of a stationary object and a moving object, a combined peak flag indicating a combined peak is set, and the process proceeds to step 330. .
[0118]
On the other hand, in step 300, since the moving object prediction flag is not set, that is, since the neighborhood sum Sum2 is equal to or less than the threshold Thp and the moving object prediction flag is set, it is determined that the spectrum peak is the spectrum peak of the stationary object. Then, a stationary object flag indicating a stationary object is set, and the routine proceeds to step 330.
[0119]
In step 330, it is determined whether or not the processing in steps 280 to 310, that is, the processing for determining what the spectral peak means, has been completed for the number of spectral peaks desired to be determined. Here, if a negative determination is made, the processing of step 280 and subsequent steps is repeated, while if an affirmative determination is made, the present processing is temporarily terminated.
[0120]
As described above, in the radar device 2 of the present embodiment, when determining a stationary object, a plurality of frequency shift amounts (Sn-1, Sn, Sn + 1) are set in consideration of an error of a vehicle speed sensor or the like. The true frequency shift amount TSn is obtained from the evaluation function using the evaluation function composed of the expressions (6) to (9). Then, using the up spectrum and the down spectrum corresponding to the true frequency shift amount TSn, the spectrum peaks of both spectra are evaluated to separate the moving object and the stationary object. Thereby, a moving object and a stationary object can be accurately distinguished.
[0121]
That is, in the present embodiment, correction is performed in consideration of the direction of the beam, and further, in consideration of the error due to the vehicle speed sensor and the like, a plurality of frequency shift amounts are set instead of a single frequency shift amount as in the related art. In addition, by evaluating the frequency shift amount using the sum Sum2 and the total sum Sum1 of the spectra obtained from the evaluation value | Vp |, the true frequency shift amount TSn excluding errors and the like can be determined.
[0122]
In addition, moving objects can be eliminated by using the upstream and downstream spectrums corresponding to the true frequency shift amount TSn and comparing the sum Sum2 of each pair of spectral peaks with the threshold value Thp in both spectra. Therefore, it is possible to accurately distinguish whether each target is a moving object or a stationary object.
[0123]
In addition, by using the moving object prediction flag, it is possible to eliminate the combined flag of the moving object and the stationary object. Thereby, it is possible to reliably recognize only the stationary object, and from that point, there is an advantage that the accuracy of the stationary object determination is improved.
(Example 2)
Next, a second embodiment will be described, but the description of the same parts as in the first embodiment will be omitted.
[0124]
In the first embodiment, processing such as shift is performed on the waveform of the spectrum peak (power spectrum) to determine the stationary object. However, in order to store the spectrum waveforms of the rising part and the falling part, a large-capacity memory for storing the waveform shape in the radar device is required. Further, when the frequency analysis is performed by a processor different from the recognition process, it may not be possible to transmit all the waveform data due to the communication speed.
[0125]
As a countermeasure, in this embodiment, the peak information of the spectrum waveform, that is, the peak information about the peak of the spectrum peak (peak frequency indicating the peak, amplitude at the peak (peak level), phase difference indicating the peak between the corresponding spectral peaks). (Phase difference)) was used to determine the stationary object. The details will be described below.
[0126]
a) First, the principle of the present embodiment will be described.
In the present embodiment, the downlink spectrum is shifted using the plurality of frequency shift amounts obtained by the equation (5) of the embodiment 1, and the degree of coincidence between the corresponding spectrum peaks of the uplink spectrum and the downlink spectrum is compared. Here, instead of the waveform information having a predetermined width of each spectrum peak, as shown by the bold line in FIG. 11, only the peak information which is the value corresponding to the peak of each spectrum peak is used to evaluate the degree of spectrum matching.
[0127]
{Circle around (1)} Specifically, first, as in the first embodiment, the length | Vp | of the evaluation vector Vp having the amplitude evaluation value Y and the phase difference evaluation value X is set as the evaluation value.
Here, as shown in FIG. 11, spectral peaks existing in the upstream spectrum are denoted by Pu1, Pu2, Pu3, and Pu4, and spectral peaks present in the downstream spectrum are denoted by Pd1, Pd2, Pd3, and Pd4 (where 1,2, 3 is a stationary object, and 4 is a spectrum peak of a moving object.
[0128]
In this case, when the downlink spectrum is shifted by a plurality of frequency shift amounts, the result is as shown in FIG. In FIG. 12, three kinds of frequency shift amounts of Sn ± 1 are set for the basic shift amount Sn.
Then, for the uplink and downlink spectra corresponding to the respective frequency shift amounts, that is, for a pair of spectral peaks corresponding to both the uplink and downlink spectra, evaluation of peak information corresponding to the peak of each spectral peak. The value | Vp | is calculated.
[0129]
{Circle around (2)} After calculating the respective evaluation values | Vp |, instead of calculating the neighborhood sum Sum2 as in the first embodiment, the respective evaluation values | Vp | Is obtained. The evaluation value sum SumVp corresponds to the total spectrum sum Sum1 of the first embodiment.
[0130]
That is, in this embodiment, since the neighborhood sum Sum2 is unnecessary, this evaluation value sum SumVp corresponds to the total spectrum sum of the first embodiment.
Then, the frequency shift amount having the smallest sum of the evaluation values SumVp is defined as a true frequency shift amount TSn. Therefore, in FIG. 12, for example, the frequency shift amount Sn corresponding to the center spectrum is selected as the true frequency shift amount TSn.
[0131]
{Circle around (3)} Separation of a moving object and a stationary object using the true frequency shift amount TSn is performed as follows.
As shown in FIG. 12, when the evaluation value | Vp | of a certain spectrum peak is equal to or less than the threshold value THp1, it is determined that the spectrum peaks of the upstream and downstream spectra coincide with each other with respect to the peak level and the beam direction (phase difference). Considering this, a stationary object determination that the coincident spectrum peak is a stationary object is performed. On the other hand, when the evaluation value | Vp | of a certain spectrum peak exceeds the threshold value THp1, it is determined that the combination of the peaks of the moving object and the stationary object or the combination of the peaks due to noise or the like, and the stationary object is not determined.
[0132]
b) Next, a stationary object determination process performed based on the above principle will be described with reference to a flowchart of FIG.
This process is a process for determining whether or not the spectrum peak of the obstacle (target) recognized by the radar device 2 corresponds to a stationary object based on the above-described principle.
[0133]
First, in step 400 in FIG. 13, similarly to the first embodiment, the second corrected shift amount obtained by correcting the basic shift amount, that is, the width of the frequency shift amount is determined.
In the following step 410, for example, the frequency is shifted sequentially from the shift width Sn-1. In the following step 420, for the predetermined spectrum peak to be evaluated in the spectrum shifted in frequency in step 410, the evaluation value | Vp | is sequentially determined using only the peak information of the peak of the spectrum peak. calculate.
[0134]
In the following step 440, it is determined whether the process of calculating the evaluation value | Vp | has been completed for the number of spectrum peaks desired to be determined. If the determination is affirmative, the process proceeds to step 450, while if the determination is negative, the process returns to step 420.
In the subsequent step 450, the evaluation values | Vp | of the predetermined spectral peaks calculated in the above step 420 are summed up, and the sum SumVp of the evaluation values is calculated.
[0135]
In a succeeding step 460, it is determined whether or not the shift has been performed the shift width times. If an affirmative determination is made here, the process proceeds to step 470, while if a negative determination is made, the process returns to step 410.
In step 470, the values of all the evaluation value sums SumVp are compared, and the frequency shift amount corresponding to the smallest value is set as the true frequency shift amount TSn.
[0136]
In the following step 480, it is determined whether or not the evaluation value | Vp | of the predetermined peak is equal to or less than the threshold value THp1 with respect to the spectrum shifted in frequency by the true frequency shift amount TSn (in FIG. 12, the spectrum of the central shift amount Sn). I do. If the determination is affirmative, the process proceeds to step 490, while if the determination is negative, the process proceeds to step 520.
[0137]
In step 520, since the evaluation value | Vp | of the spectrum peak is larger than the threshold value THp1, a moving object flag indicating that the object is a moving object is set, and the process proceeds to step 530.
On the other hand, in step 490, it is determined whether or not the moving object prediction flag is set at the position of the spectrum peak. If the determination is affirmative, the process proceeds to step 510, and if the determination is negative, the process proceeds to step 500.
[0138]
In step 510, since the moving object prediction flag has been set, a combined peak flag indicating a combined peak is set, and the process proceeds to step 530.
On the other hand, in step 500, since the moving object prediction flag is not set, a stationary object flag indicating a stationary object is set, and the process proceeds to step 530.
[0139]
In step 530, it is determined whether or not the processing of all the spectral peaks desired to be determined has been completed. Here, if a negative determination is made, the process returns to step 480, while if an affirmative determination is made, the present process is temporarily terminated.
As described above, in the present embodiment, the stationary object determination is performed using only the peak information of the peak, instead of using the waveform information (such as near the peak of the spectral peak) as in the first embodiment. .
[0140]
Therefore, there is an effect that the calculation speed is high. Further, a large-capacity memory is not required, and the cost can be reduced. Furthermore, even when the frequency analysis is performed by a processor different from the recognition process, only the peak information needs to be transmitted, and there is an advantage that the transmission speed is hardly affected.
(Example 3)
Next, a third embodiment will be described, but the description of the same parts as those in the first and second embodiments will be omitted.
[0141]
In the second embodiment, the stationary object determination is performed using only the peak information of the apex of the spectrum peak. However, for example, as illustrated in FIG. 14, the peak shape is not steep like a spectrum from a continuous guardrail, In the case of a shape with a gentle and wide width, the peak of the spectrum peak fluctuates slightly and the peak of the spectrum peak does not match in the rising part and the falling part (mismatch between (1) and (2)). I do.
[0142]
As a countermeasure against this, in the present embodiment, after the frequency shift, the peak of the spectrum peak of the downlink spectrum existing within a predetermined width Pw around the peak frequency of the spectrum peak of the reference uplink spectrum is evaluated as an evaluation value. The stationary object was determined by using | Vp | as the target for calculation. The details will be described below.
[0143]
a) First, the principle of the present embodiment will be described.
Specifically, first, as in the first embodiment, the length | Vp | of the evaluation vector Vp having the amplitude evaluation value Y and the phase difference evaluation value X is set as the evaluation value.
Here, as shown in FIG. 14, spectral peaks existing in the upstream spectrum are denoted by Pu1, Pu2, Pu3, and Pu4, and spectral peaks existing in the downstream spectrum are denoted by Pd1, Pd2, Pd3, and Pd4 (where 1,2, 3 is a stationary object, and 4 is a spectrum peak of a moving object.
[0144]
In this case, when the downlink spectrum is shifted by the basic shift amount Sn, the result is as shown in FIG.
Then, an evaluation value | Vp | of the peak information corresponding to the peak of each spectrum peak is calculated for a pair of spectrum peaks corresponding to both the up spectrum and the down spectrum, which is a reference here. It was checked whether or not the peak of the corresponding spectrum peak of the downstream spectrum exists within a predetermined width Pw around the peak frequency of each spectrum peak of the upstream spectrum.
Specifically, paying attention to the peak frequency of the peak of the predetermined spectrum peak of the upstream spectrum, the amplitude value in the range of ± Pw / 2 from the peak frequency at the corresponding spectrum peak of the downstream spectrum, and its width Pw Investigate sequentially in small sections. In this case, the peak level, which is the peak information of the apex, is stored, but the amplitude value of the waveform at other positions is “0”.
[0145]
Therefore, for example, when there is a peak of the descending A peak corresponding to the peak of the rising A peak within the width Pw, such as the peak A, the evaluation value | Vp | Is smaller than the evaluation value | Vp | Therefore, here, the evaluation value | Vp | is calculated, and the minimum value is obtained. That is, the minimum value of the evaluation value | Vp | of the present embodiment corresponds to the evaluation value | Vp | of the second embodiment (in an ideal state).
[0146]
The separation of the moving object and the stationary object is the same as that of the second embodiment except that the stationary object determination is not performed.
b) Next, the stationary object determination process performed based on the above principle will be described with reference to the flowchart of FIG.
[0147]
First, in step 600 in FIG. 16, similarly to the first and second embodiments, the second corrected shift amount obtained by correcting the basic shift amount, that is, the width of the frequency shift amount is determined.
In the following step 610, the frequency is shifted by the basic shift amount Sn.
In the following step 620, the evaluation value | Vp | is calculated using the peak information of the apex of a predetermined spectrum peak to be evaluated in the spectrum shifted in frequency in the step 610. At this time, as described above, the minimum value of the evaluation value | Vp | in the predetermined width Pw is obtained.
[0148]
In the following step 630, it is determined whether or not the calculation of the minimum value of the evaluation value | Vp | has been completed for the number of spectrum peaks desired to be determined. If the determination is affirmative, the process proceeds to step 640, and if the determination is negative, the process returns to step 620. In step 640, it is determined whether or not the minimum value of the evaluation value | Vp | of the spectrum peak is equal to or smaller than the threshold value THp2. If the determination is affirmative, the process proceeds to step 650, whereas if the determination is negative, the process proceeds to step 680.
[0149]
In step 680, since the minimum value of the evaluation value | Vp | of the spectrum peak is larger than the threshold value THp2, a moving object flag indicating a moving object is set, and the process proceeds to step 690.
On the other hand, in step 650, it is determined whether or not the moving object prediction flag is set at the position of the spectrum peak. If an affirmative determination is made here, the process proceeds to step 670, while if a negative determination is made, the process proceeds to step 660.
[0150]
In step 670, since the moving object prediction flag is set, the synthetic peak flag indicating the synthetic peak is set, and the flow advances to step 690.
On the other hand, in step 660, since the moving object prediction flag is not set, a stationary object flag indicating a stationary object is set, and the process proceeds to step 690.
[0151]
In step 690, it is determined whether or not the processing of all the spectral peaks desired to be determined has been completed. If a negative determination is made here, the process returns to the step 640. On the other hand, if a positive determination is made, the present process is temporarily ended.
As described above, in the present embodiment, since the stationary object determination is performed using only the peak information of the peak of the spectrum peak, the same effect as that of the second embodiment can be obtained.
[0152]
Further, in the present embodiment, it is checked whether or not the peak of the spectrum peak of the rising part has a peak of the spectrum peak of the falling part within the predetermined width Pw, and the minimum value of the evaluation value | Vp | Still object determination is performed using the minimum value of the evaluation value | Vp |.
[0153]
Therefore, there is an advantage that a stationary object can be more reliably determined even when the peak of the spectrum peak may fluctuate, for example, as in the case of a guardrail.
In the case where the stationary object determination is not performed, another unit (for example, a process using a pair shift) can be performed to separately perform the stationary object determination.
(Example 4)
Next, a fourth embodiment will be described, but the description of the same parts as those in the first to third embodiments will be omitted.
[0154]
This embodiment basically performs the same processing as that of the third embodiment. However, since the method for obtaining the minimum value of the evaluation value | Vp | is different, only the different points will be described.
That is, also in the present embodiment, the evaluation value | Vp | of the peak information corresponding to the peak of each spectrum peak is calculated for a pair of spectrum peaks corresponding to both the upstream and downstream spectra. Then, instead of shifting the entire downlink spectrum at once, individual spectrum peaks of the downlink spectrum are shifted within a frequency shift amount (for example, Sn) ± Pw / 2 to evaluate the degree of spectrum matching.
[0155]
More specifically, as shown in FIG. 15B, in the case where, for example, the C peak in the descending portion is shifted in the range of Sn-Pw / 2 to Sn + Pw / 2, and shifted in that range, the third embodiment is described. In the same manner as described above, a process of obtaining the minimum value of the evaluation value | Vp | in the predetermined width Pw is performed. Note that the other spectral peaks are similarly shifted to evaluate the degree of spectral matching.
[0156]
In this embodiment, the same effects as those of the third embodiment can be obtained.
It is needless to say that the present invention is not limited to the above-described embodiment, and can be implemented in various modes without departing from the technical scope of the present invention.
Also,In the above embodiment, the FMCW radar device has been described. However, the present invention may be applied to a recording medium storing a means for executing control by this device.Is applicable.
[0157]
For example, examples of the recording medium include various recording media such as an electronic control unit configured as a microcomputer, a microchip, a floppy disk, a hard disk, and an optical disk.
In other words, there is no particular limitation as long as means such as a program that can execute the control of the FMCW radar apparatus described above is stored.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating an entire configuration of a radar apparatus according to a first embodiment.
FIG. 2 is a graph showing a change in frequency of a transmission signal.
FIG. 3 is an explanatory diagram showing data stored in a RAM.
FIG. 4 is a flowchart illustrating an obstacle detection process.
FIG. 5 is an explanatory diagram illustrating correction based on a beam direction of a laser radar.
FIG. 6 is an explanatory diagram showing evaluation values.
FIG. 7 is an explanatory diagram showing spectra of an ascending portion and a descending portion.
FIG. 8 is an explanatory diagram showing a spectrum when shifting is performed with three shift widths.
FIG. 9 is an explanatory diagram showing a method of setting a moving object prediction flag.
FIG. 10 is a flowchart illustrating a stationary object determination process according to the first embodiment.
FIG. 11 is an explanatory diagram showing spectra of an ascending portion and a descending portion in Example 2.
FIG. 12 is an explanatory diagram showing a spectrum when shifting is performed with three shift widths.
FIG. 13 is a flowchart illustrating still object determination processing according to the second embodiment.
FIG. 14 is an explanatory diagram showing spectra of an ascending portion and a descending portion according to the third embodiment.
15A and 15B are diagrams illustrating a case of shifting by a basic shift width, wherein FIG. 15A is an explanatory diagram illustrating a spectrum according to a third embodiment, and FIG. 15B is an explanatory diagram illustrating a spectrum according to a fourth embodiment.
FIG. 16 is a flowchart illustrating still object determination processing according to the third embodiment.
FIG. 17 is an explanatory diagram illustrating the principle of an FMCW radar.
FIG. 18 is an explanatory diagram showing a beat signal spectrum by the FMCW radar.
[Explanation of symbols]
2 ... Radar device 10 ... Transceiver
12: transmitter 12a: modulator
12b: voltage controlled oscillator 12c, 12d: power divider
12e: transmitting antenna 14, 16: receiver
14a, 16a ... receiving antenna 14b, 16b ... mixer
14c, 16c: Preamplifier 14d, 16d: Low-pass filter
14e, 16e: post-amplifier 20: signal processing unit
22 ... triangular wave generator 24a, 24b ... A / D converter
26: microcomputer 28: arithmetic processing unit

Claims (17)

A transmission unit that generates a transmission signal whose frequency gradually increases and decreases periodically with a predetermined modulation width, and transmits the transmission signal as a radar wave,
Receiving means for receiving the radar wave reflected by the target to generate a reception signal, and mixing the reception signal with the transmission signal to generate a beat signal;
A spectrum creating unit that creates an uplink spectrum from an uplink beat signal at the time of uplink modulation in which the frequency of the transmission signal increases, and creates a downlink spectrum from a downlink beat signal at the time of downlink modulation in which the frequency of the transmission signal decreases.
Detection means for shifting at least one spectrum peak of the upstream spectrum and the downstream spectrum by a predetermined frequency shift amount, comparing corresponding spectrum peaks of both spectra, and detecting a moving state of the target,
In the FMCW radar device provided with
Based on the speed of the vehicle equipped with the FMCW radar device, when setting the frequency shift amount, a plurality of shift amount setting means for setting a plurality of frequency shift amounts in consideration of measurement errors,
For each of the uplink spectrum and the downlink spectrum corresponding to each set frequency shift amount, an evaluation unit that evaluates the degree of spectrum matching,
Determining means for determining the highest frequency shift amount of the spectrum matching degree based on the evaluation result,
Using the up spectrum and the down spectrum corresponding to the determined frequency shift amount, a stillness determination unit that performs the stillness determination of the target,
An FMCW radar device comprising: When setting the frequency shift amount based on the speed of a vehicle equipped with the FMCW radar device, a basic frequency shift amount and a frequency shift amount deviated by a predetermined amount from the basic frequency shift amount are set. The FMCW radar device according to claim 1, wherein In the evaluation means, in the up spectrum and the down spectrum corresponding to each frequency shift amount, the spectrum matching degree is evaluated for each pair of corresponding spectrum peaks of both spectrums, and based on the evaluation of the spectrum matching degree for each spectrum peak, The FMCW radar device according to claim 1 or 2, wherein a spectrum matching degree for each frequency shift amount is evaluated. In the stationary determination means, in the up spectrum and the down spectrum corresponding to the determined frequency shift amount, for each of a pair of corresponding spectrum peaks of both spectra, the spectrum matching degree is evaluated, and based on the evaluation, the object is evaluated. The FMCW radar device according to any one of claims 1 to 3, wherein the stationary determination of the target is performed. The FMCW radar apparatus according to any one of claims 1 to 4, wherein the evaluation of the degree of spectrum matching is performed based on information in a frequency band having a predetermined width of a spectrum peak. The FMCW radar device according to any one of claims 1 to 5, wherein the evaluation of the degree of spectrum matching is performed based on amplitude of a spectrum peak and azimuth information of a target. The amplitude evaluation value based on the amplitude of the spectrum peak is set to Y, and the phase matching value based on the azimuth information of the target is set to X. The FMCW radar device according to claim 6, wherein The sum of the absolute values of the evaluation vectors related to the spectrum peaks to be evaluated is obtained for each spectrum peak, and the total sum of the sums of the sums having the smallest total spectrum is regarded as the true frequency shift amount. The FMCW radar device according to claim 7, wherein: Compare the neighborhood sum of the absolute value of the evaluation vector with respect to the spectrum peak to be evaluated with a predetermined threshold, and when the neighborhood sum is equal to or less than the threshold, determine the target as a stationary object. An FMCW radar device according to claim 7. The FMCW radar device according to any one of claims 1 to 9, wherein the frequency shift amount is set in consideration of a beam direction of the FMCW radar device. For a target already recognized as a moving object, the current moving position is predicted and a moving object prediction flag is set, and the moving object prediction flag is set for a spectrum peak to be determined this time. 11. The FMCW radar device according to claim 1, wherein the target is not determined to be a stationary object in such a case. A transmission unit that generates a transmission signal whose frequency gradually increases and decreases periodically with a predetermined modulation width, and transmits the transmission signal as a radar wave,
Receiving means for receiving the radar wave reflected by the target to generate a reception signal, and mixing the reception signal with the transmission signal to generate a beat signal;
A spectrum creating unit that creates an uplink spectrum from an uplink beat signal at the time of uplink modulation in which the frequency of the transmission signal increases, and creates a downlink spectrum from a downlink beat signal at the time of downlink modulation in which the frequency of the transmission signal decreases.
Detection means for shifting at least one spectrum peak of the upstream spectrum and the downstream spectrum by a predetermined frequency shift amount, comparing corresponding spectrum peaks of both spectra, and detecting a moving state of the target,
In the FMCW radar device provided with
Based on the speed of the vehicle equipped with the FMCW radar device, when setting the frequency shift amount, a beam shift amount setting means for setting the frequency shift amount in consideration of the direction of the radar beam,
Using the up spectrum and the down spectrum corresponding to the frequency shift amount, a stillness determination unit that performs the stillness determination of the target,
An FMCW radar device comprising: In the stillness determination means, in the up spectrum and the down spectrum corresponding to the frequency shift amount, the spectrum matching degree is evaluated for each of a pair of corresponding spectrum peaks of both spectra, and based on the evaluation, the stationary state of the target is determined. 13. The FMCW radar device according to claim 12, wherein the determination is performed. 14. The FMCW radar apparatus according to claim 13, wherein the evaluation of the degree of spectrum matching is performed based on information in a frequency band having a predetermined width of a spectrum peak. The FMCW radar device according to claim 8 or 9, wherein the evaluation of the degree of spectrum matching is performed based on amplitude of a spectrum peak and azimuth information of a target. The amplitude evaluation value based on the amplitude of the spectrum peak is set to Y, and the phase matching value based on the azimuth information of the target is set to X. The FMCW radar device according to claim 15, wherein Compare the neighborhood sum of the absolute value of the evaluation vector with respect to the spectrum peak to be evaluated with a predetermined threshold, and when the neighborhood sum is equal to or less than the threshold, determine the target as a stationary object. The FMCW radar device according to claim 16 .

Images