KR20210135753A - Method and apparatus for location estimation of lidar-based vehicles - Google Patents

Abstract

A method for a vehicle position estimation apparatus operated by at least one processor to estimate a position of a vehicle passing through a tunnel using lidar mounted in the vehicle, includes the following steps of: generating a reference linear pattern indicating a basic structure of an entered tunnel based on distance information and vertical angle information of point cloud data collected through lidar; receiving the point cloud data collected through the lidar by scan timing during the movement of a vehicle and generating a linear pattern based on distance information and vertical angle information of the point cloud data by scan timing; recognizing an abnormal pattern different from the reference linear pattern through a comparison between the reference linear pattern and the linear pattern by scan timing, and calculating height values with respect to a vertical line of the point cloud data corresponding to the abnormal pattern; and estimating a spot, at which a calculated height value is matched within an error range, as the same structure in the tunnel among abnormal patterns recognized at different scan timings, and calculating a distance change value between the spot estimated as the same structure and a position of the lidar as a travel distance of the vehicle in response to the different scan timings.

Description

LIDAR-based vehicle location estimation method and apparatus

The present invention relates to a technology for estimating a lidar-based vehicle location in a GPS shadow area.

In order to operate an autonomous vehicle or create a precise map for autonomous driving, high-accuracy vehicle location measurement technology is essential.

Accordingly, the position of the vehicle is measured using a global positioning system (GPS), which is one of the most accurate identification systems. However, there is a limit to maintaining the accuracy of location measurement while the vehicle is driving in a shaded area such as a forest of buildings, under an overpass, or a tunnel. In order to solve this problem, research is being conducted on a technology for installing a lidar sensor, a camera, a radar sensor, an inertial sensor, etc. on a vehicle or measuring the exact vehicle location using the vehicle's own driving information (On Board Diagnostics, OBD).

However, vehicle information including speed, accelerator strength, steering wheel rotation angle, and brake status measured in the vehicle itself can be checked through the vehicle's own OBD, but whether it is linked with the vehicle's own OBD It is difficult to apply universally because it is often difficult to provide information in the case of a vehicle manufactured for a different and special task.

In addition, when determining the geometry of the plane and the edge for a point cloud scanned with LiDAR and determining the movement path according to the distance and angle change over time, the feature point is to determine the geometry of the plane and the edge. to extract At this time, in a road or underground parking lot, the performance of location estimation is obtained by extracting the face and corner of the building as feature points, but in the case of a closed space such as a tunnel, the distinction between the face and the edge is ambiguous, and it is difficult to extract the feature points according to the movement change. As a result, the performance is very low.

Therefore, in the case of using LiDAR, most of them have high performance in an indoor space or a limited road environment. Measurement errors for distances or rotations will occur. In addition, it is difficult to maintain the safety of the performance of lane recognition based on the reflection characteristics of the lidar sensor because it is difficult to physically recognize the lane because there are many moving vehicles in the actual road environment.

In particular, in the case of a closed space such as a tunnel, the distinction between surfaces and edges is ambiguous, it is difficult to extract feature points according to movement changes, and unlike general roads that can recognize a large number of objects such as street trees, people, and sign boards, it is very limited. Therefore, there is a limit to the method of estimating the position through object recognition.

The problem to be solved is that when a vehicle equipped with a lidar enters a tunnel, the same structure in the tunnel is estimated based on the point cloud data obtained from the lidar for each scan time, and the same structure as the position of the lidar is corresponding to the scan time. To provide a method for estimating the location of a vehicle according to a distance change value between points estimated as .

According to an embodiment, there is provided a method for estimating the position of the vehicle passing through a tunnel using a lidar mounted on the vehicle by a vehicle position estimating device operated by at least one processor, wherein the point cloud data collected through the lidar is Creating a reference linear pattern representing the basic structure of the entered tunnel based on vertical angle information and distance information. While the vehicle is moving, the point cloud data collected through LiDAR is input at each scan time point, and the generating a linear pattern based on vertical angle information and distance information Calculating the height values, and estimating a point where the calculated height values match within the error range among abnormal patterns recognized at different scan times as the same structure in the tunnel, and the position of the lidar corresponding to the different scan times and calculating a distance change value to a point estimated by the same structure as the moving distance of the vehicle.

The point cloud data may include a scan time of the lidar, reflection intensity, distance information with the lidar, range of distance information, horizontal angle information, and vertical angle information for each point cloud data.

The generating of the linear pattern includes determining an area of the point cloud data in which distance information increases at regular intervals as the vertical angle information increases as the floor area, and connecting the point cloud data having an angle greater than or equal to the vertical angle information of the floor area. linear patterns can be obtained.

The generating of the linear pattern includes generating a linear pattern based on vertical angle information and reflection intensity of the point cloud data for each scan point, and calculating the height value for the vertical line of the point cloud data includes the reflection intensity of the reference linear pattern and the A linear pattern having different reflection intensities may be recognized as an abnormal pattern.

The step of calculating the height values for the vertical line of the point cloud data may include calculating the height values for the vertical line orthogonal to the horizontal line connected to the lidar based on the vertical angle information of the lidar and the distance information with the lidar for each point cloud data. can

In the step of calculating the moving distance of the vehicle, if distance information or reflection intensity of the point cloud data corresponding to the abnormal pattern has a constant ratio continuously, other point cloud data other than one point cloud data having the highest vertical angle information can be excluded. have.

In the step of generating the linear pattern, when the floor area is determined, information on the lane in which the vehicle travels can be estimated by estimating the cross ratio for left and right and the width of each lane based on the location of the vehicle in the floor area. can

In the generating of the linear pattern, when vertical angle information determined by the crossing ratio and the floor area is simultaneously changed, it may be determined that the driving lane is changed.

If distance information or reflection intensity has a constant ratio based on vertical angle information among point cloud data of an abnormal pattern recognized for each scan period set in the lidar, vertical angle information of the point cloud data, distance information from the lidar, and height of the point cloud data The method may further include calculating an interval for the point cloud data of the abnormal pattern using the values, and resetting the scan period of the lidar in proportion to the interval for the point cloud data.

As a program stored in a computer-readable storage medium and executed by a processor according to an embodiment, a reference line indicating a basic structure of an entered tunnel based on vertical angle information and distance information of point cloud data collected through lidar An operation of generating a pattern, an operation of receiving the point cloud data collected through the lidar at each scan time point while the vehicle is moving, and generating a linear pattern based on the vertical angle information and distance information of the point cloud data at each scan time point; An operation of recognizing an abnormal pattern different from the reference linear pattern by comparing the linear pattern with the reference linear pattern, calculating height values for vertical lines of point cloud data corresponding to the abnormal pattern, and among abnormal patterns recognized at different scan points The point where the calculated height values match within the error range is estimated as the same structure in the tunnel, and the distance change value between the position of the lidar and the point estimated to be the same structure in response to different scan times is calculated as the moving distance of the vehicle. Contains instructions that are described to perform an action.

The operation of obtaining the linear patterns is performed by determining an area of point cloud data in which distance information increases at regular intervals as vertical angle information increases as the floor area, and connecting point cloud data having an angle greater than or equal to the vertical angle information of the floor area. linear patterns can be obtained.

The reference linear pattern includes the scan timing of the lidar according to the basic structure of the tunnel, reflection intensity, distance information from the lidar, and horizontal or vertical angle information. A linear pattern having a reflection intensity different from the reflection intensity of the linear pattern may be recognized as an abnormal pattern.

In the operation of calculating the moving distance of the vehicle, if distance information or reflection intensity of the point cloud data corresponding to the abnormal pattern has a constant ratio continuously, other point cloud data other than one point cloud data having the highest vertical angle information can be excluded. have.

According to the present invention, it is possible to accurately estimate the moving distance and location of the vehicle in a closed space such as a tunnel without separately constructed map information by utilizing the lidar linked to the vehicle.

According to the present invention, actual movement in the GPS shaded area is estimated by selecting an anomaly pattern from linear patterns of point cloud data and estimating the movement distance for a part that is difficult to use in estimating the movement distance due to the ambiguous distinction between the face and the edge of the closed space. It can improve the accuracy of the distance.

1 is an exemplary diagram illustrating an apparatus for estimating a vehicle location of a vehicle equipped with a lidar sensor entering a tunnel according to an exemplary embodiment.
2 is a graph illustrating a correlation between vertical angle information and distance information stored in the collected point cloud data according to an exemplary embodiment.
3 is an exemplary diagram illustrating point cloud data scanned through a lidar sensor according to an exemplary embodiment.
4 is a flowchart illustrating a method of measuring a moving distance of a vehicle according to an exemplary embodiment.
5 is an exemplary diagram for explaining a process of determining a period of an abnormal pattern according to an exemplary embodiment.
6 is an exemplary diagram for explaining a process of excluding from a period of an abnormal pattern according to an exemplary embodiment.
7 is an exemplary diagram for explaining a process of calculating a moving distance of a vehicle according to an exemplary embodiment.
8 is a hardware configuration diagram of a computing device according to an embodiment.

Hereinafter, with reference to the accompanying drawings, the embodiments of the present invention will be described in detail so that those of ordinary skill in the art to which the present invention pertains can easily implement them. However, the present invention may be embodied in several different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part "includes" a certain element, it means that other elements may be further included, rather than excluding other elements, unless otherwise stated. In addition, terms such as “…unit”, “…group”, and “module” described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. have.

In the specification, a vehicle means any type of vehicle equipped with a LiDAR sensor, and here, it is assumed that the lidar is mounted on the front of the vehicle, but the description is not necessarily limited thereto.

LiDAR stands for Light Detection And Ranging or Laser Imaging, Detection and Ranging, which measures the time and intensity it takes for a pulsed laser to be emitted at a target and the light to return. It is a technology that detects distance, direction, speed, temperature, material distribution and concentration characteristics by measuring. LiDAR calculates the results of work performance as point cloud data such as so-called 'point clouds'.

The specification relates to a method of measuring the moving distance and location of a vehicle using a lidar in an area where a GPS signal is shielded, such as a tunnel, and it is assumed that the vehicle has entered the tunnel or is driving in the tunnel.

In general, the determination that the vehicle has entered the tunnel is a part that can be determined based on camera, illuminance sensor, GIS information, etc., and the criteria for determining whether or not to enter the tunnel are omitted.

Hereinafter, a lidar-based vehicle location estimation apparatus will be described in detail with reference to FIG. 1 .

1 is a structural diagram of an apparatus for estimating a vehicle position based on a lidar according to an embodiment.

As shown in FIG. 1 , the vehicle location estimation apparatus 100 is a computing device operated by at least one processor, and is connected to a lidar mounted in the vehicle through a wired/wireless network to transmit and receive data.

A computing device suffices as a device capable of executing a software program written to carry out the present invention.

The vehicle position estimating apparatus 100 collects a lidar point cloud (point cloud) obtained at each set scan time point from the lidar mounted on the vehicle.

In addition, the vehicle position estimation apparatus 100 may collect a frame scanned once for every scan time set by the lidar and point cloud data constituting the frame. Here, point IDs to be scanned in each location direction in the set up/down/left/right scan range of the lidar are designated, and the vehicle position estimating apparatus 100 provides point cloud data generated through one scan of the lidar. can be defined for each frame

And the set scan time of the lidar can be reset through communication with the vehicle location estimation apparatus 100 .

The vehicle position estimating apparatus 100 may determine the spatial shape by recognizing the bottom surface of the tunnel based on the vertical angle of the bottom of the tunnel and recognizing the inner side and upper side of the tunnel.

In this case, the vehicle position estimating apparatus 100 may automatically convert the angle, height, etc. derived differently in response to the position of the lidar mounted on each vehicle according to a preset reference value. In other words, the vehicle position estimation apparatus 100 may automatically perform calibration according to the installation height angle in order to eliminate the difference in the height angle of the lidar mounted on each vehicle.

In addition, the vehicle position estimating apparatus 100 may extract a shape according to a change in distance on a straight line according to a specific point to which vertical, horizontal, and vertical angles are previously designated based on the defined frame. In other words, the vehicle position estimation apparatus 100 generates linear patterns based on vertical angle information and distance information stored for each point cloud data.

Here, the vertical angle information is vertical angle information of the lidar, and the distance information means distance information between the point where the emitted laser is reflected and the lidar.

In other words, in the point cloud data collected through the lidar, the scan timing of the lidar, the reflection intensity, the distance information with the lidar, the range of the distance information, the horizontal angle information, and the vertical angle information are stored, and the vehicle position estimation apparatus 100 ) may generate a linear pattern using information stored in these point cloud data.

In this case, the vehicle position estimating apparatus 100 may generate a linear pattern generated based on vertical angle information and distance information of point cloud data as a reference linear pattern when first entering the tunnel.

In detail, the reference preceding pattern is information that serves as a reference for extracting feature points from inside the tunnel in which the vehicle moves in the longitudinal (front/back) direction. can be created with

In this case, the lidar may scan in the lateral direction to generate the reference linear pattern of the tunnel and then scan in the longitudinal direction after generating the reference linear pattern in the tunnel.

In other words, the vehicle position estimating apparatus 100 sets the scan direction of the lidar to scan in the lateral direction until the reference linear pattern is generated after the vehicle enters the tunnel, and when the reference linear pattern is generated, the scan of the lidar You can set the direction to vertical.

Meanwhile, the reference linear pattern may be information set in advance corresponding to the GPS closed area. For example, a reference linear pattern may be preset and stored for each tunnel, and when a vehicle enters a tunnel, it may access a device or a server to collect the stored reference linear pattern corresponding to the tunnel.

A configuration for generating or collecting such a reference linear pattern can be easily changed and designed by a user later.

In addition, a linear pattern may be generated in real time based on vertical angle information and distance information of the point cloud data for each scan point using the point cloud data collected for each scan point while the vehicle passes through the tunnel.

As such, the vehicle position estimating apparatus 100 compares the reference preceding pattern with the linear pattern generated in real time, and recognizes a preceding pattern that is different from the reference preceding pattern by more than an error range as an abnormal pattern.

The abnormal pattern is a pattern to be used as a reference point when calculating the movement distance in an environment where the distinction between surfaces and edges is ambiguous, such as in tunnel space, and it is difficult to extract feature points according to movement changes. .

In other words, the abnormal pattern refers to point cloud data that is reflected by a sign, a ventilation mechanism, and an installed structure that can become a singularity in the tunnel by the laser emitted from the lidar.

Therefore, the vehicle position estimating apparatus 100 may recognize the corresponding linear pattern as an abnormal pattern when distance information of the reference linear pattern and distance information of the linear pattern are different or reflection intensity is different based on the same vertical angle.

The vehicle position estimating apparatus 100 may recognize an abnormal pattern through comparison for each frame through one scan. In addition, the vehicle position estimating apparatus 100 may store a scan time of the recognized abnormal pattern, horizontal or vertical angle information of the lidar, distance information with the lidar, reflection intensity, and the like in an interlocked database.

In addition, the vehicle position estimating apparatus 100 calculates height values with respect to the vertical line of the point cloud data corresponding to the abnormal pattern.

Here, the height value is a value calculated by applying the vertical angle information of the lidar and the distance information to the lidar for each point cloud data to a trigonometric function, and means a height value for a vertical line orthogonal to the horizontal line connected to the lidar.

The vehicle position estimating apparatus 100 checks whether the calculated height values match within an error range (eg, +/-0.01 mm) set based on the precision value of the corresponding lidar. This error range can be set by the type, characteristic, precision, etc. of the interlocking lidar, and can be easily changed and set later by the use environment.

In other words, the vehicle position estimating apparatus 100 compares the abnormal pattern having the first scan time and the abnormal pattern having the second scan time, and estimates a point where the calculated height values match within the error range as the same structure in the tunnel. can do.

In addition, the vehicle position estimating apparatus 100 calculates a distance change value from a point estimated to be the same structure in response to different scan times as the moving distance of the vehicle.

In detail, the vehicle position estimating apparatus 100 applies the vertical angle information of the lidar, the distance information where the emitted laser is reflected and returned, and the height value of the reflection point to the trigonometric function to apply the position of the lidar and the horizontal line connected to the lidar It is possible to calculate the length between the intersections of vertical lines orthogonal to .

In this way, the vehicle position estimating apparatus 100 may calculate distance values between a point estimated to be the same structure at different time points and a lidar position, and calculate a change value of the calculated distance values as the moving distance of the vehicle. have.

In addition, the vehicle location estimation apparatus 100 may estimate location information on which the vehicle is located in the tunnel by applying the moving distance of the vehicle.

In addition, the vehicle position estimating apparatus 100 is based on the width of the lane and the floor area of the tunnel cut based on the vertical angle of the floor of the tunnel, the cross ratio of the left and right and the lane based on the position of the vehicle in the floor area According to the width, information on the driving lane of the vehicle may be estimated.

On the other hand, the lidar (not shown) linked to the vehicle includes all 3D lidar sensors, and generates point cloud data through a scan within a range of preset horizontal and vertical angles and transmits them to the vehicle position estimation apparatus 100 . do.

For example, the lidar can perform a scan by setting 360 degrees as a horizontal angle and 30 degrees as a vertical angle. can do.

Such a scan period can be reset through communication with the vehicle location estimation apparatus 100 .

2 is a graph illustrating a correlation between vertical angle information and distance information stored in the collected point cloud data according to an exemplary embodiment.

2 is a graph comparing vertical angle information and distance information of each point cloud data for one frame acquired by one scan when information scanned from the bottom of the vertical angle is extracted. FIG. 2 shows the relationship between vertical angle and distance information of point cloud data for a tunnel having a round upper end.

In general, since the bottom surface is formed to be flat, it is represented by a pattern in which the distance increases at regular intervals as the vertical angle increases under the premise that the bottom surface is flat.

In other words, like the region of FIG. 2 , the bottom surface area is formed in a straight line shape having a constant inclination. Although the actual floor area may be inclined, the floor area appears as an increasing straight line in a constant shape even in a situation where the floor area is inclined.

Therefore, as the vertical angle information increases, the vehicle position estimation apparatus 100 determines an area in which distance information is increased at regular intervals as the floor area.

Next, the vehicle position estimating apparatus 100 may identify the side and oblique fronts of the tunnel at angles greater than or equal to the vertical angle θ of the floor area.

For example, when the vehicle travels in the central region of the tunnel under the condition that the distance is the same, as shown in FIG. 3 , the distance information on the side of the tunnel shows a constant shape even though the vertical angle information increases.

When the shape of the wall of the tunnel is a semicircle, the same shape as shown in FIG. 2 appears.

On the other hand, for the upper front area of the tunnel, as shown in FIG. 2 , the distance information is reversely decreased at an angle equal to or greater than the vertical angle θ of the floor area.

The reason why the straight line of Da in FIG. 2 is shorter than my straight line is that data for the upper region directly above the vehicle cannot be obtained when scanning in the front direction due to the distance recognition limit of the lidar.

Accordingly, the starting point of the dotted arrow in FIG. 2 indicates the cross-section located the furthest, and the end point of the arrow indicates the cross-section relatively close to the lidar sensor.

At this time, depending on the shape of the tunnel, the area between me and Da in FIG. 2 may appear as a curved line or a straight line.

For example, if the vehicle is traveling in a lane other than the center or has a semi-elliptical shape of a tunnel, it may appear as a curved line, and if the tunnel has a rectangular shape, it may appear as a straight line.

As such, the vehicle position estimating apparatus 100 determines a floor area if the vertical angle information θ is less than or equal to the vertical angle information θ based on the vertical angle information θ determined as the floor area area, and when the vertical angle information θ is greater than the vertical angle information θ, the side surface of the tunnel It can be determined as a region or an upper frontal region.

In other words, the vehicle position estimating apparatus 100 may generate a reference linear pattern having a predetermined shape corresponding to the basic structure of the tunnel as shown in FIG. 2 , and may recognize different abnormal patterns based on the reference linear pattern.

Fig. 3 shows the point cloud data obtained from the actual lidar for such a pattern.

3 is an exemplary diagram illustrating point cloud data scanned through a lidar sensor according to an exemplary embodiment.

3 (a) is a tunnel having a rectangular shape, (b) is an image of a tunnel having a semicircular shape.

The vehicle location estimation apparatus 100 may collect point cloud data from the lidar in real time as shown in FIG. 3A . In addition, the vehicle location estimation apparatus 100 compares the reference linear pattern in the A->B direction or B->A direction based on the shape of the tunnel composed of the collected point cloud data.

Since the shape of FIG. 3 is based on the point at which the emitted laser pulse is reflected, the difference in the shape of the pattern means that the laser pulses emitted with constant vertical angle information are reflected on structures having different heights.

Therefore, the vehicle position estimating apparatus 100 compares the point cloud data of the upper front area AB recognized corresponding to the point cloud data of the lidar received in real time with the point cloud data forming the reference linear pattern, which is different than the error range. explore the part

Referring to (a) of FIG. 3 , the left linear patterns are the same as the right reference linear pattern in the region B, but it can be seen that the linear pattern is disconnected differently from the right reference linear pattern in the A region.

At this time, it can be estimated that each disconnected part is different from the reference linear pattern because it corresponds to a sign, a ventilation mechanism, a structure installed on the upper surface of the tunnel, and the like.

Similarly, the vehicle position estimation apparatus 100 compares the point cloud data of the upper front area AB located on the left and the point cloud data forming the reference linear pattern located on the right in the same manner as in FIG. Search for a part that is different than the error range.

In this case, the point cloud data forming the linear patterns has a reflection intensity in addition to distance information from the lidar, and when point cloud data having different reflection intensities is found, the corresponding pattern may be recognized as an abnormal pattern.

For example, in each point cloud data, information may be stored as shown in Equation 1 in response to a frame ID and a point ID according to a frame acquired through one scan.

[Equation 1]

Each frame has point cloud data, and information of scan time, point ID, range, and reflection intensity. And each point has its own horizontal and vertical angle information. In addition, each point cloud data includes distance information between the lidar sensor and the point that reflects the laser pulse emitted. Here, the horizontal angle or vertical angle information means angle information within the scan range of the lidar sensor.

4 is a flowchart illustrating a method of measuring a moving distance of a vehicle according to an exemplary embodiment.

As shown in FIG. 4 , the vehicle location estimation apparatus 100 generates a reference linear pattern representing the basic structure of the entered tunnel based on vertical angle information and distance information of the point cloud data collected through the lidar (S110). ).

First, the vehicle location estimation apparatus 100 determines the floor area of the tunnel through the point cloud data of the lidar. Since the height and horizontal degree of the vehicle may be different for each vehicle, the vehicle position estimating apparatus 100 must first recognize the floor area based on the threshold of the floor vertical angle.

In detail, the apparatus 100 for estimating the vehicle position may recognize an area having a pattern in which the distance increases at regular intervals as vertical angle information increases as the floor area.

In addition, the vehicle position estimating apparatus 100 generates a linear pattern based on the lidar vertical angle information of each point cloud data in the upper front area and the distance information with the lidar.

The vehicle position estimating apparatus 100 may store a pattern maintained for a preset time based on a linear pattern initially acquired when entering a corresponding tunnel as a reference linear pattern. Here, the reference linear pattern becomes a criterion for recognizing a singularity inside a tunnel such as a sign, a structure, a sign, an electric light, a traffic light, and the like.

Next, the vehicle position estimating apparatus 100 generates a linear pattern based on vertical angle information and distance information of point cloud data collected through the lidar for each scan time point while the vehicle is moving ( S120 ).

The vehicle position estimation apparatus 100 may generate a linear pattern using point cloud data having vertical angle information having an angle greater than or equal to the vertical angle θ of the floor area.

In other words, a linear pattern may be generated based on the upper central region excluding the bottom surface of the tunnel, and in some cases, a linear pattern may also be generated on the side region of the tunnel.

Next, the vehicle position estimating apparatus 100 compares the linear pattern for each scan point with the reference linear pattern, recognizes an abnormal pattern different from the reference linear pattern, and calculates height values for the vertical line of point cloud data corresponding to the abnormal pattern ( S130).

First, the vehicle position estimating apparatus 100 may recognize a linear pattern having different distance information or different reflection intensity based on the same vertical angle information as an abnormal pattern.

For example, in a tunnel installed with the same structure on the inner wall, even if the distance information is the same, if it has a different reflective intensity from the surroundings, such as tiles or glass, it is possible to measure the distance to a point showing different reflective intensity within a specific period range. Therefore, it can be a criterion for estimating the movement path. Therefore, when at least one of distance information or reflection intensity of the point cloud data is different, the vehicle position estimation apparatus 100 may recognize the corresponding linear pattern as an abnormal pattern.

In addition, the vehicle position estimating apparatus 100 calculates height values for a vertical line of the point cloud data by using the point cloud data forming each abnormal pattern.

The vehicle position estimation apparatus 100 applies the vertical angle information of the lidar and the distance information to the lidar for each point cloud data to a trigonometric function to obtain height values for a vertical line orthogonal to a horizontal line connected to the lidar from each point cloud data. can be calculated.

Meanwhile, the vehicle position estimating apparatus 100 checks whether distance information or reflection intensity from the lidar has a constant ratio based on the vertical angle information of the lidar of the point cloud data corresponding to the abnormal pattern.

When the distance information or the reflection intensity has a constant ratio based on the vertical angle information of the lidar, when the vertical angle information changes due to the movement of the vehicle, since the distance information changes by the corresponding ratio, each point cannot be specified.

In other words, when distance information or reflection intensity has a constant ratio based on vertical angle information of the lidar, height values for a vertical line orthogonal to a horizontal line connected to the lidar from each point cloud data have the same height.

Therefore, when the vehicle position estimating apparatus 100 continuously checks point cloud data having a constant ratio of distance information or reflection intensity, other point cloud data other than one point cloud data is excluded.

Next, the vehicle position estimating apparatus 100 calculates a distance change value between the position of the lidar and the point at which the calculated height value from among the abnormal patterns recognized at different scan points coincides within the error range as the moving distance of the vehicle. (S140).

The vehicle position estimating apparatus 100 estimates a point having the same height value among abnormal patterns recognized at successive or different scan points within a predetermined range as the same structure in the tunnel.

In this case, the vehicle position estimating apparatus 100 may estimate the same structure with one height value, but when some of the patterns of the corresponding height values match using one or more height values, one of the same height values is selected. Thus, the distance difference value can be calculated.

In other words, when the height value of the point cloud data is compared with the abnormal pattern calculated as 1-3-4-3-1 at the first time point and the abnormal pattern calculated as 4-3-1 at the second time point, the same point is In 4-3-1, since some patterns match, the moving distance from the first viewpoint to the second viewpoint may be calculated based on the height value 4 .

In addition, the vehicle position estimating apparatus 100 sets the estimated same structure as a reference point, and calculates a distance change value between the position of the lidar and the estimated point of the same structure.

In detail, the vehicle position estimating apparatus 100 applies the vertical angle information, the distance information of the lidar, and the height value of the reflection point to the trigonometric function to determine the distance between the position of the lidar and the intersection of the vertical line orthogonal to the horizontal line connected to the lidar. length can be calculated.

In other words, the vehicle position estimating apparatus 100 calculates a length between the position of the lidar and a point (intersection) estimated to be the same structure for each scan time point, and calculates a difference between the calculated lengths as the moving distance of the vehicle. .

Although the description has been made based on distance information of the vehicle for convenience of description, when a linear pattern having different reflection intensity as well as distance information is recognized, it may be recognized as an abnormal pattern capable of estimating a position.

Meanwhile, the vehicle position estimating apparatus 100 may infer information on the lane in which the vehicle currently travels by comparing the left and right sides by determining the ratio of the floor surfaces of both ends and the width of the lane.

At this time, when the lane is changed while the vehicle is moving and the refraction ratio of the reference cross-section of the wall surface is changed, the cross-section of the floor and the reference angle with respect to the floor are simultaneously changed.

Therefore, the vehicle position estimating apparatus 100 may recognize the lane change of the vehicle, generate a new wall reference cross-section, and continuously determine the left, right, front and rear movement distances.

Hereinafter, a process for recognizing an abnormal pattern and a configuration for calculating a moving distance of a vehicle will be described in detail with reference to FIGS. 5 to 7 .

For convenience of explanation, although structures above the tunnel are repeatedly illustrated in the same shape in FIGS. 5 to 7 , the shape or spacing of the structures is not necessarily limited.

5 is an exemplary diagram for explaining a process of determining a period of an abnormal pattern according to an embodiment, and FIG. 6 is an exemplary diagram for explaining a process of excluding from a period of an abnormal pattern according to an embodiment.

FIG. 5(a) is a diagram in which P (a point of an abnormal pattern) is shaped according to vertical angle information in one frame (point cloud data acquired through one scan), P(P1, P1', P2, P3) ) indicates the points reflected by the protruding structure. In addition, (b) of FIG. 5 is a graph showing point cloud data having distance information according to vertical angle information, and (c) is a graph showing height values of the point cloud data of (b).

Referring to (b), which is displayed based on the vertical angle information and distance information of points P1, P2, P1', P3 , and P1' that are reflected by the structure in (a) of FIG. 5 , the distance value is different from the reference linear pattern. It can be seen that they are marked with different points.

Through this, the vehicle position estimating apparatus 100 recognizes an abnormal pattern formed at different points. In (b), P1 and P1' have the same change ratio even if the distance information from the lidar is different, whereas P2 and P3 are different. have a shape

In detail, even if P1 and P1' are points detected by different structures, as the vehicle moves in a linear pattern with respect to structures located at the same height, even if point P1' is detected instead of point P1, point P1 and point P1' difficult to distinguish

In this case, the vehicle location estimation apparatus 100 may exclude the point P1' other than the point P1. However, since the point P2 or P3 is detected between the point P1 and the point P1', it may be distinguished in the form of a pattern including an adjacent point instead of a single point.

For example, when the points of P1, P2, P1', P3, and P1' are recognized at the first time point and the points of P1', P3, P1' are recognized at the second time point, the vehicle location estimation apparatus 100 is P1 at the first time point and P1' at the second time point may be recognized as the same structure, but may be recognized as different structures in consideration of the values of adjacent points (P2 and P3, respectively).

In other words, when distance information or reflection intensity of the point cloud data discontinuously has a constant ratio, it is possible to estimate whether the structure is the same including information on adjacent points.

Accordingly, P1'(H1)-P3(H3)-P1'(H1) having the same height value at the first time point and the second time point within the error range may be recognized as a common point sensed at each time point.

Therefore, the vehicle position estimating apparatus 100 may select one point from among the same points P1'-P3-P1' and calculate a distance change value between the corresponding point and the position of the lidar as the moving distance of the vehicle.

Meanwhile, the vehicle location estimation apparatus 100 may estimate how far the same structure is located through P1 and P1' in FIG. 5A . This can estimate the distance between P1 and P1' through vertical/horizontal angle information and distance information stored in each point cloud data transmitted from LiDAR.

In addition, even when repeated structures are spaced apart by several tens of meters and the corresponding structures are not scanned at the same time in one frame, the vehicle position estimation apparatus 100 stores time information, distance information, reflection information, or It is possible to estimate the distance of the same structure by using the vertical angle information of Ida, time information stored as an abnormal pattern in the next frame, distance information reflection positive information, or vertical angle information of lidar.

When the distance between the same structures is estimated in this way, the vehicle location estimation apparatus 100 may reset the scan period of the lidar as necessary to reduce power consumption and computational load of the lidar.

For example, when entering a tunnel, the vehicle location estimation apparatus 100 sets the scan period as the fastest first scan period to acquire abnormal patterns, and then sets distance information or reflection intensity at a constant rate for each scan period set in the lidar. By recognizing abnormal patterns having the same value, it is possible to estimate the spacing of the structures, which is the interval between the abnormal patterns. Then, the vehicle location estimation apparatus 100 may reset the scan period of the lidar in proportion to the distance between the structures.

In detail, if the distance between the structures is large, the vehicle location estimation apparatus 100 may reset the scan period of the lidar to a second scan period that is slower than the first scan period.

On the other hand, as shown in (a) of FIG. 6 , even if the meaningful patterns are different from the reference cross-sectional information, if they have the same shape except for the difference in the ratio, it is determined that they are the same structure. hard to be

6(a) shows that the abnormal patterns different from the reference linear pattern are all obtained at the height of a similar structure, and as shown in (b), the different degrees have a constant ratio and appear continuously, so the height according to each vertical angle information When the value is expressed, it appears as the same height H1 as in (c).

Therefore, when the point cloud data successively have the same distance information or reflection intensity at a constant ratio, the vehicle position estimating apparatus 100 excludes other patterns except for one of the abnormal patterns from the process of calculating the moving distance. In this case, the vehicle position estimating apparatus 100 may select one abnormal pattern having the highest vertical angle information.

In the case of FIG. 6 , in addition to the abnormal pattern corresponding to P1, other abnormal patterns indicated by P1' may be excluded.

Next, FIG. 7 is an exemplary diagram for explaining a process of calculating a moving distance of a vehicle according to an exemplary embodiment.

7(a) shows the point cloud data of P1, P2, and P1` obtained from the lidar when the vehicle is at the x1 point, and (b) is the P2 obtained from the lidar when the vehicle is at the x2 point. , represents the point cloud data for P1'.

And, (c) of FIG. 7 shows a state in which the vehicle moves from the point x1 to the point x2.

The vehicle location estimation apparatus 100 separately stores the point cloud data P1, P2, P1' included in the abnormal pattern determined at the point x1, and the point cloud data P2, P1' included in the abnormal pattern determined at the point x2. ) are stored separately.

In addition, the vehicle position estimating apparatus 100 compares the height values of the point cloud data included in the abnormal pattern at the x1 point or the x2 point to check the P2 point having the same height value within the error range.

Accordingly, the vehicle position estimating apparatus 100 applies the vertical angle information of the lidar stored in the point cloud data for the P2 point at the lidar scan time corresponding to the x1 point, the distance information with the lidar, and the height value (H2) to the trigonometric function. Input to calculate the distance value from point P2 to lidar. And the vehicle position estimating apparatus 100 calculates the vertical angle information of the lidar stored in the point cloud data for the P2 point at the lidar scan time corresponding to the x2 point, the distance information with the lidar, and the height value (H2) to the trigonometric function. Input to calculate the distance value from point P2 to lidar.

Accordingly, the apparatus for estimating the vehicle location may calculate the actual driving distance D of the vehicle as a difference between the distance value of X1-P2 and the distance value of X2-P2.

In addition, when the vehicle position estimation apparatus 100 uses a trigonometric function through distance information, reflection information, and vertical angle information of each point cloud data in FIG. can be obtained, and the structure spacing between P1 and P1` can also be obtained.

Meanwhile, the vehicle position estimating apparatus 100 may estimate the moving speed of the vehicle by using a difference value between the calculated moving distance of the vehicle and the time point of the lidar scan.

Since the vehicle location estimating apparatus 100 does not have information on the actual driving speed of the vehicle, it is difficult to set the lidar scan period based on the driving speed of the vehicle from the beginning or use it to calculate an abnormal pattern, but after estimating the speed A guideline for recognizing an abnormal pattern may be provided.

In general, in a situation in which the same abnormal patterns exist, recognition of abnormal patterns may vary according to a change in vehicle speed.

For example, when a vehicle travels at a slow speed in a tunnel, even if a characteristic structure is densely formed, a linear pattern for each structure is recognized, so even a dense structure may be an abnormal pattern. On the other hand, when the vehicle travels at a high speed, when the characteristic structures are dense, each structure cannot be recognized, and it may be difficult to recognize as an abnormal pattern even for the characteristic structures separated by a certain distance or more. .

As such, the relationship between the vehicle speed, the lidar scan period, and the characteristic structure repetition period will be described in detail using Table 1 below.

Table 1 shows that the repetition recognized as an anomaly pattern when scanning (predefined) is performed once between 0.2 and 0.05 seconds for the VLP-16 lidar model, assuming that the vehicle travels at a speed of 36 km/h to 108 km/h. This is a table measuring the minimum distance for the structure spacing.

speed
(km/h) speed
(m/s) LiDAR Minimum Scan Period(s) LiDAR Max Scan Period(s) Minimum structure repetition period (m) Maximum structure repetition period (m) 36 10 0.2 0.05 2 0.5 39.6 11 0.2 0.05 2.2 0.55 43.2 12 0.2 0.05 2.4 0.6 46.8 13 0.2 0.05 2.6 0.65 50.4 14 0.2 0.05 2.8 0.7 54 15 0.2 0.05 3 0.75 57.6 16 0.2 0.05 3.2 0.8 61.2 17 0.2 0.05 3.4 0.85 64.8 18 0.2 0.05 3.6 0.9 68.4 19 0.2 0.05 3.8 0.95 72 20 0.2 0.05 4 One 75.6 21 0.2 0.05 4.2 1.05 79.2 22 0.2 0.05 4.4 1.1 82.8 23 0.2 0.05 4.6 1.15 86.4 24 0.2 0.05 4.8 1.2 90 25 0.2 0.05 5 1.25 93.6 26 0.2 0.05 5.2 1.3 97.2 27 0.2 0.05 5.4 1.35 100.8 28 0.2 0.05 5.6 1.4 104.4 29 0.2 0.05 5.8 1.45 108 30 0.2 0.05 6 1.5

As shown in Table 1, as the moving speed of the vehicle increases, the scan interval naturally increases, so that the structure cannot be recognized in detail.

Therefore, as shown in Table 1, the interval (repetition period of the structure) at which structures appearing as abnormal patterns are repeatedly measured can be expressed as movement speed (m/s) * scan period (s).

In other words, if the movement speed is 10m/s and the maximum (fast) LiDAR scan period is multiplied by 0.05s, the maximum repeating structure period is 0.5m. This means that if the interval between the repeated structures is 0.5m or more, it can be recognized as an abnormal pattern. In addition, structures installed at intervals of 0.5 m or less, even if there is an actual separation, due to the movement speed and scan cycle, the lidar is judged as a cross-section rather than a structure, and it becomes a meaningless pattern.

And in Table 1, when the moving speed of 100 km/h or more and the lidar scan cycle is 0.2 seconds, the structure must be at least 6 m apart to be recognized as an abnormal pattern.

As such, when the vehicle position estimating apparatus 100 estimates the moving speed of the vehicle, the condition for an abnormal pattern for the interval of the repeating structure may be primarily set according to the set LiDAR scan period.

Based on the analysis result of Table 1, the vehicle position estimation apparatus 100 calculates the conditions for determining the abnormal pattern (a) and the meaningless pattern (b) for the moving speed of the vehicle, the lidar scan period, and the interval of the repeating structure using the following math It can be expressed as Equation 2.

[Equation 2]

Here, V_car(m/s) is the movement speed, Lidar_frequency(s) is the scan period of the lidar, and Pattern(m) is the structure (pattern) interval.

As shown in Equation 2, the vehicle position estimating apparatus 100 may first classify an abnormal pattern or a meaningless pattern by using the distance between the structures using the vehicle moving speed information and the lidar scan period information.

Equation 2 may be applied as a concept such as a guideline for discriminating an abnormal pattern or a meaningless pattern.

At this time, when the vehicle location estimating apparatus 100 collects vehicle driving information (OBD: On Board Diagnostics), when the vehicle speed is collected and the distance between structures is estimated, among the abnormal patterns satisfying the condition of Equation 2 An abnormal pattern according to step S130 may be recognized.

8 is a hardware configuration diagram of a computing device according to an embodiment.

Referring to FIG. 8 , the vehicle location estimation apparatus 100 is a computing device 300 operated by at least one processor, and executes a program including instructions described to execute the operation of the present invention. .

The hardware of the computing device 300 may include at least one processor 310 , a memory 320 , a storage 330 , and a communication interface 340 , and may be connected through a bus. In addition, hardware such as an input device and an output device may be included. The computing device 300 may be loaded with various software including an operating system capable of driving a program.

The processor 310 is a device for controlling the operation of the computing device 300 , and may be a processor 310 of various types for processing instructions included in a program, for example, a central processing unit (CPU), an MPU ( It may be a micro processor unit), a micro controller unit (MCU), a graphic processing unit (GPU), or the like. The memory 320 loads the corresponding program so that the instructions described to execute the operation of the present invention are processed by the processor 310 . The memory 320 may be, for example, read only memory (ROM), random access memory (RAM), or the like. The storage 330 stores various data, programs, etc. required for executing the operation of the present invention. The communication interface 340 may be a wired/wireless communication module.

The apparatus for estimating the vehicle location of the present invention can accurately estimate the moving distance and location of the vehicle in the tunnel, which is a closed space, by using only the lidar sensor.

In addition, the vehicle position estimating device is capable of accurately estimating the speed of the vehicle and the lane unit based on the obtained abnormal pattern of the lidar shape according to the moving distance and the time change value according to the recognition period.

In this way, the vehicle location estimation apparatus 100 can be utilized for autonomous driving or V2X by estimating the moving distance and driving lane information of the vehicle inside the tunnel.

In addition, it is easy to combine with the GPS and IMU attached to the existing autonomous vehicle, and through this, it is possible to estimate the location with high accuracy.

The embodiment of the present invention described above is not implemented only through the apparatus and method, but may be implemented through a program for realizing a function corresponding to the configuration of the embodiment of the present invention or a recording medium in which the program is recorded.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improved forms of the present invention are also provided by those skilled in the art using the basic concept of the present invention as defined in the following claims. is within the scope of the right.

Claims (13)

A method for estimating the location of the vehicle passing through a tunnel using a lidar mounted on the vehicle, by a vehicle location estimation device operated by at least one processor,
generating a reference linear pattern representing a basic structure of an entered tunnel based on vertical angle information and distance information of the point cloud data collected through the lidar;
receiving the point cloud data collected through the lidar at each scan time point while the vehicle is moving, and generating a linear pattern based on vertical angle information and distance information of the point cloud data for each scan time point;
Comparing the linear pattern for each scan point with the reference linear pattern, recognizing an abnormal pattern different from the reference linear pattern, and calculating height values for a vertical line of point cloud data corresponding to the abnormal pattern; and
A point in which the calculated height value coincides within an error range among abnormal patterns recognized at different scan times is estimated to be the same structure in the tunnel, and the position of the lidar and the same structure estimated to be the same structure in response to different scan times are estimated. calculating a distance change value from a point as a moving distance of the vehicle
A vehicle location estimation method comprising a. In claim 1,
The point cloud data are
A vehicle position estimation method including a scan time point of the lidar, reflection intensity, distance information with respect to the lidar, a range of the distance information, horizontal angle information, and vertical angle information for each point cloud data. In claim 2,
The step of generating the linear pattern comprises:
determining an area of point cloud data in which distance information increases at regular intervals as the vertical angle information increases as the floor area;
A vehicle position estimation method for obtaining linear patterns connecting point cloud data having an angle greater than or equal to the vertical angle information of the floor area. In claim 3,
The step of generating the linear pattern comprises:
A linear pattern is generated based on the vertical angle information and reflection intensity of the point cloud data for each scan point,
Calculating a height value for a vertical line of the point cloud data includes:
A vehicle position estimation method for recognizing a linear pattern having a reflection intensity different from that of the reference linear pattern as an abnormal pattern. In claim 4,
The step of calculating height values for the vertical line of the point cloud data includes:
A vehicle position estimation method for calculating height values of a vertical line orthogonal to a horizontal line connected to the lidar based on vertical angle information of the lidar and distance information from the lidar for each of the point cloud data. In claim 5,
Calculating the moving distance of the vehicle comprises:
When distance information or reflection intensity of the point cloud data corresponding to the abnormal pattern has a continuous constant ratio, the vehicle position estimation method excludes other point cloud data other than one point cloud data having the highest vertical angle information. In claim 3,
The step of generating the linear pattern comprises:
When determining the floor area, the vehicle position estimation method for estimating lane information in which the vehicle travels by estimating the cross ratio for left and right and the width of each lane on the basis of the location of the vehicle in the floor area . In claim 7,
The step of generating the linear pattern comprises:
When the vertical angle information determined by the crossing ratio and the floor area is changed at the same time, the vehicle position estimation method for determining that the driving lane is changed. In claim 1,
If distance information or reflection intensity has a constant ratio based on vertical angle information among point cloud data of an abnormal pattern recognized for each scan period set in the lidar, vertical angle information of the lidar and distance information of the lidar of the point cloud data And calculating an interval for the point cloud data of the abnormal pattern using the height values, and
The method of estimating the vehicle position further comprising the step of resetting the scan period of the lidar in proportion to the interval for the point cloud data. A program stored in a computer-readable storage medium and executed by a processor, comprising:
An operation of generating a reference linear pattern representing the basic structure of an entered tunnel based on vertical angle information and distance information of point cloud data collected through lidar;
An operation of receiving the point cloud data collected through the lidar at each scan time point while the vehicle is moving and generating a linear pattern based on vertical angle information and distance information of the point cloud data at each scan point point;
Comparing the linear pattern for each scan point with the reference linear pattern, recognizing an abnormal pattern different from the reference linear pattern, and calculating height values for a vertical line of point cloud data corresponding to the abnormal pattern; and
A point in which the calculated height value coincides within an error range among abnormal patterns recognized at different scan times is estimated to be the same structure in the tunnel, and the position of the lidar and the same structure estimated to be the same structure in response to different scan times are estimated. Calculating a distance change value from a point as a moving distance of the vehicle
A program comprising instructions written to execute a program. in paragraph 10
The operation of obtaining the linear patterns is,
determining an area of point cloud data in which distance information increases at regular intervals as the vertical angle information increases as the floor area;
A program for obtaining linear patterns connecting point cloud data having an angle greater than or equal to the vertical angle information of the floor area. In claim 11,
The reference linear pattern includes a scan timing of the lidar according to the basic structure of the tunnel, reflection intensity, distance information from the lidar, and horizontal or vertical angle information,
The operation of calculating a height value for a vertical line of the point cloud data includes:
A program for recognizing a linear pattern having a reflection intensity different from that of the reference linear pattern as an abnormal pattern. In claim 12,
The operation of calculating the moving distance of the vehicle is,
A program for excluding other point cloud data other than one point cloud data having the highest vertical angle information when distance information or reflection intensity of the point cloud data corresponding to the abnormal pattern has a constant ratio continuously.

Images