JP2019087829A - Decoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a decoding device capable of more accurately decoding data encoded by compression using compressed sensing. SOLUTION: A compressed data receiving unit 241 receives compressed data obtained by converting an N-dimensional vector as original data into M (M <N) dimensions using a predetermined compression matrix at any time. The N-dimensional vector before compression is, for example, data showing the positions of objects existing around the source in a grid map format (hereinafter referred to as a grid map). The distribution estimation unit 246 estimates the current position of each object in the vicinity of the transmission source based on the grid map obtained by restoring the compressed data received in the past. The filter generation unit 247 generates an N-dimensional filter vector based on the estimation result of the distribution estimation unit 246. When the filter generation unit 247 can generate the filter vector, the decoding processing unit 242 decodes the compressed data using the filter vector and the compression matrix. [Selection diagram] FIG. 5

Description

  The present disclosure relates to a decoding apparatus that decodes data encoded using a compressed sensing technique.

  In recent years, compressed sensing technology has attracted attention as a compression technology that can be applied to data having sparsity. For example, Patent Document 1 discloses a system in which a captured image of a vehicle-mounted camera compressed using a compression sensing technology is shared by inter-vehicle communication.

  In addition, as a known decoding algorithm used in the technical field of compressed sensing, Matching Pursuit (MP), Orthogonal Matching Pursuit (OMP), Iterative Hard Thresholding (IHT), Weak MP, LS-OMP, Basis Pursuit (BP) Various algorithms such as Least Angle Regression (LARS) and Iterated-Reweighted-Least-Squares (IRLS) Approximate Message Passing (AMP) have been proposed. MP, OMP, IHT, Weak MP, and LS-OMP are decoding algorithms applicable to the greedy method. BP, IRLS, and LARS are a type of convex relaxation method that approximates a cost function as a convex function.

Unexamined-Japanese-Patent No. 2013-145954

  Decoding processing in compressed sensing corresponds to an optimization problem for which an optimal solution can not be obtained unless all combinations are tried, and is NP-hard. Therefore, as described above, OMP, IHT, etc. have been proposed as practical decoding algorithms. However, in these heuristics, it is assumed that the output solution contains an error. Therefore, as a decoding apparatus used in the field of compressed sensing technology, a configuration for restoring original data with higher accuracy is required.

  The present disclosure has been made based on the above circumstances, and an object of the present disclosure is to provide a decoding device that decodes compressed and encoded data more accurately using compressed sensing.

  A decoding device according to the present disclosure for achieving the object is a decoding device that decodes compressed data that is data that is compressed using compressed sensing, and the compressed data has temporal continuity. While pre-compressed original data obtained by converting state quantity data, which is data indicating a state at a certain point in time with respect to a dynamically changing predetermined state quantity, into one column vector having a predetermined number of elements, It is data compressed and encoded by a predetermined compression matrix, and the compressed data receiving unit (241) that receives compressed data sequentially from another device, and the compressed data that is received each time the compressed data receiving unit receives compressed data. A decoding processing unit (242) that generates decoded data having the number of elements by executing a predetermined decoding algorithm on data, and based on the decoded data generated by the decoding processing unit And an estimation unit (246) for estimating the current state of the state quantity based on a plurality of state quantity data generated by the restoration unit within a predetermined period of time. A filter generation unit (247) for generating a filter vector having one non-zero element at a position corresponding to the current state of the state quantity estimated by the distribution estimation unit, which is one column vector including a number; When the filter generation unit can generate the filter vector, the decoding processing unit generates an estimated solution corresponding to the original data using the filter vector, the compressed data, and the compression matrix as a decoding algorithm, and also generates an estimated solution. The residual, which is an index indicating the degree of deviation between the original and the original data, is calculated, and the process of correcting the estimated solution using the filter vector and the compression matrix until the predetermined termination condition is satisfied is repeated. It is configured to generate the decoded data using an algorithm to implement.

  The decoding processing unit included in the above configuration decodes the compressed data using the filter vector generated by the filter generation unit when the filter generation unit can generate the filter vector. The filter vector used here is a vector having non-zero elements at positions according to the current state of the state quantity estimated by the distribution estimation unit. According to another aspect, such a filter vector corresponds to a vector indicating the result of estimating the position of the non-zero element in the received uncompressed data of the compressed data.

  That is, the above configuration corresponds to a configuration in which the compressed data is decoded using the result of estimating the position of the non-zero element in the received data of the compressed data before compression. Further, according to the configuration in which decoding is performed using the result of estimating the position of the nonzero element in the data before compression, information on the position at which the nonzero element is arranged in the data before compression of the received compressed data It can be expected that the decoding accuracy is higher than in the configuration in which the decoding is performed without using anything. Therefore, according to the above configuration, it is possible to more accurately decode data that has been compression-encoded using compression sensing. The case where the filter generation unit can generate the filter vector corresponds to the case where there is at least one state quantity data generated by the restoration unit within a predetermined period of time in the past. This is because generation of the filter vector requires at least one of a plurality of state quantity data generated by the restoration unit within a predetermined time in the past.

  In addition, the code | symbol in the parentheses described in the claim shows correspondence with the specific means as described in embodiment mentioned later as one aspect, Comprising: The technical scope of this indication is limited. is not.

BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram which shows schematic structure of the communication system which concerns on this embodiment. FIG. 1 is a block diagram showing a schematic configuration of a transmission side system 1; It is a figure for demonstrating surrounding object data. It is a figure which shows the matrix corresponding to the surrounding object data shown in FIG. FIG. 2 is a block diagram showing a schematic configuration of a receiving side system 2; It is a conceptual diagram for demonstrating the action | operation of the object tracking part 245. FIG. It is a conceptual diagram for demonstrating the operation | movement of the distribution estimation part 246. FIG. It is a figure which shows an example of an occupancy probability map. It is a figure which shows an example of an occupancy probability map. It is a figure which shows an example of an occupancy probability map. It is a flowchart for demonstrating the decompression | restoration process which the decoding apparatus 24 implements. It is the figure which put together OMP. It is the figure which put together the filter combined OMP as a proposal method. It is the figure which put together the filter combination IHT as a proposal method.

  Hereinafter, embodiments of the present disclosure will be described using the drawings. FIG. 1 is a diagram illustrating an example of a schematic configuration of a communication system according to the present disclosure. As shown in FIG. 1, the communication system includes a transmission side system 1 mounted on a vehicle Ma and a reception side system 2 mounted on a vehicle Mb. For convenience, the vehicle Ma on which the transmission side system 1 is mounted is also referred to as a transmission vehicle Ma. Moreover, the vehicle Mb carrying the receiving side system 2 is described also as the receiving vehicle Mb. When the transmitting vehicle Ma and the receiving vehicle Mb are not distinguished from one another, they are described as vehicles. Although only one transmission side system 1 is illustrated in FIG. 1 for convenience, a plurality of transmission side systems 1 may exist. Along with this, there may be a plurality of transmission vehicles Ma. Similarly, a plurality of receiving systems 2 and receiving vehicles Mb may be present. The transmitter system 1 corresponds to another device for the receiver system 2.

  The receiving vehicle Mb is a vehicle that travels on the road. The receiving vehicle Mb may be a four-wheeled vehicle, a two-wheeled vehicle, a three-wheeled vehicle or the like. Motorcycles also include motor bikes. In the present embodiment, the receiving vehicle Mb is a four-wheeled vehicle as an example. Similarly to the receiving vehicle Mb, the transmitting vehicle Ma is also a vehicle traveling on the road, and the specific type thereof is not limited. As an example, the transmission vehicle Ma is also a four-wheeled vehicle.

  The transmitting side system 1 and the receiving side system 2 are configured to be capable of short-range communication using radio waves of a predetermined frequency band. Here, the short range communication is direct wireless communication not via the wide area communication network. The frequency band used for short range communication is, for example, the 760 MHz band. In addition, 2.4 GHz and 5.9 GHz bands can also be used. A standard that defines the frequency, modulation scheme, and the like for realizing short-range communication can adopt any standard.

  Here, as an example, based on the fact that both the transmitting side system 1 and the receiving side system 2 are mounted on a vehicle, the short range communication is the standard of WAVE (Wireless Access in Vehicular Environment) disclosed in IEEE 1609 etc. Shall be implemented in accordance with. WAVE is a standard for direct wireless communication between vehicles (so-called inter-vehicle communication). The wireless communication between the transmitting system 1 and the receiving system 2 corresponds to inter-vehicle communication according to another aspect.

  Each of the plurality of vehicles sequentially broadcasts a communication packet (hereinafter referred to as a vehicle information packet) indicating vehicle information such as current position, traveling speed, traveling direction, etc. by inter-vehicle communication as short-range communication The vehicle information packet transmitted from the vehicle is sequentially received. The vehicle information packet includes information such as transmission time of the communication packet and transmission source information in addition to the vehicle information. The transmission source information is an identification number (so-called vehicle ID) assigned to a vehicle corresponding to the transmission source.

  Further, the transmitting system 1 and the receiving system 2 are the positions of an object such as a vehicle, a pedestrian, or a falling object existing around the transmitting vehicle Ma (in other words, the transmitting system 1) by inter-vehicle communication as short range communication. It is configured to share peripheral object data in the form of a grid map. The peripheral object data is, in other words, data indicating the traffic condition and the traveling environment around the transmission vehicle Ma. Traffic conditions change dynamically with temporal continuity. That is, the traffic condition around the transmission side system 1 corresponds to a dynamically changing state quantity with temporal continuity. Further, the traveling environment (the positional relationship with the falling object or the structure, etc.) of the transmission side system 1 corresponds to a state quantity dynamically changing with continuity as the transmission vehicle Ma moves. Therefore, peripheral object data corresponds to state quantity data. Here, the periphery of the transmission vehicle Ma corresponds to a predetermined range determined based on the current position of the transmission vehicle Ma.

<About the Configuration of Sender System 1>
As shown in FIG. 2, the transmission side system 1 includes a short range communication module 11, a periphery monitoring sensor 12, a grid map generation unit 13, and a compression processing unit 14. The short range communication module 11 is a communication module for performing the short range communication described above. The short range communication module 11 is communicably connected to the compression processing unit 14. The short range communication module 11 is also communicably connected to the grid map generation unit 13. The short range communication module 11 modulates the data input from the compression processing unit 14 and the like, and wirelessly transmits the data via an antenna (not shown). In addition, when receiving the vehicle information packet transmitted from the other vehicle, the short range communication module 11 provides the grid map generation unit 13 with data (hereinafter, other vehicle data) indicated in the vehicle information packet.

  The periphery monitoring sensor 12 is a sensor that detects an object present around the transmission vehicle Ma. The periphery monitoring sensor 12 may be, for example, a periphery surveillance camera for imaging a predetermined range around the transmission vehicle Ma, a millimeter wave radar for transmitting a survey wave to a predetermined range around the transmission vehicle Ma, a sonar, LIDAR (Light Detection and Ranging / Laser Imaging) Detection and Ranging) sensor. The periphery surveillance camera is a device that outputs a captured image to be captured sequentially as sensing information. A sensor that transmits a search wave such as sonar, millimeter wave radar, LIDAR or the like is a sensor that outputs, as sensing information, a scan result based on a reception signal y obtained when a reflection wave reflected by an obstacle is received. The sensing information output from the surrounding area monitoring sensor 12 is input to the grid map generation unit 13.

  The grid map generation unit 13 sequentially specifies the position of an object present around the transmission side system 1 based on the sensing information output by the surrounding area monitoring sensor 12. As a more preferable aspect, the grid map generation unit 13 of the present embodiment specifies the position of an object present around the transmission side system 1 also using other vehicle data provided from the short range communication module 11. And as shown to (B) of FIG. 3, the data (namely, periphery object data) which represented the position of the object which exists around transmission side system 1 in a grid map format are produced | generated. The objects here include pedestrians as well as vehicles. The surrounding object data can function as data indicating a space in which the vehicle can travel, in other words.

  The peripheral object data is data in which a predetermined range determined based on the transmitting side system 1 is divided into a plurality of cells, and whether or not an object is present in each cell is represented by a numerical value, a color, a hatching pattern, etc. is there. (A) of FIG. 3 is a conceptual diagram schematically showing the traffic situation around the transmission side system 1 showing the degree of deviation between the estimated solution and the original data before compression, and (B) of FIG. It is surrounding object data corresponding to the traffic condition illustrated to (A) of 2. The hatched cells represent cells in which an object is present (hereinafter, occupied cells). The non-hatched cells represent cells in which no object exists (hereinafter, blank cells). In FIG. 3B, whether or not an object exists is represented by the presence or absence of hatching.

  The number of rows, the number of columns, and the total number of cells (hereinafter, the total number of cells) N constituting the grid map as peripheral object data may be appropriately designed according to the size of the range represented by the grid map, etc. . Here, as an example, it is assumed that the grid map is configured by cells of six rows and ten columns. That is, the total number N of cells is 6 × 10 = 60. The size of the real space represented by the grid map may be appropriately designed according to the range considered to be the periphery of the transmission vehicle Ma.

  The size of the real space represented by the unit cell may also be designed appropriately. Here, as an example, one cell corresponds to a 2 meter square real space. Since the grid map of the present embodiment includes six lines in the vehicle width direction and ten rows in the vehicle longitudinal direction, the grid map is within a rectangular range of 12 m in the vehicle width direction and 20 m in the vehicle longitudinal direction in real space. It functions as data indicating the position of an existing object.

  Here, in order to simplify the description, peripheral object data is data representing two states of whether or not an object exists in each cell, but the present invention is not limited to this. In another aspect, the surrounding object data may represent the probability that an object exists in each cell. In that case, a numerical value (for example, a decimal such as 0.5) indicating the probability that an object exists may be inserted in each cell.

  Also, the grid map as the peripheral object data may be represented by a matrix as shown in FIG. 4, for example. The matrix as peripheral object data comprises elements corresponding to each cell of the grid map. In the matrix as peripheral object data, the value of the element corresponding to the blank cell is set to 0, and the element corresponding to the occupied cell is set to a non-zero value (here, 1). That is, an element for which a non-zero value is set in the matrix shown in FIG. 4 corresponds to an occupied cell which is a hatched cell in the grid map shown in FIG. 3B.

  The grid map as the peripheral object data generated by the grid map generation unit 13 is provided to the compression processing unit 14. The grid map generation unit 13 specifies the position of an object present around the transmission side system 1, and sequentially performs processing of generating a grid map as peripheral object data at predetermined generation intervals. The generation interval may be set to a value that does not significantly change the traffic condition around the transmission side system 1. For example, the generation interval is preferably set to a value within 3 seconds, such as 100 milliseconds, 200 milliseconds, 500 milliseconds, 1 second, 2 seconds, and so on. Here, as an example, the generation interval is set to 250 milliseconds. The generation interval may be appropriately designed in accordance with the size of the real space represented by the grid map.

  The compression processing unit 14 generates compressed data, which is data obtained by compression encoding a grid map as peripheral object data, using a predetermined compression matrix (so-called observation matrix). Specifically, it is as follows. The compression processing unit 14 converts the grid map as the peripheral object data into one column vector x having the number of elements (in other words, the number of dimensions) equal to the total number N of cells. The transformation from the grid map to the column vector may be realized by linking column vectors constituting the grid map in order of the column number. Of course, as another aspect, the column vector x may be generated by transposing an N-dimensional row vector formed by connecting row vectors constituting the grid map in row number order.

  That is, the compression processing unit 14 converts a grid map as peripheral object data into an N-dimensional vector x. This N-dimensional vector x itself is also data indicating the distribution of the object, and corresponds to the original data in compressed sensing. An N-dimensional vector x formed by converting a grid map as peripheral object data corresponds to pre-compression original data. Then, the compression processing unit 14 generates an M-dimensional (M <N) vector y obtained by linearly converting the N-dimensional vector x using a predetermined compression matrix. The compression matrix A is an M × N matrix, and an M-dimensional vector y corresponds to compressed data. The compressed data corresponds to linear transformation of a grid map as peripheral object data using a predetermined compression matrix A. The compression processing unit 14 outputs the generated compressed data to the short range communication module 11. As a result, compressed data is broadcasted in a cycle corresponding to the generation interval.

  The compressed data may be transmitted in a state where information indicating the number k of non-zero elements included in the peripheral object data before compression is added. According to such an aspect, the receiving system 2 can specify the non-zero element number k of the received data. In view of such circumstances, as a more preferable aspect in the present embodiment, it is assumed that information indicating the number k of non-zero elements included in peripheral object data before compression is added to compressed data as transmission data. Of course, the compressed data as transmission data may not have the information indicating the non-zero element number k.

  Each of the grid map generation unit 13 and the compression processing unit 14 is provided in the transmission side system 1 as a functional block which is realized by the CPU (not shown) executing predetermined software. The grid map generation unit 13 and the compression processing unit 14 may be realized as hardware. The aspect implemented as hardware also includes an aspect implemented using one or more ICs.

<Regarding Configuration of Receiving-side System 2>
Next, the configuration of the receiving system 2 will be described with reference to FIG. As shown in FIG. 5, the reception side system 2 includes a short range communication module 21, a map data storage unit 22, a locator 23, a decoding device 24, and a grid map storage unit 25. The short range communication module 21 is a communication module for performing the short range communication described above. The short range communication module 21 is communicably connected to the decoding device 24. When the short range communication module 21 receives the compressed data sequentially transmitted from the transmission side system 1, the short range communication module 21 provides the compressed data to the decoding device 24. When the short-range communication module 21 receives a vehicle information packet transmitted from another vehicle, the short-range communication module 21 decodes the data (hereinafter referred to as other vehicle data) indicated in the vehicle information packet. Provided to a predetermined ECU (Electronic Control Unit) installed in the

  The map data storage unit 22 is a non-volatile memory storing map data indicating the connection relationship of roads and the like. The map data has, for example, node data on a point (hereinafter referred to as a node) at which a plurality of roads intersect, merge or branch, and link data on a road (hereinafter referred to as a link) connecting the points. The link data is data including node coordinates (latitude / longitude) of the start end and the end of the link, as well as the length and width of the link, the number of lanes, the road type and the like. The map data may be configured to include a three-dimensional map including a road shape and a point group of feature points of a structure. The receiving system 2 may be configured to acquire map data from an external server or the like.

  The locator 23 is a device that calculates the current position of the receiving vehicle Mb. The locator 23 includes, for example, a GNSS (Global Navigation Satellite System) receiver and an inertial sensor. The GNSS receiver receives positioning signals from multiple satellites. The inertial sensor includes, for example, a gyro sensor and an acceleration sensor. The locator 23 sequentially measures the current position of the receiving vehicle Mb by combining the positioning signal received by the GNSS receiver and the measurement result of the inertial sensor. The locator 23 may be configured to complementarily use the vehicle speed information and the travel distance, etc., which are sequentially output from the vehicle speed sensor, for positioning. The positioning result of the locator 23 is provided to the decoding device 24.

  The decoding device 24 is configured to generate decoded data by executing a predetermined decoding algorithm on the compressed data received by the short range communication module 21. The decoded data is data obtained by decoding (in other words, restoring) an M-dimensional vector as compressed data into an N-dimensional vector. The decoded data indirectly show the distribution of the object in the vicinity of the transmitting vehicle Ma, similarly to the original data represented as the N-dimensional vector x.

  The decoding device 24 is realized using a computer provided with a CPU, a RAM, a flash memory and the like. The decoding device 24 includes a compressed data reception unit 241, a decoding processing unit 242, a grid map restoration unit 243, a storage processing unit 244, an object tracking unit 245, a distribution estimation unit 246, and a filter generation unit 247 as finer functions. Each function with which the decoding device 24 is provided is realized by a CPU (not shown) executing predetermined software. Note that part or all of the functions of the decoding device 24 may be realized as hardware. The aspect implemented as hardware also includes an aspect implemented using one or more ICs.

  The compressed data receiving unit 241 is configured to acquire the compressed data received by the short range communication module 21. Here, as described above, the compressed data is an M-dimensional vector y obtained by compression encoding of an N-dimensional vector indicating the distribution of an object around the transmission vehicle Ma using the compression matrix A.

  The decoding processing unit 242 is configured to decode compressed data using a predetermined decoding algorithm. The details of the operation of the decoding processing unit 242 will be described later separately. The decoding processing unit 242 generates a N-dimensional vector z corresponding to an estimated solution of the original data from the M-dimensional vector as the received data by executing a predetermined decoding algorithm. This N-dimensional vector z corresponds to the decoded data for the received compressed data. The N-dimensional vector z as decoded data generated by the decoding processing unit 242 is provided to the grid map restoring unit 243.

  The grid map restoration unit 243 converts the N-dimensional vector z into a grid map. Thereby, the grid map as peripheral object data generated by the transmission side system 1 is restored. The conversion from the N-dimensional column vector z to the grid map (in other words, restoration) may be performed in the reverse procedure of the conversion from the grid map to the N-dimensional column vector. The grid map restoration unit 243 corresponds to a restoration unit.

  The storage processing unit 244 associates the grid map generated by the grid map restoration unit 243 with information indicating the transmission source of the grid map (in fact, the compressed data thereof) (that is, transmission source information) and reception time information, It is configured to be stored in the map storage unit 25.

  The object tracking unit 245 assigns a unique object number to each object shown in the received grid map, and tracks the same object by using a known object tracking method. The received grid map refers to the grid map reproduced from the compressed data received by the compressed data receiving unit 241 in cooperation with the grid map restoring unit 243 and the decoding processing unit 242 (in other words, cooperation). .

  FIG. 6 is a diagram conceptually showing the operation of the object tracking unit 245. As shown in FIG. The ID in the figure represents an object number. Thus, each object is identified using the object number, and the previously received grid map is associated with the object shown in the newly received grid map. Then, based on the degree of change of the position of the same object between a plurality of consecutive frames, the relative movement direction and movement speed of the object are specified. The moving direction and moving speed of each object may be held in association with the object number.

  The distribution estimation unit 246 estimates the current position of an object present around the transmission vehicle Ma based on a plurality of grid maps from the same transmission vehicle Ma generated by the grid map restoration unit 243 within a predetermined period of time in the past. . The distribution estimation unit 246 corresponds to an estimation unit.

  For example, as shown in FIG. 7, when the object to which the object number 001 is assigned moves one cell to the right on the paper surface every time the object having the object number 001 is received as shown in FIG. It is estimated that the position has been moved by one cell to the right of the position shown in the grid map. When an object moves 2 cells to the right on the paper surface every time a grid map is received, the current position moves 2 cells to the right from the position shown in the previously received grid map. It is estimated that In the figure, t represents the reception time of the previous grid map, and t-1 represents the reception time of the last previous grid map. In FIG. 7, t + 1 represents the current time.

  The distribution estimation unit 246 estimates the current position of an object present in a predetermined range around the transmission vehicle Ma by performing the above-described estimation processing of the current position for each object. The current position of each object can be estimated based on the moving direction and the moving speed specified by the object tracking unit 245.

  Then, the distribution estimation unit 246 generates a grid map (hereinafter referred to as an occupancy probability map) representing the probability that an object is present in a predetermined range corresponding to the periphery of the transmission vehicle Ma. The occupancy probability map includes the same number of rows and columns as the grid map generated by the transmission vehicle Ma. That is, the configuration itself of the occupancy probability map is the same as the grid map generated by the transmission vehicle Ma. Therefore, the number of cells included in the occupancy probability map is also N = 60.

  In each cell constituting the occupancy probability map, a numerical value representing the probability that an object is present in the cell (that is, occupancy probability) is inserted. A cell whose occupancy probability is set to 0 means that it is estimated that there is no object in the cell. In the occupancy probability map, cells corresponding to the current position of each object estimated based on the received grid map are non-zero elements.

  The occupancy probability of a cell corresponds to the presence probability which is the probability that an object is present in the cell. The distribution estimation unit 246 estimates the current position for each of the tracked objects, and calculates the presence probability for each cell for each of the objects. The existence probability of an object in a cell is the probability that the object is present in the cell. The existence probability of each cell for a certain object is determined according to the current position of the object or the like. It is preferable to set the existence probability of the object for each cell to a higher value for cells located in the traveling direction of the object being detected (in other words, tracked).

  For example, since the object with the object number 001 shown in FIG. 7 moves in the right direction on the paper surface, the adjacent cells in the horizontal direction are set to higher values than the cells adjacent in the vertical direction of the estimated current position. . FIG. 8 shows the occupancy probability map corresponding to the estimation result shown in FIG. 7 in a matrix form for convenience. The broken line in FIG. 8 represents the current position of the object with object number 001 estimated by the distribution estimation unit 246.

  Further, based on the fact that the object with object number 001 is moving to the right side of the drawing, the occupancy probability map corresponding to the estimation result shown in FIG. 7 is a cell adjacent to the left side of the estimated current position as shown in FIG. Rather, the value may be set so that the cell adjacent on the right side has a higher value. In this embodiment, the presence probability of each cell of the object is calculated in consideration of the moving direction of the object.

  Furthermore, it is preferable that the existence probability of the object per cell be set to be distributed widely along the moving direction as the moving speed of the object is higher. For example, although an object with an object number 001 as an object is assumed to be moved to the right by one cell in FIG. 7, if it is moved to the right by two cells, it is estimated as shown in FIG. Set a non-zero value as the occupancy probability from the current position to the right 2 cells.

  The distribution estimation unit 246 generates the occupancy probability map indicating the occupancy probability for each cell by adding the presence probability for each cell for each object calculated according to the above rule. The sum of the existence probabilities of the respective objects may be set as the occupancy probabilities of the cells in the cells in which the existence probabilities of a plurality of objects overlap each other. That is, the distribution estimation unit 246 generates an occupancy probability map by calculating the presence probability for each cell according to the current position of each object. The occupancy probability map corresponds to data indicating the probability that each cell is an occupied cell in the grid map as the original data of the reception signal y.

  The filter generation unit 247 generates a filter vector F obtained by converting the occupancy probability map generated by the distribution estimation unit 246 into an N-dimensional column vector. Since the occupancy probability map itself is data representing the existence probability of an object for each point (specifically, for each cell) in the vicinity of the transmitting vehicle Ma, this filter vector F is also a point corresponding to the peripheral region of the transmitting vehicle Ma Indicates the existence probability of each object. The filter vector F has non-zero elements at positions according to the distribution of objects in the periphery of the transmitting vehicle Ma.

  The vector x as the original data corresponding to the reception signal y is data indicating the distribution of an object present around the transmission vehicle Ma, and the filter vector F is for each point corresponding to the peripheral region of the transmission vehicle Ma Data indicating the existence probability of the object of And the point which each element which filter vector F and vector x show shows in real space corresponds mutually.

  That is, according to another aspect, the filter vector F corresponds to a vector indicating the result of estimating the position of the nonzero element in the vector x as the original signal of the reception signal y. The numerical value of each element of the filter vector F corresponds to the probability that the element is set to a non-zero value in the vector x as the original signal. The filter vector F generated by the filter generation unit 247 is used by the decoding processing unit 242.

  The grid map storage unit 25 stores a grid map as peripheral object data received from the transmission vehicle Ma as another vehicle. The grid map storage unit 25 is realized using a rewritable storage medium such as a RAM or a flash memory. The grid map is stored in association with information indicating the transmission source of the compressed data (that is, transmission source information) and reception time information. The grid map is stored for at least a predetermined time from the reception time, and thereafter deleted at any time. The storage period is, for example, 30 seconds or 1 minute.

<Decryption processing>
Next, the restoration process performed by the decoding device 24 will be described using the flowchart shown in FIG. The restoration process is a process for restoring a grid map from the compressed data acquired by the compressed data receiving unit 241. The restoration process includes steps S101 to S107 as shown in FIG. The restoration process may be started, for example, triggered by the compressed data receiving unit 241 receiving the compressed data from the short range communication module 21.

  First, in step S101, two grid maps are received from the same transmission vehicle Ma as the transmission source of the compressed data received this time, and the reception time is within a predetermined extraction condition time from the present. As described above, it is determined whether or not it is stored in the grid map storage unit 25. The extraction condition time may be appropriately designed according to the generation interval of the grid map in the transmission vehicle Ma, and here, is set to a value four times the generation interval, that is, 2 seconds. In this step S101, it is possible to generate a filter vector F whether or not the current position of an object present around the transmission vehicle Ma can be estimated based on the grid map that the distribution estimation unit 246 has received. It corresponds to the process of determining whether or not

  A grid map received from the same transmission vehicle Ma as the transmission source of the compressed data (that is, received signal y) received this time in step S101 and the reception time is within the extraction condition time from the present time If two or more are present, step S103 is executed. On the other hand, there are 0 or 1 grid maps that are received from the same transmission vehicle Ma as the transmission source of the compressed data received this time, and the reception time is within the extraction condition time from the present. In the case, step S102 is performed.

  In the present embodiment, in order to generate the occupancy probability map (as a result, the filter vector F) using the movement direction or movement speed of the object, it is assumed that two or more grid maps received within the extraction condition time from the current time exist. The conditions for generating the occupancy probability map are not limited thereto. When generating the occupancy probability map without considering the moving speed or moving direction, at least one grid map from the same transmitting vehicle Ma as the transmission source of the compressed data received this time is within the extraction condition time from the current time It should just be able to receive. That is, in step S101, at least one grid map is a grid map received from the same transmission vehicle Ma as the transmission source of the compressed data received this time, and the reception time is within the extraction condition time from the present. It may be configured to determine whether it is stored in the map storage unit 25 or not.

  In step S102, the decoding processing unit 242 decodes the compressed data as the reception signal y using a predetermined decoding algorithm known in the technical field of compressed sensing. Known decoding algorithms used in the technical field of compressed sensing include Matching Pursuit (MP), Iterative Hard Thresholding (IHT), Orthogonal MP (so-called OMP), Weak MP, LS-OMP, Basis Pursuit (BP), These include Iterated-Reweighted-Least-Squares (IRLS), Least Angle Regression (LARS), and Approximate Message Passing (AMP). Note that MP, IHT, OMP, Weak MP, and LS-OMP are decoding algorithms corresponding to the greedy method. BP, IRLS, and LARS are a type of convex relaxation method that approximates a cost function as a convex function.

Here, as an example, the decoding processing unit 242 decodes the received signal y using OMP. In OMP, among the N column vectors a _i (i = 1, 2,..., N) included in the compression matrix A, the k have a strong correlation with the residual r determined based on the received signal y. This is a method of recovering a vector x (that is, original data) as original data using a column vector (k is a natural number).

FIG. 12 summarizes the OMP algorithm, and the specific procedure is as shown in FIG. As shown in FIG. 12, the OMP includes processes 1 to 4. Process 1 is a process of updating a support set S which is a set of element numbers for which non-zero values are set in the original data. In processing 1, a column vector a _i having the highest correlation with the residual r in the compression matrix A (the absolute value of the inner product is large) is specified, and a subscript i of the column vector a _i is added to the support set S. Although the apparent variable to be added to the support set S in the process 1 is j, the variable j is set to the subscript i satisfying the above condition. Therefore, the process 1 is a process of adding the subscript i of the column vector a _i which substantially satisfies the above condition to the support set S. Note that i and j are variables set to integers from 1 to N.

  Process 2 is a process of calculating the best solution in the support set S. Process 3 is a process of calculating the residual provided by the estimated solution calculated in process 2, and process 4 is a process of determining whether the result of process 3 satisfies the termination condition. The termination condition may be designed as appropriate. Here, it is assumed that when the residual r becomes smaller than the approximation threshold ε, or when the number of repetitions p exceeds the upper limit q repeatedly, the process of repetition is ended. Specific values of the approximation threshold ε and the repetition upper limit q may be designed as appropriate.

  The residual r in OMP is the result of subtracting a signal (specifically, an M-dimensional column vector) obtained by compression-transforming the provisional estimated solution with the compression matrix A from the reception signal y. The residual r is a parameter that functions as an index indicating the degree of deviation between the estimated solution and the original data. The residual r determined in the compression matrix A based on the received signal y and the strongly correlated column vector are column vectors having large absolute values of the inner product with the residual r. That is, OMP is a method of adopting as a support in order from the column having the largest absolute value of the inner product with the residual r. The number of non-zero elements k may be added to the compressed data. In addition, the non-zero element number k may be a preset value. When the calculation process in step S102 is completed, step S106 is executed.

  In this embodiment, as an example, if it is not possible to estimate the current position of an object present in the vicinity of the transmission vehicle Ma based on the received grid map, the compressed data is decoded using OMP. Not limited to. An algorithm other than OMP may be adopted as a known decoding algorithm (hereinafter referred to as an alternative algorithm) used when it is not possible to estimate the current position of an object present around the transmission vehicle Ma based on the received grid map. For example, the alternative algorithm may be an IHT. Also, the alternative algorithm may be BP or the like. Various known decoding algorithms can be employed.

  IHT roughly calculates a residual r for an estimated solution which is temporary decoded data, and repeats the process of correcting the estimated solution with the residual r to obtain a final solution (that is, decoding Data) algorithm. In IHT, in the correction phase of the estimated solution, the non-zero element number k of the original data is known, and in order to keep the estimated solution k-sparse, the absolute values of the elements of the estimated solution corrected are large. Set to 0 except for k. The residual r is a vector obtained by subtracting a signal obtained by linearly converting an estimated solution using a compression matrix from a reception signal y as compressed data.

  In step S103, the distribution estimation unit 246 uses an occupancy probability map indicating the existence probability of an object for each point corresponding to the periphery of the transmission vehicle Ma based on a plurality of grid maps from the transmission vehicle Ma received within a predetermined time in the past. Generate When the process in step S103 is completed, step S104 is executed.

  In step S104, the filter generation unit 247 generates, based on the occupancy probability map generated by the distribution estimation unit 246, a filter vector F indicating the existence probability of an object for each point corresponding to the periphery of the transmission vehicle Ma. When the process in step S104 is completed, step S105 is executed.

In step S105, the decoding processing unit 242 decodes compressed data as the reception signal y by using an algorithm shown in FIG. 13 as a proposed method. The algorithm as the proposed method in the present embodiment is a new algorithm based on OMP, and is referred to as filter combined OMP for convenience. FIG. 13 summarizes the filter combined OMP, and a specific procedure is as shown in FIG. As shown in FIG. 13, the filter combination OMP includes processing 1 to processing 4. Process 1 is a process of updating the support set S, and in the compression matrix A, a column vector a _i whose value obtained by multiplying the inner product with the residual r by the value fi of the i-th element of the filter vector F is maximum. The subscript i of is added to the support set S. The processes 2 to 4 are the same as the processes 2 to 4 of the OMP shown in FIG.

That is, the difference between the filter combined OMP and the conventional OMP is that weighting by the filter vector F is introduced in the selection (update) of the support set in processing 1 and is as follows. In OMP, a subscript i of a column vector a _i having the highest correlation with the residual r in the compression matrix A (the absolute value of the inner product is large) is added to the support set S. On the other hand, in the proposed method, in the compression matrix A, the subscript i of the column vector a _i for which the value obtained by multiplying the inner product with the residual r by the value fi of the i-th element of the filter vector F becomes the support set S to add.

  According to such a configuration, it is possible to reduce the possibility that the subscript i of the column vector having a weak correlation with the vector x as the original signal in the compression matrix A is selected as the support. That is, there is a high possibility that the column vector component having a weak correlation with the vector x as the original data can be excluded from the process of optimization of the estimated solution. As a result, estimation errors can be suppressed.

  In step S106, the grid map restoration unit 243 converts the N-dimensional vector z as decoded data obtained in the above processing into a grid map, and in step S107, the storage processing unit 244 generates the grid map generated in step S106. This flow is stored in the grid map storage unit 25 in association with the transmission source information and the reception time information, and this flow is ended.

  Note that the grid map as peripheral object data handled in the present embodiment has high spatial correlation on the time axis. The present disclosure focuses on the spatio-temporal correlation of data to be transmitted and received, and the decoding accuracy is obtained by using a filter vector F determined using the spatio-temporal correlation of data to be transmitted and received. Increase. Furthermore, in the present embodiment, it is possible to perform more accurate decoding by weaving the movement feature amount (for example, movement speed and movement direction) of the object into the filter vector F.

  By the way, although the filter combination OMP is used as an example in the present embodiment as an algorithm for correcting the estimated solution using the filter vector F and the compression matrix, the present invention is not limited to this. The filter vector F is applicable to various decoding algorithms, and contributes to the improvement of the decoding accuracy and the convergence speed to the optimum solution. For example, the method of updating the support set S using weighting by the filter vector F is also applicable to MP which is an algorithm underlying the OMP. It is also applicable to other algorithms based on MP (eg, Weak MP, LS-OMP, etc.).

  Furthermore, updating of the estimated solution using the filter vector F is also applicable to IHT. FIG. 14 shows an algorithm (hereinafter, filter combined IHT) in which weighting processing by a filter vector F is applied to update of an estimated solution in IHT as a second proposed method. The filter combination IHT includes processes 1 to 5 as shown in FIG. Process 1 is a process of calculating a basic update vector v. The basic update vector is a vector obtained by linearly mapping the residual r with the transposed matrix of the compression matrix A. Process 2 weights the basic update vector v generated in process 1 with the filter vector F to generate a weighted update vector c. In process 3, the estimated solution is updated using the weighted update vector generated in process 2. Process 4 is a process of calculating the residual r. Process 5 is a process of determining whether the termination condition is satisfied.

  The differences between the filter combined IHT as the proposed method and the normal IHT are as follows. In the ordinary IHT, the process of setting elements other than k having a large absolute value to 0 among elements of the vector obtained by adding the basic update vector v defined in FIG. 14 to the estimated solution is repeated. Then, update the estimated solution. On the other hand, in the second proposed method, as defined as processing 2, a vector obtained by multiplying each element included in the basic update vector v and each element included in the filter vector F is added to the estimated solution, and Among the elements included in the vector after addition, the process of setting elements other than k having a large absolute value to 0 is repeated to update the estimated solution. According to such an aspect, the same effect as that of the filter combined OMP described above can be obtained. An algorithm that repeatedly performs processing of updating an estimated solution using a filter vector F and a compression matrix A, such as a filter combined OMP and a filter combined IHT, corresponds to a filter combined algorithm.

  The embodiment of the present disclosure has been described above, but the present disclosure is not limited to the above-described embodiment, and various modifications described below are also included in the technical scope of the present disclosure. Also, various modifications can be made without departing from the scope of the invention. For example, various modifications described below can be implemented in combination as appropriate as long as technical contradiction does not occur.

  In addition, about the member which has the function same as the member described in the above-mentioned embodiment, the same code | symbol is attached | subjected and the description is abbreviate | omitted. In addition, when only a part of the configuration is mentioned, the configuration of the embodiment described above can be applied to the other parts.

[Modification 1]
In the above-described proposed method, when the number of iterations p exceeds a predetermined defiltering threshold value, decoding using ordinary OMP or ordinary IHT is performed using an estimated solution at that time (in other words, as an initial solution) You may control to continue. That is, in step S105, while the estimated solution is calculated by the proposed method (for example, filter combined OMP or filter combined IHT) until the number of repetitions p exceeds the filter cancellation threshold, after the number of repetitions p exceeds the filter cancellation threshold In, the estimation solution at that time point is taken over, and the calculation of the estimation solution in ordinary OMP etc. is continued. Such a configuration corresponds to a configuration in which the proposed method and the conventional method are combined and decoded.

  The operation and effect of the modification 1 are as follows. While using the filter vector F, the proposed method promotes convergence to the optimal solution, but can also work in the direction of excluding the presence of a novel object that has not been detected so far. According to the configuration of the present modification to such an issue, it is possible to flexibly cope with the appearance of a new object which has not been detected until now. That is, the effect described in the above-described embodiment and the flexibility of handling the newly detected object can be compatible.

  Note that a new object that has not been detected is an object that has not been represented by a saved grid map, for example, an object that has entered the real space represented by the grid map, or is being tracked (in other words, detected) Object separated from the object). An object separated from an object being tracked is a falling object from a vehicle as an object being tracked, a person who got out of the vehicle, or the like.

  The filter removal threshold value applied in the above configuration may be appropriately designed in a range smaller than the repetition upper limit value q. For example, the filter release threshold may be set to a value such as one half, one third, or one fourth of the repetition upper limit value q. The defiltering threshold may be dynamically set according to the number k of non-zero elements of the received grid map to suppress excessive filter effects. For example, as the non-zero element number k is smaller, the filter removal threshold is also set to a smaller value.

[Other modifications]
The device for distributing the peripheral object data may be a device installed on the road / along the road (so-called roadside device). In that case, peripheral mobile object data (strictly speaking, its compressed data) generated by the roadside device may be provided to the receiving system 2 by road-to-vehicle communication.
-Although the aspect which implements several decoding algorithm in order in the above decoding process was disclosed, it is not restricted to this. Multiple decoding algorithms may be implemented in parallel. Also, multiple decoding algorithms may be implemented in parallel, and errors may be detected by majority decision.
The receiver system 2 and the transmitter system 1 may be mounted on one vehicle. According to another aspect, such a configuration corresponds to a configuration in which the receiving system 2 including the decoding device 24 also includes the function of the transmitting system 1.
-Although the aspect which compression-codes and transmits the data which show the position of the mobile which exists around the transmission side system 1 above was disclosed, the data to be compression-coded are not limited to this. Apply the above configuration to data at a certain point in time for a predetermined amount of state that changes dynamically while having temporal continuity, such as temperature distribution and grid map showing distribution of rainfall amount be able to.
-The receiving side system 2 does not need to be mounted in a vehicle, and may be built in the roadside machine. In addition, the receiving side system 2 may be a portable terminal such as a smartphone or a wearable device.
· Transmission and reception of compressed data may be performed, for example, in accordance with a known short distance wireless communication standard such as Wi-Fi (registered trademark) (IEEE 802.11a / b / g / n / p) or Bluetooth (registered trademark) good. Also, transmission and reception of compressed data may be performed via a wide area communication network.

Reference Signs List 1 transmission side system 2 reception side system 11 narrow area communication module 12 perimeter monitoring sensor 13 original data generation unit 14 compression processing unit 21 narrow area communication module 22 map data storage unit 23 locator 24 decoding device , 25 grid map storage unit, 241 compressed data reception unit, 242 decoding processing unit, 243 grid map restoration unit (restoration unit), 244 storage processing unit, 245 object tracking unit, 246 distribution estimation unit (estimation unit), 247 filter generation Department

Claims (6)

A decoding apparatus that decodes compressed data, which is data that has been compressed using compressed sensing, comprising:
The compressed data is one column vector having state quantity data, which is data indicating a state at a certain point in time with respect to a dynamically changing predetermined state quantity while having temporal continuity, and a predetermined number of elements It is data obtained by compression encoding of the pre-compression original data obtained by
A compressed data receiving unit (241) for sequentially receiving the compressed data from another device;
Each time the compressed data receiving unit receives the compressed data, the decoding processing unit (242) generates decoded data including the number of elements by executing a predetermined decoding algorithm on the received compressed data. When,
A restoration unit (243) for restoring the state quantity data based on the decoded data generated by the decoding processing unit;
An estimation unit (246) for estimating the current state of the state quantity based on the plurality of state quantity data generated by the restoration unit within a predetermined past time;
A filter generation unit (247) for generating a filter vector including one non-zero element at a position corresponding to the current state of the state quantity estimated by the estimation unit, which is one column vector including the number of elements; Equipped with
When the filter generation unit can generate the filter vector, the decoding processing unit corresponds to the pre-compression original data using the filter vector, the compressed data, and the compression matrix as the decoding algorithm. Using the filter vector and the compression matrix until a predetermined end condition is satisfied, while generating an estimated solution and calculating a residual that is an index indicating the degree of deviation between the estimated solution and the original data before compression. A decoding device configured to generate the decoded data using an algorithm that repeatedly performs the process of correcting the estimated solution. The decoding apparatus according to claim 1, wherein
The state quantity data is peripheral object data representing, in a grid map form, the presence or absence of an object at each point within a predetermined range determined according to the current position of the other device, or the presence probability of the object.
The decoding processing unit converts the decoded data into the peripheral object data,
The estimation unit estimates the distribution of the object within the predetermined range, based on the plurality of pieces of peripheral object data generated by the restoration unit within a predetermined time in the past.
The decoding device, wherein the filter generation unit is configured to generate, as the filter vector, a vector including non-zero elements at positions according to the distribution of the object estimated by the estimation unit. The decoding apparatus according to claim 2,
The filter generation unit identifies the movement direction for each object indicated in the peripheral object data based on the plurality of peripheral object data generated by the restoration unit within a predetermined time in the past, and the identified movement direction A decoding device configured to estimate a current position for each of the objects within the predetermined range based on. The decoding apparatus according to claim 3, wherein
The filter generation unit specifies the movement speed for each object indicated in the peripheral object data based on the plurality of peripheral object data generated by the restoration unit within a predetermined time in the past, and the movement for each object A decoding device configured to estimate a current position for each object within the predetermined range based on a direction and a movement speed. The decoding device according to any one of claims 1 to 4, wherein
The decryption processing unit
When the filter generation unit can not generate the filter vector, one of Orthogonal Matching Pursuit (OMP), Matching Pursuit (MP), and Iterative Hard Thresholding (IHT) used in the compressed sensing is used. While generating the decoded data,
When the filter generation unit can generate the filter vector, in the compression matrix, an integer j that satisfies the following expression and an element number for which a non-zero value is set in the raw data before compression By adding to the support set, which is a set, the estimated solution is generated, and a residual, which is an index indicating the degree of deviation between the estimated solution and the original data before compression, is calculated, and a predetermined end condition is satisfied. A decoding device configured to generate the decoded data using a filter combination algorithm that repeatedly performs a process of correcting the estimated solution using the filter vector and the compression matrix.

The decoding apparatus according to claim 5, wherein
The decoding processing unit is configured to generate the filter vector when the filter generation unit can generate the filter vector, and when the number of repetitions of the process exceeds a predetermined filter cancellation threshold in the filter combination algorithm. A decoding device configured to generate the decoded data using any one of the OMP, the MP, and the IHT as an initial solution for the estimated solution obtained at a point in time.

Images