Information processing facility, information processing system, information processing procedure and information processing program

Abstract

Information processing facility (100), comprising: an object position information procurement unit (110) for procuring obstacle position information showing the position of at least one obstacle that is present around a vehicle (CA), and vehicle position information showing the position of the vehicle (CA); an obstacle potential map generation unit (140) for generating an obstacle potential map of a traffic accident risk for the at least one obstacle based on the obstacle position information; and an obstacle movement route prediction unit (150) for predicting a movement route of the at least one obstacle based on the obstacle potential map and for generating obstacle movement route prediction information; and a vehicle potential map generation unit (130) to generate a dynamic potential map of a traffic accident risk for the vehicle (CA) based on the vehicle position information and the obstacle movement route prediction information in a case where the at least one obstacle moves along the predicted movement route shown by the obstacle movement route prediction information; wherein the object position information acquisition unit (110) as the obstacle position information position information, which includes stationary obstacle position information showing a position of at least one stationary obstacle from within the at least one obstacle, and moving obstacle position information showing a position of at least one moving obstacle from within the at least one obstacle, the obstacle potential map generation unit (140) as the obstacle potential map generates a moving obstacle potential map of a traffic accident risk for the at least one moving obstacle based on the moving obstacle position information, the obstacle movement route prediction unit (150) as the obstacle movement route prediction information generates movement route prediction information of a moving obstacle, which predicts a movement route of the at least one moving obstacle based on the moving obstacle potential map, and The vehicle potential map generation unit (130) comprises: a static potential map generation unit to generate a static potential map of a traffic accident risk for the vehicle (CA) based on at least the stationary obstacle position information in addition to the vehicle position information, wherein the risk is caused by at least one stationary obstacle, and a dynamic potential map generation unit (132) to generate the dynamic potential map based on the static potential map and the motion route prediction information of the moving obstacle.

Description

Technical field

[0001] The present disclosure relates to an information processing device that predicts movement routes of obstacles around a vehicle, an information processing system that uses this, an information processing method for it, and an information processing program for it. Technical background

[0002] One technique used by vehicles, particularly autonomous vehicles, to avoid obstacles along a route is the so-called potential method (e.g., patent document 1). In this method, two-dimensional data is first generated in a plan called a potential map, which depicts the vehicle from above. Areas that the vehicle should avoid are defined by the magnitude of the potential values. The movements of the obstacles are predicted, for example, using a Kalman filter, and a potential map is generated for the vehicle based on these predicted movement paths. The higher the potential value in the potential map, the greater the need for safe obstacle avoidance.

[0003] In the potential method, the vehicle's driving control is then carried out using the potential map in such a way that the vehicle passes the route with the minimum total sum of potential values ​​among the routes that the vehicle can travel.

[0004] DE 10 2013 013 747 A1 discloses a driver assistance system for a vehicle, with a detector designed to detect at least one object in the vicinity of the vehicle.

[0005] DE 10 2004 009 085 A1 discloses a traffic flow simulation system and a method for simulating flows of vehicles. Literature on the state of the art; patent documents

[0006] Patent document 1 Japanese patent application no. 2019-523880 (also as WO 2020 / 044 512 A1) Summary of the invention Problems to be solved by the invention

[0007] In the technique described in Patent Document 1, the movement paths of obstacles, such as oncoming vehicles, located around a vehicle are predicted under the assumption that they are in linear motion with constant acceleration. However, actual obstacles are not always in linear motion with constant acceleration. Therefore, the technique described in Patent Document 1 has a problem: the prediction may be inaccurate in some cases.

[0008] The present disclosure serves to solve the above problem and aims to provide an information processing device with improved prediction accuracy for the movement routes of obstacles. Means to solve the problem

[0009] The information processing device according to the present disclosure comprises: an object position information acquisition unit for acquiring obstacle position information showing the position of at least one obstacle present around a vehicle, and vehicle position information showing the position of the vehicle; an obstacle potential map generation unit for generating an obstacle potential map of the risk of a traffic accident for the at least one obstacle based on the obstacle position information; and an obstacle movement route prediction unit for predicting a movement route of the at least one obstacle based on the obstacle potential map and for generating obstacle movement route prediction information; and a vehicle potential map generation unit.to generate, based on vehicle position information and obstacle movement route prediction information, a dynamic potential map of the risk of a traffic accident for the vehicle in a case where the at least one obstacle moves along the predicted movement route shown by the obstacle movement route prediction information, wherein the object position information acquisition unit acquires position information as the obstacle position information, which includes stationary obstacle position information showing a position of at least one stationary obstacle from within the at least one obstacle, and moving obstacle position information showing a position of at least one moving obstacle from within the at least one obstacle,The obstacle potential map generation unit comprises a moving obstacle potential map, which generates a moving obstacle potential map of a traffic accident risk for the at least one moving obstacle based on the moving obstacle position information; the obstacle movement route prediction unit comprises movement route prediction information for a moving obstacle, which predicts a movement route of the at least one moving obstacle based on the moving obstacle potential map; and the vehicle potential map generation unit comprises a static potential map generation unit to generate a static potential map of a traffic accident risk for the vehicle based on at least the stationary obstacle position information in addition to the vehicle position information.wherein the risk is caused at least by at least one stationary obstacle, and a dynamic potential map generation unit to generate the dynamic potential map based on the static potential map and the motion route prediction information of the moving obstacle. Advantageous effects of the invention

[0010] The information processing device according to the present disclosure comprises the obstacle potential map generation unit for generating an obstacle potential map of a traffic accident risk for the at least one obstacle based on the obstacle position information, and the obstacle movement route prediction unit for predicting a movement route of the at least one obstacle based on the obstacle potential map and for generating obstacle movement route prediction information. Thus, the prediction accuracy of the movement route of the at least one obstacle can be improved by taking the environment of the at least one obstacle into account. Brief description of the drawings Fig. Figure 1 is a diagram showing a configuration example of an information processing system 1000 according to embodiment 1. Fig. Figure 2 is a top view showing a vehicle CA and obstacles around the vehicle CA from above. Fig. Figure 3 is a conceptual diagram showing a concrete example of a static potential map. Fig. Figure 4 is a top view showing the vehicle CA and an obstacle around the vehicle CA from above. Fig. Figure 5 is a conceptual diagram showing a concrete example of an obstacle potential map. Fig. Figure 6 is a conceptual diagram showing a concrete example of obstacle movement routes given by obstacle movement route prediction information. Fig. Figure 7 is a conceptual diagram showing a concrete example of a dynamic potential map. Fig. Figure 8 is a diagram showing an example of a hardware configuration of an information processing device 100 according to embodiment 1. Fig. Figure 9 is a flowchart showing the operation of the information processing device 100 according to embodiment 1. Fig. Figure 10 is a top view showing the vehicle CA and obstacles around the vehicle CA from above. Fig. Figure 11 is a conceptual diagram showing a concrete example of an obstacle potential map. Fig. Figure 12 is a conceptual diagram showing a concrete example of obstacle movement routes shown by the obstacle movement route prediction information. Fig. Figure 13 is a conceptual diagram showing a concrete example of the dynamic potential map. Description of embodiments Embodiment 1

[0011] Fig. Figure 1 is a diagram showing a configuration example of an information processing system 1000 according to embodiment 1. The information processing system 1000 comprises an information processing unit 100, a localizer 200, a camera 300, a radar 400, a LiDAR (light detection and ranging) 500, and a vehicle control unit 600. In embodiment 1, the information processing system 1000 is an in-vehicle system to be mounted on a vehicle CA (not shown).

[0012] The localizer 200 is a device that locates the position of the vehicle CA, which may be equipped with, for example, a GNSS (Global Navigation Satellite System) receiver, an inertial navigation system, an odometry system, or similar equipment. In embodiment 1, the localizer locates the position of the vehicle CA by decoding and analyzing the information received from the GNSS receiver, the inertial navigation system, or the odometry system and transmits vehicle position information showing the position of the vehicle CA to the information processing device 100.

[0013] Alternatively, the localizer can use 200 pieces of information from various sensors, which will be described later, to estimate the position of a vehicle in question.

[0014] The Camera 300, the Radar 400 and the LiDAR 500 are sensors that capture an environment around the vehicle CA.

[0015] The camera 300 takes pictures of the area around the vehicle CA. In embodiment 1, the camera 300 transmits the image data obtained by taking pictures to the information processing unit 100.

[0016] The radar 400 is a device that detects the environment around the vehicle CA using radio waves, e.g., millimeter waves. It measures distances, speeds, angles, etc., of obstacles. Here, distances are those between the obstacles present around the vehicle CA and the vehicle CA itself. In embodiment 1, the radar 400 transmits the radar data obtained through detection to the information processing unit 100.

[0017] The LiDAR 500 is a device that uses laser light to scan the environment around the vehicle CA and measures the distances between obstacles present around the vehicle CA and the vehicle CA, as well as the speeds, angles, etc., of the obstacles. In general, the radar 400 is very robust against environmental influences, while the LiDAR 500 can measure distance and orientation with high spatial resolution. In embodiment 1, the LiDAR 500 transmits the point cloud data obtained by scanning to the information processing unit 100. Hereinafter, the image data obtained from the camera 300, the radar data obtained from the radar 400, and the point cloud data obtained from the LiDAR 500 are collectively referred to as sensor data.

[0018] The vehicle control unit 600 is a device that performs various controls for the vehicle CA to enable automated driving. For example, it uses an electronic control unit (ECU) for vehicles, which controls the steering, accelerator pedal, brakes, and similar functions. In embodiment 1, the vehicle control unit 600 controls the vehicle based on a predicted movement route of the vehicle CA when the vehicle CA is operating automatically, the movement route being predicted by the information processing unit 100.

[0019] The information processing device 100 is a device that predicts movement routes of obstacles present around the vehicle CA, and in embodiment 1 the information processing device 100 is an in-vehicle device that is mounted on the vehicle CA.

[0020] The information processing unit 100 comprises an object position information acquisition unit 110, a map data acquisition unit 120, a vehicle potential map generation unit 130, an obstacle potential map generation unit 140, an obstacle movement route prediction unit 150 and a vehicle movement route prediction unit 160.

[0021] The object position information retrieval unit 110 is a unit that retrieves object position information showing the positions of objects. In embodiment 1, the object position information retrieval unit 110 retrieves vehicle position information showing the position of the vehicle and obstacle position information showing the positions of obstacles present around the vehicle CA. The obstacle position information retrieved by the object position information retrieval unit 110 includes stationary obstacle position information showing the positions of stationary obstacles beneath the obstacles and moving obstacle position information showing the positions of moving obstacles beneath the obstacles.Stationary obstacles include obstacles such as a guardrail and a building that are stationary, and moving obstacles include obstacles such as a vehicle and a pedestrian that are moving.

[0022] Furthermore, the object position information acquisition unit 110 recognizes obstacle attributes in the vehicle's vicinity based on sensor data and acquires obstacle attribute information that displays the obstacles' attributes. Hereinafter, the obstacle position information and the obstacle attribute information are collectively referred to as obstacle information. An obstacle attribute is information indicating its classification, such as whether the obstacle is stationary or moving, its type (e.g., pedestrian, vehicle, or bicycle), and its physical size (e.g., predicted mass and area as seen from the vehicle CA), and similar characteristics.

[0023] The object position information acquisition unit 110 obtains vehicle position information from the localizer 200 and sensor data from the camera 300, the radar 400, and the LiDAR 500 to detect obstacles around the vehicle CA and thereby obtain obstacle information. The preceding description stated that the object position information acquisition unit 110 obtains the vehicle position information by receiving it from the localizer 200. However, the object position information acquisition unit 110 can also obtain the vehicle position information by calculating the position of the vehicle CA itself based on the information from the various sensors.

[0024] This section describes how the object position information acquisition unit 110 obtains object position information based on sensor data acquired by the various sensors. However, the object position information acquisition unit 110 can also obtain object position information by communicating with an external server or another vehicle. Obstacle information, for example, can be obtained by communicating with another vehicle and exchanging information showing their respective positions, or from dynamic map data if the map data acquisition unit 120, described later, is responsible for obtaining the dynamic map data.

[0025] The map data acquisition unit 120 is a unit that acquires map data about the vehicle CA. In embodiment 1, the map data acquisition unit 120 acquires the map data about the vehicle CA based on the vehicle position information acquired by the object position information acquisition unit 110. In embodiment 1, it is described that the map data acquisition unit 120 acquires the map data from an external server. However, the information processing unit 100 can include a storage unit that stores map data and acquire the map data by reading it from the storage unit.

[0026] The vehicle potential map generation unit 130 is a unit that generates a vehicle potential map showing the risk of a traffic accident for the vehicle CA. In embodiment 1, the vehicle potential map generation unit 130 generates a dynamic potential map based on the vehicle position information and obstacle movement route prediction information described later. This dynamic potential map reflects the vehicle potential map as the obstacles move along their predicted movement routes, as shown by the obstacle movement route prediction information.

[0027] The potential map is a spatial representation of information that quantifies the risk of a traffic accident for each spatial position. A traffic accident includes contact between vehicle CA and an obstacle, vehicle CA leaving the roadway, and similar events. For example, the risk of a traffic accident increases as vehicle CA approaches an oncoming vehicle or the edge of the road. Therefore, in the potential map, where vehicle CA is centered, the higher values ​​are given for the positions of oncoming vehicles or the road edges. The potential map is generally generated as two-dimensional data in a horizontal plane, viewed from above.However, the potential map can be generated as one-dimensional data if obstacles can be avoided solely by acceleration or deceleration, or as three-dimensional data by adding height to the horizontal plane if height information is important, e.g., if there is a tunnel with height restrictions.

[0028] In embodiment 1, the vehicle potential map generation unit 130 comprises a static potential map generation unit 131 and a dynamic potential map generation unit 132.

[0029] The Static Potential Map Generation Unit 131 is a unit that, based on at least the stationary obstacle position information in addition to the vehicle position information, generates a static potential map that shows the risk of a traffic accident for vehicle CA due to stationary obstacles. Furthermore, the Static Potential Map Generation Unit 131 can generate a static potential map based on the moving obstacle position information to include the risk of a traffic accident for vehicle CA due to moving obstacles at the current time.By including the risk due to moving obstacles at the current time in the static potential, the static potential is generated as a vehicle potential map, which shows the risk of a traffic accident for the vehicle CA due to the stationary obstacles and the moving obstacles that are present at their position coordinates at the current time.

[0030] When generating a vehicle potential map, the Static Potential Map Generation Unit 131 first obtains the lane center of the vehicle CA based on the vehicle position information and the map data. Based on the lane center information obtained above and the obstacle information, the Static Potential Map Generation Unit 131 generates a potential map in which the vehicle CA is centered. If the map data contains lane center data, it is advantageous to read only the lane center data to obtain the lane center. However, if the map data does not contain lane center data, it is advantageous to create a curved line as the lane center midway between the edge line at the side of the road and the center line of the road.

[0031] The static potential map generation unit 131 is currently generating a two-dimensional potential map R(x) based on the vehicle position information and obstacle information obtained by the object position information acquisition unit 110. Generally, the potential map is obtained by calculating potential values ​​at the current time T for two-dimensional coordinates x over a predefined area. In this case, the predefined area is, for example, an area with a radius of 100 m around the vehicle. This area is considered important for controlling the vehicle CA by the designer of the information processing unit 100. It is determined by the designer through presetting.

[0032] The static potential map can be generated according to the following formula 1, using, for example, a normal distribution whose center is an obstacle position X.k This assumes that there are n+1 obstacles. R(x)=∑k=0n{αk∗exp(−(Xk−x)2 / σk2)}+ω(1−exp(−(x1−Yr)2 / σr2)

[0033] Here, x is a two-dimensional coordinate in the potential map, α k is a weighting factor that is based on an attribute K k of the object is determined, and a two-dimensional standard deviation σ k The normal distribution is a value proportional to the width and height of the object. ω is a weighting factor with respect to a potential value of the road center, x1 is a first component of the two-dimensional coordinates x in the potential map, Y r represents the lane center for vehicle CA, and σ r , a preset value, is a constant to provide the potential value at the position furthest from the center of the lane.

[0034] A concrete example of the static potential map is shown using Fig. 2 and Fig. 3 described. Fig. Figure 2 is a top view showing the vehicle CA and the obstacles around the vehicle CA from above and Fig. Figure 3 is a conceptual diagram that provides a concrete example of the static potential map for the situation of Fig. Figure 2 shows. For better understanding of the spatial relationship, this is shown in Fig. 2 vehicles shown CA in Fig. 3 displayed.

[0035] The object position information acquisition unit 110 and the static potential map generation unit 131 detect the obstacles around the vehicle CA, the wall surfaces and the lane center, as shown in Fig. 2 is shown by using the information from the various sensors. Then the static potential map generation unit 131 generates a static potential map according to formula 1 based on the various detected pieces of information.

[0036] In Fig. 2. Around vehicle CA are vehicle MO21, vehicle MO22, and vehicle MO23, acting as moving obstacles. In Fig. 2 The leftmost straight line is a road edge line LE21, the rightmost straight line is a road edge line LE22, the middle straight line is the center line CL21 and the dashed line between the road edge line LE21 and the center line CL21 is the lane center LA21.

[0037] In Fig. 3. The risk potential RMO21 is a risk potential for the vehicle MO21 in Fig. 2. The risk potential RMO22 is a risk potential for the vehicle MO22 in Fig. 2 and the risk potential RMO23 is a risk potential for the vehicle MO23 in Fig. 2. Furthermore, there is a risk potential RLA21 in the areas on both the left and right sides of the vehicle due to the deviation from the lane center LA21. Fig. 2.

[0038] The Dynamic Potential Map Generation Unit 132 is a unit that generates the dynamic potential map based on the static potential map and obstacle movement route prediction information. The obstacle movement route prediction information is generated by the Obstacle Movement Route Prediction Unit 150. Therefore, the configurations of the Obstacle Potential Map Generation Unit 140 and the Obstacle Movement Route Prediction Unit 150 are described first, followed by a description of the Dynamic Potential Map Generation Unit 132.

[0039] The Obstacle Potential Map Generation Unit 140 is a unit that generates an obstacle potential map showing the risk of a traffic accident for an obstacle based on obstacle position information. If there are multiple obstacles, an obstacle potential map is generated for each one. Based on moving obstacle position information, the Obstacle Potential Map Generation Unit 140 can only generate moving obstacle potential maps, each showing the risk of a traffic accident for the moving obstacle. This is because the obstacle potential map is used to predict the movement path of the target obstacle, and by definition, a stationary obstacle is assumed not to be moving at all.When reference is made below to an obstacle potential map, this refers to a moving obstacle potential map.

[0040] Just as the Static Potential Map Generation Unit 131 receives the lane center of vehicle CA, the Obstacle Potential Map Generation Unit 140 receives the lane centers of other vehicles representing the moving obstacles.

[0041] The obstacle potential map generation unit 140 generates an obstacle potential map U i (x), which is seen from each of the obstacles. For example, the obstacle potential map U i (x), in which the obstacle is at a position X i centered, as expressed in Formula 2 below. Ui(x)=∑k=0n{αk∗βki∗exp(−(Xk−x)2 / σk2)}+{γk∗exp(−(C−x)2 / σc2)}+ω{1 −exp(−(x1−Yi)2 / σr2)

[0042] The first term represents potential values ​​of other obstacles, which differ from the one at X.i The second term represents the potential value of a vehicle in question, positioned from the obstacle seen from X. i The third term represents a potential value of the current lane, which is seen from the positioned obstacle, and the third term represents a potential value of the current lane, which is determined by the one at X. i The obstacle is seen from the positioned position.

[0043] β ki is a coefficient that is 0 when k = i, and 1 when this is not the case, γ k is a weighting value for a potential value of the vehicle in question, C is a position of the vehicle in question on the map, and σ is a two-dimensional standard deviation. c Y is a value that is proportional to the width and height of the vehicle in question, based on a normal distribution. i represents a lane center for the obstacle.

[0044] The Obstacle Movement Route Prediction Unit 150 is a unit that predicts the movement routes of obstacles based on the obstacle potential map and generates obstacle movement route prediction information. This information includes the predicted movement routes of moving obstacles. In this case, the Obstacle Movement Route Prediction Unit 150 can only predict movement routes for moving obstacles. This is because stationary obstacles are defined as obstacles that do not move and remain stationary.

[0045] In embodiment 1, the obstacle movement route prediction unit 150 predicts the movement route of each of the target obstacles based on the obstacle potential map U generated by the obstacle potential map generation unit 140. i(x) before. The movement route of the target obstacle is determined by the cost function shown below, where its steering value and its acceleration value at time t are e.g. with r. i (t) and a i (t) will be designated. Gi(Sit(ri(T),ai(T)),,,Sint(ri(T+mu),ai(T+mu))) =∑k=0m{Ui(Sik(ri(T+ku),ai(T+ku)),T+ku)+wr∗ri(T+ku)2+wa∗ai(T+ku)2}

[0046] This includes S it (r(t), a(t)) the two-dimensional coordinates at time t in a case where a steering value r(t) and an acceleration value a(t) are given, and w r and w a These are weighting values ​​for changes to a steering value r and an acceleration value a, which are preset. The obstacle movement route prediction unit 150 provides a movement route S. it (ri(t), a i (t)), ···, S int (r i (nt), a i(nt)) which minimizes the cost function. The movement route S is... it (r(t), a(t)) is a function of the steering value r(t) and the acceleration value a(t). The steering value r(t) and the acceleration value a(t) are its independent variables. The motion route S it (r(t), a(t)) is determined by obtaining the steering value r(t) and the acceleration value a(t) that minimize the cost function. T is a present time, u is a preset time interval, and m is a positive integer. The designer of the information processing device 100 can set u and m according to a rule of thumb or based on data obtained from experiments. Specifically, u can be determined by how frequently they want to know the change in position of an obstacle within a given time interval, and m can be determined by how far into the future they want to know the position of the obstacle.

[0047] Although the above description uses the movement route that minimizes the cost function, a multitude of movement routes can be determined as predicted routes. Alternatively, it is also possible to prepare a multitude of cost functions different from the one above for predicting the movement routes and to predict the movement routes based on their respective cost functions. That is, the obstacle movement route prediction unit 150 can predict a multitude of movement routes for an obstacle at any given time. In the following, the obstacle movement route prediction unit 150 determines as the predicted routes the movement route where the cost function represented by formula 3 reaches a minimum and the movement route where the cost function represented by formula 3 reaches a second minimum.

[0048] A concrete example of a process for predicting the movement paths of an obstacle is given using the Fig. 4, Fig. 5 to Fig. 6 described. Fig. Figure 4 is a top view of the vehicle CA and an obstacle around the vehicle CA from above. Fig. Figure 5 is a conceptual diagram showing a concrete example of the obstacle potential map and Fig. Figure 6 is a conceptual diagram showing a concrete example of the movement routes of an obstacle as shown by the obstacle movement route prediction information.

[0049] In Fig. Vehicle 4 is surrounded by vehicle CA by vehicle MO41, which is a moving obstacle. Vehicle MO41 is an oncoming vehicle for vehicle CA. Fig. Figure 5 is a conceptual diagram showing an obstacle potential map, which illustrates the risk of a traffic accident for vehicle MO41, and the vehicle MO41 in Fig. Number 4 is displayed to make the positional relationship easier to understand. The RCA risk potential is a risk potential due to the in Fig. 4. Vehicle CA shown. The grey areas on both the left and right sides of vehicle MO41 indicate the risk potential posed by vehicle MO41, which deviates from the center of the lane.

[0050] Fig. Figure 6 shows the predicted movement routes of vehicle MO41. The movement route AR61, indicated by the dashed arrow, is a movement route in the case where vehicle MO41 travels straight ahead, and the movement route AR62, indicated by the dash-dot-dash arrow, is a movement route in the case where vehicle MO41 turns right. If, for example, as in the conventional technique, it is assumed that vehicle MO41 is in a constant acceleration motion, only movement route AR61 can be predicted as the movement route of vehicle MO41. However, when using the obstacle potential map as in the information processing device 100 according to this embodiment, in addition to movement route AR61 when traveling straight ahead, movement route AR62 when turning right can also be predicted as a route with a low risk potential value.

[0051] Next, the Dynamic Potential Map Generation Unit 132 will be described.

[0052] As described above, the Dynamic Potential Map Generation Unit 132 is a unit that generates the dynamic potential map based on the static potential map and the obstacle movement route prediction information. In embodiment 1, the Dynamic Potential Map Generation Unit 132 generates the vehicle potential map by overlaying, on the static potential map, the risk of a traffic accident for the vehicle CA caused by the obstacles when positioned on their movement routes shown by the obstacle movement route prediction information.

[0053] In embodiment 1, the dynamic potential map generation unit 132 receives the positions of the obstacles at predetermined multiple time points from the movement routes of the obstacles predicted by the obstacle movement route prediction unit 150. Subsequently, the dynamic potential map, which reflects the movement prediction results of the obstacles, is generated by overlaying the risk potentials of the obstacles at each time point onto the static potential map. A static potential map can be generated, for example, according to formula 4 below. R'(x)=R(x)+∑i=0n∑k=1m{αi∗δk∗exp(−Sik(ri(T+ku),ai(T+ku))−x)2 / σi2}

[0054] Here, as described above, is α i a weighting factor derived from the object attribute K i is determined, and a two-dimensional standard deviation σ iIn a normal distribution, δ is a value that is proportional to the width and height of an object. k is a weighting factor for the risk posed by an obstacle in relation to the potential value when the risk is displayed on the static potential map at any given time. For example, the weighting δ is set to T at time t = T, which is closest to the present time, is large and has a weighting δ mT at time t = mT, which is furthest away from the present time, is small.

[0055] A concrete example of the dynamic potential map will be given with reference to Fig. 7 described.

[0056] Fig. Figure 7 is a conceptual diagram showing a concrete example of the dynamic potential map that is described in Fig. 4, Fig. 5 to Fig. This corresponds to the situation shown in section 6. For a better understanding of the spatial relationship, the vehicle CA, the vehicle MO41, and each road edge line shown in the diagram are shown. Fig. 4 are displayed. Fig. 7 is a risk potential RMO41, the risk potential for vehicle MO41 at the present time; a risk potential RAR61 (a risk potential RAR611, a risk potential RAR612 and a risk potential RAR613) is the risk potential for a predicted movement route AR61 in a case where vehicle MO41 travels straight ahead; a risk potential RAR62 (a risk potential RAR621, a risk potential RAR622 and a risk potential RAR623) is the risk potential for a predicted movement route AR62 in a case where vehicle MO41 turns right. Here, risk potential RAR611 and risk potential RAR621 are the risk potentials of vehicle MO41 at time T + u, risk potential RAR612 and risk potential RAR622 are the risk potentials of vehicle MO41 at time T + 2u, and risk potential RAR613 and risk potential RAR623 are the risk potentials of vehicle MO41 at time T + 3u.Furthermore, in the areas on both the left and right sides of vehicle CA, as in the other figures, there is a potential risk due to vehicle CA deviating from the center of the lane.

[0057] The vehicle motion route prediction unit 160 is a unit that predicts the motion routes of the vehicle CA based on the vehicle potential map and generates vehicle motion route prediction information. In embodiment 1, it predicts the motion routes of the vehicle CA based on the dynamic potential map generated by the vehicle potential map generation unit 130. The prediction of the motion routes of the vehicle CA can be performed in the same way as the prediction of the motion routes of obstacles, as shown in Formula 3. In this case, the motion route prediction unit 160 can transmit only the information indicating a motion route to the vehicle control unit 600 so that the vehicle control unit 600 can recalculate the steering angle and speed of the vehicle CA.Alternatively, the vehicle movement route prediction unit 160 can add the steering values ​​and speeds to the vehicle movement route prediction information in addition to the movement route in order to transmit the information to the vehicle control unit 600.

[0058] Furthermore, the vehicle movement route prediction unit 160 can output a signal to the vehicle control unit 600 to cause the vehicle CA to stop when the minimum value of the cost function calculated according to formula 3 is equal to or greater than a predetermined threshold.

[0059] Next, a hardware configuration of the information processing device 100 according to embodiment 1 is described. Each function of the information processing device 100 is implemented by a computer. Fig. Figure 8 is a diagram showing an example of the hardware configuration of the computer that implements the information processing unit 100.

[0060] The in Fig. 8 Hardware shown includes a processing unit 10000, such as a central processing unit (CPU), and a storage unit 10001, such as a read-only memory (ROM) and a hard disk.

[0061] The object position information acquisition unit 110, the map data acquisition unit 120, the vehicle potential map generation unit 130, the obstacle potential map generation unit 140, the obstacle movement route prediction unit 150, and the vehicle movement route prediction unit 160, shown in Fig. 1, are realized by the processing unit 10000, which executes a program stored in the storage unit 10001.

[0062] Furthermore, the method for implementing each function of the information processing device 100 is not limited to a combination of hardware and a program as described above. Instead, the processing device can be implemented by a single piece of hardware, such as a highly integrated circuit (LSI) in which a program is implemented. Alternatively, some of the functions of the processing device 100 can be implemented by dedicated hardware, and the other functions can be implemented by a combination of the processing device and a program.

[0063] The information processing unit 100 is set up as described above.

[0064] Next, the operation of the information processing unit 100 according to embodiment 1 will be described.

[0065] Fig. Figure 9 is a flowchart showing the operation of the information processing unit 100 according to embodiment 1.

[0066] Here, the operation of the information processing equipment 100 corresponds to the information processing procedure, and the program that causes the computer to execute the information processing procedure corresponds to the information processing program.

[0067] First, in an object position information acquisition step, or step S1, the object position information acquisition unit 110 obtains information from the various sensors and receives the position information of objects. More specifically, the object position information acquisition unit 110 obtains the vehicle position information, which shows the position of the vehicle CA, from the localizer 200 and identifies the positions of the obstacles present around the vehicle CA based on the sensor data received from the camera 300, the radar 400, and the LiDAR 500 to obtain the obstacle information, which shows the positions of the obstacles. Additionally, the object position information acquisition unit 110 recognizes the attributes of the obstacles from the sensor data.

[0068] Next, in step 2, the map data acquisition unit 120 obtains map data about the vehicle CA based on the vehicle position information of the vehicle CA obtained by the object position information acquisition unit 110. If the map data has already been obtained from an external source and stored in the information processing unit 100, it is sufficient to simply read the data.

[0069] Next, in step S3, the Static Potential Map Generation Unit 131 generates a static potential map showing the risk of a traffic accident for vehicle CA at the present time. More precisely, the Static Potential Map Generation Unit 131 first obtains the lane center of vehicle CA based on the vehicle position information and the map data. Similarly, the lane center of each obstacle is obtained based on the obstacle information and the map data. Based on the information showing the lane centers and obstacle information obtained as described above, the Static Potential Map Generation Unit 131 generates a potential map in which vehicle CA is centered, as given by Formula 1.

[0070] Next, in an obstacle potential map generation step, or step S4, the obstacle potential map generation unit 140 creates an obstacle potential map indicating the risk of a traffic accident for each obstacle. Steps S4 through S6 are executed in a loop and repeated until processing is complete for all obstacles. The order in which the obstacles are processed is predefined by the designer. This processing order can be arbitrary. For example, one option is to process the obstacles in order of their proximity to the vehicle CA, and another is to process them in descending order of the weighting factor assigned to them based on their attributes in the risk potential calculation.

[0071] Next, in an obstacle movement route prediction step, or step S5, the obstacle movement route prediction unit 150 predicts the movement routes of the obstacles. In step S6, the dynamic potential map generation unit 132 generates a dynamic potential map based on the movement routes of the obstacles predicted by the obstacle movement route prediction unit 150 and the static potential map.

[0072] As described above, the processes from step S4 to step S6 are executed in a loop until processing for all obstacles is complete. To determine whether processing for all obstacles is complete, it is useful to simply count the number of obstacles at the time the object position information retrieval unit 110 first retrieves the obstacle position information, and then simply count the processed obstacles each time processing from step S4 to step S6 is complete to determine whether the number of processed obstacles has reached the total number of obstacles counted in advance. In the above description, the dynamic potential map is generated within the loop.In other words, the predicted movement route is reflected in the vehicle potential map each time the processing of a single obstacle is complete. However, the dynamic potential map can also be generated by reflecting the information about the predicted movement routes all at once in the static potential map after the prediction of movement routes for all obstacles is complete.

[0073] After the above loop processing for all obstacles is completed, the vehicle motion route prediction unit 160 predicts the motion route of the vehicle CA based on the dynamic potential map and transmits the vehicle motion route prediction information to the vehicle control unit 600 in step S7.

[0074] Through the above operations, the information processing device 100 according to embodiment 1 generates the obstacle potential maps, each of which is a potential map in which an obstacle is centered, and predicts the movement routes of the obstacles based on the obstacle potential maps, thereby making it possible to improve the prediction accuracy of the movement routes of the obstacles.

[0075] Furthermore, the information processing device 100 according to embodiment 1 generates the dynamic potential map that indicates the risk of a traffic accident for a vehicle, the risk existing when the obstacles move along the movement paths predicted by the above method. This makes it possible to calculate the risk of a traffic accident for the vehicle CA based on the movement paths of the obstacles predicted by the differentiated method. Furthermore, the information processing device 100 according to embodiment 1 predicts the movement paths of a vehicle based on the aforementioned dynamic potential map. By predicting the movement paths of the vehicle CA based on the movement paths of the obstacles predicted from the obstacle potential maps, the prediction accuracy of the movement paths of the vehicle CA can also be improved.

[0076] For example, if in the Fig. In the situation depicted in Figure 4, assuming that the vehicle MO4 is in a constant acceleration motion, it is predicted that the vehicle MO4 will travel straight ahead. However, the information processing device 100 according to embodiment 1 can also predict a movement path if the obstacle turns right by predicting the obstacle's movement paths using the obstacle potential map. As shown in Figure 4, the information processing device 100 can predict the movement path of the obstacle if the obstacle turns right. Fig. As shown in Figure 7, it is furthermore possible to accurately predict the movement routes of vehicle CA by generating the dynamic potential map showing the risk of a traffic accident for vehicle CA, using the predicted movement route information of the obstacle obtained on the basis of the obstacle potential map, taking into account not only the case in which vehicle MO41 travels straight ahead, but also the case in which it turns right.

[0077] The following section describes the beneficial effects produced by the information processing device 100 in another specific situation, with reference to the Fig. 10, Fig. 11, Fig. 12 to Fig. 13 described. Fig. Figure 10 is a top view showing the vehicle CA and the obstacles around the vehicle CA from above; Fig. Figure 11 is a conceptual diagram showing a concrete example of the obstacle potential map for a vehicle MO101, which corresponds to the situation of Fig. 10 corresponds to; Fig. Figure 12 is a conceptual diagram showing a concrete example of the movement routes of vehicle MO101, based on the in Fig. The obstacle potential map shown in section 11 is predicted; and Fig. Figure 13 is a conceptual diagram showing a concrete example of the dynamic potential map based on the in Fig. The 12 predicted movement routes of the vehicle MO101 are generated.

[0078] In Fig. In scenario 10, vehicle CA is surrounded by vehicle MO101 and vehicle MO102. It is assumed that vehicle MO102 has stopped and is stationary on the road. Assuming that vehicle MO101 is accelerating continuously, it is predicted that if it decelerates, it will stop just before reaching vehicle MO102. However, in practice, vehicle MO101 may swerve to avoid vehicle MO102 and temporarily enter the lane of vehicle CA. At this point, as in Fig. As shown in Figure 11, the information processing device 100 according to embodiment 1 predicts the movement of the vehicle MO102 based on the obstacle potential map, so that it is possible to determine the routes with minimum values ​​of the cost function, which is calculated based on the risk potential values, that the two in Fig. To predict the 12 routes shown, i.e., the route of going straight ahead to stop just in front of vehicle CA, and the route of turning right to enter vehicle CA's lane. As shown in Fig. As shown in Figure 13, the results of predicting the movement of the vehicle MO101 can be reflected in a differentiated way in the vehicle potential map.

[0079] In Fig. 11 is a risk potential RMO102, the risk potential for the vehicle MO102 in Fig. 10; the risk potential RCA is the risk potential for the vehicle CA in Fig. 10; the grey areas on both sides of vehicle MO101 show the risk potential posed by vehicle MO101 deviating from the center of the lane. Fig. 12. The predicted movement route AR121 is represented by the dashed arrow in a case where the vehicle MO101 travels straight ahead, and the predicted movement route AR122 is represented by the dash-dot-dash arrow in a case where the vehicle MO101 turns right.

[0080] In Fig. 13 is the risk potential RMO102, the risk potential for the vehicle MO102 in Fig. 10; a risk potential RMO101 is the risk potential of vehicle MO101 at the present time in Fig. 10; a risk potential RAR121 (a risk potential RAR1211 and a risk potential RAR1212) is the risk potential for the predicted movement route AR121; and a risk potential RAR122 (a risk potential RAR1221 and a risk potential RAR1222) is the risk potential for the predicted movement route AR122. Here, the risk potentials RAR1211 and RAR1221 are the risk potentials of vehicle MO101 at time T + u, and the risk potentials RAR1212 and RAR1222 are the risk potentials of vehicle MO101 at time T + 2u. Furthermore, in the areas on both the left and right sides of vehicle CA, as in the other figures, there is a risk potential due to vehicle CA deviating from the lane center.

[0081] As in the Fig. 10, Fig. 11, Fig. 12 to Fig. As shown in Figure 13, the information processing device 100 according to embodiment 1 can predict the movement routes of vehicle MO101 not only in a case where it travels straight ahead and stops, but also in a case where it turns right and enters the lane of vehicle CA, and can furthermore generate a dynamic potential map, which is the vehicle potential map reflecting the routes in these different cases. By predicting the movement routes based on the dynamic potential map, the movement routes of vehicle CA, including the movement route in a case where vehicle MO101 enters the lane of vehicle CA, can also be correctly predicted.

[0082] Examples of modifications for the information processing unit 100 according to embodiment 1 are described below.

[0083] In the description above, the vehicle potential map generation unit 130 and the obstacle potential map generation unit 140 generate the potential maps using the map data acquired by the map data acquisition unit 120. However, in locations where no map data is available, e.g., on a mountain road or a newly constructed road, the potential maps can only be obtained from the object position data and information acquired from various sensors, without using the map data.

[0084] In the description above, the vehicle potential map generation unit 130 first generates a static potential map and then reflects the obstacle movement route prediction information into the generated static potential map to create the dynamic potential map. However, the dynamic potential map can be generated directly without the intermediate steps. For example, the dynamic potential map can be generated according to Formula 5. R'(x)=∑i=0n∑k=0m{αi∗δk∗exp(−Sik(ri(T+ku),ai(T+ku))−x)2 / σi2)}+ω{1 −exp(−(x1−Yr)2 / σr2)

[0085] In the description above, the vehicle potential map generation unit 130 overlays the risk potentials at several different times along the predicted movement routes of the obstacle onto the vehicle potential map at the present time. However, the vehicle potential map generation unit 130 can generate vehicle potential maps for several different times, so that the risk potentials for these multiple times along the predicted movement routes of the obstacle are displayed on the generated vehicle potential maps. The vehicle potential map can be updated, for example, according to Formula 6 shown below. It should be noted that the phrase "multiple different times" includes both continuous and discontinuous time in its meaning. R'(x,t)=∑i=0n{αi∗δt∗exp(−Sit(r(t),a(t))−x)2 / σi2)}+ω{1−exp(−(x1−Yr)2 / σr2)

[0086] When generating the vehicle potential map according to Formula 6, the vehicle movement route prediction unit 160 can also obtain the movement routes of vehicle CA in the same way as described previously, such that the cost function is minimized. Since the movement routes of vehicle CA are predicted using the time-dependent potential maps, the influence of obstacles located at the same spatial positions at different times can be neglected. This allows the movement routes of the vehicle to be predicted with greater accuracy.

[0087] Furthermore, in the above description, the process in which the obstacle motion route prediction unit 150 predicts the movement routes of the obstacles assumes that all obstacles except the target object are stationary. However, for the untargeted obstacles, the movement routes can be predicted using a conventional motion prediction method such as a Kalman filter. In this case, assuming that the positions of the untargeted obstacles or X k (t) (k ≠ i) move along the movement routes predicted by the conventional prediction method, the obstacle potential map of the target obstacle (k = i) is generated according to formula 7. Ui(x,t)=∑k=0n{αk∗βki∗exp(−(Xk(t)−x)2 / σk2)}+{γk∗exp(−(C−x)2 / σc2)}+ω{1 −exp(−(Xi1−Yi)2 / σr2)

[0088] Alternatively, assuming that the obstacles whose movement routes have already been predicted (k = 0 to i-1) move according to the movement routes predicted by formula 3 and the obstacles whose movement routes have not yet been predicted (k = i+1 to n) move according to the movement routes predicted by using a Kalman filter or the like, the obstacle potential map of the target obstacle (k = i) can be generated according to formula 8. Ui(x,t)=∑k=0i−1{αk∗exp(−(Skt(rk(t),ak(t))−x)2 / σk2)}+∑k=i+1n{αk∗ex p(−(Xk(t)−x)2 / σk2)}+{γk∗exp(−(C−x)2 / σk2)}+ω{1−exp(−(Xi1−Yi)2 / σr2)

[0089] The above description refers to the operation of the in Fig.The information processing unit 100, as depicted in Figure 9, performs steps S4 and S5 for all obstacles. However, the dynamic potential map can only be generated for some of the obstacles, such as the obstacle closest to vehicle CA, by performing steps S4 and S5. For the remaining obstacles, the dynamic potential map can be generated by performing step S6, either by considering them stationary or by assuming they move according to the motion paths predicted by conventional techniques. If motion prediction for all obstacles is performed based on obstacle potential maps, a more differentiated motion prediction can be made.If the motion prediction described above is only performed for selected obstacles, the computation costs can be reduced, while the differentiated motion prediction is performed for the critical obstacles.

[0090] Since the information processing unit 100 in embodiment 1 is an in-vehicle unit mounted on the vehicle CA, it can predict the movement routes of the obstacles and the vehicle CA even in a poor communication environment. Furthermore, because a single information processing unit 100 predicts the movement routes for operating a vehicle CA, the computational effort for this single unit can be reduced. However, if the vehicle CA is moving in an area with good communication and a computer used for the information processing unit 100 has high processing power, the information processing unit 100 can be implemented by a computer located outside the vehicle CA so that the movement prediction results for the obstacles and the vehicle CA can be transmitted to the vehicle CA.

[0091] Although the potentials of the vehicle CA and the obstacles are expressed by normal distributions in the description above, other distribution functions can also be used. For example, it is possible to use a box function or a normal distribution for a stationary object and a distribution function whose form is a Doppler-shifted normal distribution along the direction of travel, etc., for a moving object.

[0092] In the preceding description, the obstacle movement route prediction unit 150 transmits the obstacle movement route prediction information, which shows the predicted movement paths of the obstacles, to the vehicle potential map generation unit 130. The predicted steering values ​​and speeds of the obstacles can be included in the obstacle movement route prediction information to be transmitted. At this point, the vehicle potential map generation unit 130 can adjust the potential values ​​of the obstacles not only based on the positions of the obstacles, but also on their steering values ​​and speeds. Industrial applicability

[0093] The information processing device according to the present disclosure is suitable for use in an automated driving system. Reference symbol list 100 Information processing equipment, 1000 Information processing systems, 200 localizers, 300 camera, 400 radar, 500 LiDAR, 600 vehicle control unit, 110 Object position information acquisition unit, 120 map data acquisition units, 130 Vehicle Potential Map Generation Unit, 131 Static Potential Map Generation Unit, 132 Dynamic Potential Map Generation Unit, 140 Obstacle Potential Map Generation Unit, 150 obstacle movement route prediction unit, 160 Vehicle movement route prediction unit

Claims

Information processing unit (100), comprising: an object position information acquisition unit (110) for acquiring obstacle position information showing the position of at least one obstacle present around a vehicle (CA), and vehicle position information showing the position of the vehicle (CA); an obstacle potential map generation unit (140) for generating an obstacle potential map of a traffic accident risk for the at least one obstacle based on the obstacle position information; and an obstacle movement route prediction unit (150) for predicting a movement route of the at least one obstacle based on the obstacle potential map and for generating obstacle movement route prediction information; and a vehicle potential map generation unit (130).to generate, based on the vehicle position information and the obstacle movement route prediction information, a dynamic potential map of the risk of a traffic accident for the vehicle (CA) in a case where the at least one obstacle moves along the predicted movement route shown by the obstacle movement route prediction information; wherein the object position information acquisition unit (110) acquires position information as the obstacle position information, which includes stationary obstacle position information showing a position of at least one stationary obstacle from the at least one obstacle, and moving obstacle position information showing a position of at least one moving obstacle from the at least one obstacle,the obstacle potential map generation unit (140) as the obstacle potential map generates a moving obstacle potential map of a traffic accident risk for the at least one moving obstacle based on the moving obstacle position information; the obstacle movement route prediction unit (150) as the obstacle movement route prediction information generates movement route prediction information of a moving obstacle, which predicts a movement route of the at least one moving obstacle based on the moving obstacle potential map; and the vehicle potential map generation unit (130) comprises: a static potential map generation unit to generate a static potential map of a traffic accident risk for the vehicle (CA) based on at least the stationary obstacle position information in addition to the vehicle position information.wherein the risk is caused at least by at least one stationary obstacle, and a dynamic potential map generation unit (132) to generate the dynamic potential map based on the static potential map and the motion route prediction information of the moving obstacle. Information processing device (100) according to claim 1, wherein the vehicle potential map generation unit (130) generates the dynamic potential map at each of several different times. Information processing device (100) according to claim 1 or 2, wherein the obstacle movement route prediction unit (150) predicts a plurality of movement routes for each of the at least one obstacle at any given time. Information processing system (1000), comprising: a localizer (200) for localizing the position of a vehicle (CA); a sensor for sensing environments around the vehicle (CA); an object position information acquisition unit (110) for obtaining vehicle position information showing the position of the vehicle (CA) from the localizer (200) and for obtaining obstacle position information showing the position of at least one obstacle present around the vehicle (CA) based on sensor data obtained from the sensor; an obstacle potential map generation unit (140) for generating an obstacle potential map of a traffic accident risk for the at least one obstacle based on the obstacle position information; an obstacle movement route prediction unit (150),to predict a movement route of the at least one obstacle based on the obstacle potential map and to generate obstacle movement route prediction information; a vehicle potential map generation unit (130) to generate, based on the vehicle position information and the obstacle movement route prediction information, a dynamic potential map of a risk of a traffic accident for the vehicle (CA) in a case where the at least one obstacle moves along the predicted movement route shown by the obstacle movement route prediction information; a vehicle movement route prediction unit (160) to predict a movement route of the vehicle (CA) based on the dynamic potential map; and a vehicle control unit (600) to control the vehicle (CA) based on the movement route predicted by the vehicle movement route prediction unit (160).wherein the object position information acquisition unit (110) acquires position information comprising stationary obstacle position information showing the position of at least one stationary obstacle from within the at least one obstacle, and moving obstacle position information showing the position of at least one moving obstacle from within the at least one obstacle; the obstacle potential map generation unit (140) generates a moving obstacle potential map of a traffic accident risk for the at least one moving obstacle based on the moving obstacle position information; and the obstacle movement route prediction unit (150) generates movement route prediction information of a moving obstacle.which predict a movement route of the at least one moving obstacle based on the moving obstacle potential map, and the vehicle potential map generation unit (130) comprises: a static potential map generation unit (131) to generate a static potential map of a traffic accident risk to the vehicle (CA) based on at least the stationary obstacle position information in addition to the vehicle position information, wherein the risk is caused at least by the at least one stationary obstacle, and a dynamic potential map generation unit (132) to generate the dynamic potential map based on the static potential map and the movement route prediction information of the moving obstacle. Information processing procedure comprising: an object position information acquisition step (S1) of acquiring obstacle position information showing the position of at least one obstacle present around a vehicle (CA), and vehicle position information showing the position of the vehicle (CA); an obstacle potential map generation step (S4) of generating an obstacle potential map of a traffic accident risk for the at least one obstacle based on the obstacle position information; and an obstacle movement route prediction step (S5) of predicting a movement route of the at least one obstacle based on the obstacle potential map and generating obstacle movement route prediction information; and a vehicle potential map generation step (S6) of generating, based on the vehicle position information and the obstacle movement route prediction information,to generate a dynamic potential map of a traffic accident risk for the vehicle (CA) in a case where the at least one obstacle moves along the predicted movement route shown by the obstacle movement route prediction information, wherein in the object position information acquisition step (S1) position information, the stationary obstacle position information showing a position of at least one stationary obstacle from the at least one obstacle, and moving obstacle position information showing a position of at least one moving obstacle from the at least one obstacle, are acquired as the obstacle position information,In the obstacle potential map generation step (S4), a moving obstacle potential map of a traffic accident risk for the at least one moving obstacle is generated based on the moving obstacle position information, as the obstacle potential map; in the obstacle movement route prediction step (S5), movement route prediction information of a moving obstacle is generated, which predicts a movement route of the at least one moving obstacle based on the moving obstacle potential map; and the vehicle potential map generation step includes: a static potential map generation step (S3) of generating a static potential map of a traffic accident risk for the vehicle (CA) based on at least the stationary obstacle position information in addition to the vehicle position information.where the risk is caused at least by at least one stationary obstacle, and a dynamic potential map generation step (S6) of generating a dynamic potential map based on the static potential map and the obstacle movement route prediction information. An information processing program to induce a computer to perform an information processing procedure, comprising: an object position information acquisition step (S1) of acquiring obstacle position information showing the position of at least one obstacle present around a vehicle (CA), and vehicle position information showing the position of the vehicle (CA); an obstacle potential map generation step (S4) of generating an obstacle potential map of a traffic accident risk for the at least one obstacle based on the obstacle position information; an obstacle movement route prediction step (S5) of predicting a movement route of the at least one obstacle based on the obstacle potential map and generating obstacle movement route prediction information; and a vehicle potential map generation step (S6) of generating,Based on the vehicle position information and the obstacle movement route prediction information, a dynamic potential map of a traffic accident risk for the vehicle (CA) in a case where the at least one obstacle moves along the predicted movement route shown by the obstacle movement route prediction information; wherein in the object position information acquisition step (S1) position information, the stationary obstacle position information showing a position of at least one stationary obstacle from the at least one obstacle, and moving obstacle position information showing a position of at least one moving obstacle from the at least one obstacle, are acquired as the obstacle position information,In the obstacle potential map generation step (S4), a moving obstacle potential map of a traffic accident risk for the at least one moving obstacle is generated based on the moving obstacle position information, as the obstacle potential map; in the obstacle movement route prediction step (S5), movement route prediction information of a moving obstacle is generated, which predicts a movement route of the at least one moving obstacle based on the moving obstacle potential map; and the vehicle potential map generation step includes: a static potential map generation step (S3) of generating a static potential map of a traffic accident risk for the vehicle (CA) based on at least the stationary obstacle position information in addition to the vehicle position information.where the risk is caused at least by at least one stationary obstacle, a dynamic potential map generation step (S6) of generating the dynamic potential map based on the static potential map and the motion route prediction information of the moving obstacle.

Images