Surrounding environment recognition device and surrounding environment recognition method

Abstract

Improve the accuracy of risk assessment. [Solution] The surrounding environment recognition device 30 has a first calculation unit 322 that calculates the probability of a moving object reaching a divided region A after a unit of time, and the first time when the moving object reaches a divided region A, for each divided region A obtained by dividing the area around the vehicle S into predetermined sizes based on the predicted position where a moving object moves around the vehicle S; a second calculation unit 323 that calculates the second time when the vehicle S reaches each divided region A; and a third calculation unit 324 that calculates a risk level indicating the degree to which the vehicle S and the moving object will come into contact for each divided region A based on the first time and the second time. The third calculation unit 324 adds an additional value corresponding to the probability of reaching the divided region A to the risk level from the unit time before in a divided region A where the time difference between the first time and the second time is less than a threshold, and subtracts a predetermined value from the risk level from the unit time before in a divided region A where no additional value is added.

Description

Technical Field

[0001] The present invention relates to a surrounding environment recognition device and a surrounding environment recognition method.

Background Art

[0002] The surrounding environment recognition device of Patent Document 1 determines, based on the position of a moving object detected by an external sensor provided in the host vehicle and the control model of the host vehicle, a region including the positions of the host vehicle and the moving object at a time after the current time, and the time that can exist within the region. Then, based on the determined region and time, the surrounding environment recognition device calculates a risk level indicating the probability that the host vehicle contacts the moving object for each region where the host vehicle can move.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Since the position of the moving object detected by the external sensor includes observation noise and the control model includes modeling error, the surrounding environment recognition device cannot determine the positions of the host vehicle and the moving object at a time after the current time to an appropriate position. As a result, a problem occurs in that an incorrect risk level is calculated.

[0005] Therefore, the present invention has been made in view of these points, and an object thereof is to improve the accuracy of the risk level.

Means for Solving the Problems

[0006] A surrounding environment recognition device according to a first aspect of the present invention includes: an acquisition unit that acquires vehicle information including the speed of the vehicle and movement prediction information including predicted positions which predict the position where a moving object will move around the vehicle for each unit of time over a predetermined period of time; a first calculation unit that calculates, based on the predicted position, the probability that the moving object will reach the divided region after the unit of time and the first time the moving object will reach the divided region for each divided region of a predetermined size obtained by dividing the area around the vehicle into a predetermined size; a second calculation unit that calculates the second time the vehicle will reach each divided region; and a third calculation unit that calculates a risk level indicating the degree to which the vehicle and the moving object will come into contact for each divided region based on the first time and the second time, wherein the third calculation unit adds an additional value corresponding to the probability of the divided region to the risk level from the unit of time prior to the current time in the divided region where the time difference between the first time and the second time is less than a threshold, and subtracts a predetermined value from the risk level from the unit of time prior to the current time in the divided region where the additional value is not added.

[0007] The second calculation unit may calculate the second time for each of the divided regions based on the distance between each divided region and the vehicle, the speed of the vehicle, and the maximum acceleration of the vehicle stored in the memory unit.

[0008] The third calculation unit may, in the division region from the division region where the time difference is less than the threshold to the division region within a predetermined range, add to the risk level from the unit time before the current time an additional value corresponding to the probability of the division region where the time difference is less than the threshold and the distance from the division region.

[0009] The first calculation unit may calculate the probability based on the probability calculated at a time unit time prior to the current time and the predicted position obtained at the current time.

[0010] The first calculation unit may, if the predicted position is included in the divided region, calculate the first time based on the first time calculated at a time prior to the current time by the unit time and the predicted position acquired at the current time. If the predicted position is not included in the divided region, the first calculation unit may calculate the first time at the current time as a subtracted value obtained by subtracting the unit time from the first time calculated at a time prior to the current time by the unit time.

[0011] The first calculation unit may further calculate the time it takes for the moving body to pass through each of the divided regions based on the predicted position.

[0012] The movement prediction information includes a division value obtained by dividing the sum of the total length of the moving body and the length of one side of the divided area by the speed of the moving body. The first calculation unit may, if the predicted position is included in the divided area, calculate the passage time based on the passage time calculated at a time one unit time prior to the current time and the division value obtained at the current time. If the predicted position is not included in the divided area, the first calculation unit may maintain the passage time calculated at a time one unit time prior to the current time.

[0013] The second calculation unit may calculate the threshold value for each divided region based on the length of one side of the divided region, the sum of the total length of the vehicle stored in the storage unit, and the speed of the vehicle, and the passage time.

[0014] The first calculation unit may generate a moving object map showing the probability and first time for each divided region, and the third calculation unit may generate a risk map showing the risk level for each divided region.

[0015] A second aspect of the present invention relates to a surrounding environment recognition method, which includes an acquisition step in which a computer acquires vehicle information including the speed of the vehicle and movement prediction information including predicted positions, which predict the positions in which moving objects around the vehicle will move at each unit time over a predetermined period of time, for each unit time, and a first calculation step in which, based on the predicted positions, the area around the vehicle is divided into predetermined size division regions, the probability that the moving object will reach the division region after the unit time, and the first time the moving object will reach the division region, for each division region The system includes a second calculation step for calculating a second time at which the vehicle will arrive, and a third calculation step for calculating a risk level indicating the degree to which the vehicle and the moving object will come into contact for each divided region based on the first time and the second time. In the third calculation step, for divided regions where the time difference between the first time and the second time is less than a threshold, an additional value corresponding to the probability of the divided region is added to the risk level from the unit time prior to the current time, and for divided regions where the additional value is not added, a predetermined value is subtracted from the risk level from the unit time prior to the current time. [Effects of the Invention]

[0016] According to the present invention, the accuracy of risk assessment is improved. [Brief explanation of the drawing]

[0017] [Figure 1] This is a diagram illustrating the overview of the surrounding environment recognition system 1. [Figure 2] This figure shows an example of a processing sequence in the control unit 32. [Figure 3] This figure shows an example of a processing sequence for generating a moving object map. [Figure 4] This diagram illustrates the general operation of moving arrival prediction information. [Figure 5] This figure shows an example of a processing sequence for updating arrival prediction information. [Figure 6] This figure shows an example of a processing sequence for deleting arrival prediction information. [Figure 7]This figure shows an example of a processing sequence for calculating the contact position. [Figure 8] This diagram illustrates the process of calculating the second time point, te. [Figure 9] This figure shows an example of a processing sequence for generating a risk map. [Figure 10] This figure shows an example of a processing sequence for adding a value to the risk level. [Figure 11] This figure shows an example of a processing sequence for subtracting a deduction value from the risk level. [Modes for carrying out the invention]

[0018] <Overview of the surrounding environment recognition system 1> Figure 1 is a diagram illustrating the overview of the surrounding environment recognition system 1. The surrounding environment recognition system 1 shown in Figure 1 comprises a vehicle sensor 10, an external sensor 11, a moving object recognition device 12, a path generation device 20, a driving control device 21, an actuator 22, and a surrounding environment recognition device 30. The surrounding environment recognition system 1 is a system installed in an autonomously driven vehicle (hereinafter referred to as "the vehicle") and has the function of steering the vehicle so that it moves along a path that makes it less likely to come into contact with moving objects around it. Moving objects include, for example, pedestrians, light vehicles such as bicycles, and other vehicles different from the vehicle. The area around the vehicle is, for example, an area within a radius of 40 meters from the center of the vehicle, but this radius may be different from 40 meters.

[0019] The vehicle sensor 10 is a sensor for detecting when the vehicle is moving, and includes a speed sensor. The vehicle sensor 10 outputs vehicle information, including the vehicle's speed detected by the speed sensor, to the surrounding environment recognition device 30.

[0020] The external sensor 11 includes a camera, a LiDAR (Light Detection and Ranging) sensor or a millimeter-wave sensor, and a GNSS (Global Navigation Satellite System) receiver. The external sensor 11 outputs, for example, the captured image generated by the camera capturing images of the area around the vehicle, the distance between the vehicle and the moving object detected by the LiDAR sensor or millimeter-wave sensor, and the position of the vehicle received by the GNSS receiver to the moving object recognition device 12.

[0021] The moving object recognition device 12 is a device that calculates a predicted position by predicting the location where a moving object in the vicinity of the vehicle will move. For example, the moving object recognition device 12 calculates the position and speed of the moving object based on the distance between the vehicle and the moving object and the position of the vehicle, which are acquired from the external sensor 11 at unit time intervals, and calculates the predicted position if the moving object moves from that position at that speed. The unit time is, for example, 0.1 seconds. The moving object recognition device 12 then outputs the movement prediction information, including the calculated predicted position, to the surrounding environment recognition device 30.

[0022] The route generation device 20 is a device that generates a route for the vehicle to travel. The route generation device 20 obtains, for example, the degree of risk, which is the degree to which the vehicle and other vehicles are likely to come into contact in each area around the vehicle, from the surrounding environment recognition device 30. The route generation device 20 generates a route such that the vehicle travels through the area with the lowest risk on the road it is traveling on, and outputs the route to the driving control device 21.

[0023] The driving control device 21 is a device that controls the path and speed of the vehicle by controlling the actuator 22. For example, the driving control device 21 calculates the steering angle and speed required for the vehicle to move along the path generated by the path generation device 20. The driving control device 21 generates control information for the actuator 22 corresponding to the calculated steering angle and speed, and outputs this control information to the actuator 22.

[0024] The actuator 22 includes an actuator for controlling the steering angle, an actuator for controlling the opening and closing of the engine's throttle valve, and an actuator for controlling the braking force of the brakes. The actuator 22 moves the vehicle along the path generated by the path generation device 20 by controlling each actuator based on control information acquired from the driving control device 21.

[0025] The surrounding environment recognition device 30 is a computer that calculates the degree of danger in each area around the vehicle in order for the path generation device 20 to generate a path for the vehicle to move along. For example, the surrounding environment recognition device 30 calculates a first time when a moving object will reach each area around the vehicle based on the predicted position obtained from the moving object recognition device 12, and calculates a second time when the vehicle will reach each area based on the speed obtained from the vehicle sensor 10. Then, the surrounding environment recognition device 30 calculates the degree of danger based on the difference between the first time and the second time.

[0026] The speed obtained from the vehicle sensor 10 includes observation noise from the speed sensor included in the vehicle sensor 10, and the predicted position obtained from the moving object recognition device 12 includes observation noise from the millimeter-wave sensor or LiDAR sensor included in the external sensor 11. Furthermore, if the moving object recognition device 12 uses a vehicle model that shows the movement of its own vehicle in calculating the predicted position, or if the surrounding environment recognition device 30 uses the same vehicle model in calculating the second time, the predicted position and the second time will include modeling errors. As a result, the surrounding environment recognition device 30 is prone to calculating incorrect risk levels.

[0027] Therefore, the surrounding environment recognition device 30 calculates the probability that a moving object will reach each surrounding area after a unit of time (hereinafter referred to as "arrival probability") based on the arrival probability calculated at a time unit prior to the current time and the predicted position acquired at the current time. Then, the surrounding environment recognition device 30 calculates the risk level based on the calculated arrival probability and the risk level calculated at a time unit prior to the current time. By operating in this manner, the surrounding environment recognition device 30 can reduce the amount of change in the risk level calculated at each unit of time, thereby reducing the error in calculating the risk level due to observation noise and modeling errors. As a result, the surrounding environment recognition device 30 can improve the accuracy of the risk level. The configuration and operation of the surrounding environment recognition device 30 will be described in detail below.

[0028] <Configuration of the surrounding environment recognition device 30> As shown in Figure 1, the surrounding environment recognition device 30 has a storage unit 31 and a control unit 32. The control unit 32 has an acquisition unit 321, a first calculation unit 322, a second calculation unit 323, and a third calculation unit 324.

[0029] The memory unit 31 has a storage medium such as ROM (Read Only Memory), RAM (Random Access Memory), HDD (Hard Disk Drive), or SSD (Solid State Drive). The memory unit 31 stores various information for generating the program executed by the control unit 32 and the risk map.

[0030] The control unit 32 is a processor such as a CPU (Central Processing Unit) or an ECU (Electronic Control Unit). By executing a program stored in the storage unit 31, the control unit 32 functions as an acquisition unit 321, a first calculation unit 322, a second calculation unit 323, and a third calculation unit 324. The control unit 32 may consist of a single processor, or it may consist of multiple processors or a combination of one or more processors and electronic circuits.

[0031] Figure 2 shows an example of a processing sequence in the control unit 32. The processing sequence shown in Figure 2 is a processing sequence that shows the operation of the control unit 32 to calculate the risk level. First, the control unit 32 generates a moving object map (S1). The moving object map is a map that shows, for each divided region of a predetermined size obtained by dividing the area around the vehicle into a divided region, the probability of a moving object reaching the divided region after a unit of time, the first time the moving object reaches the divided region, and the passage time the moving object passes through the divided region. The predetermined size is, for example, a square with sides of 0.5m. If there are multiple moving objects around the vehicle, the moving object map shows the probability of arrival, the first time, and the passage time for each moving object in each divided region.

[0032] Next, the control unit 32 calculates the position where the vehicle and the moving object will come into contact (S2). The control unit 32 calculates the second time when the vehicle reaches each divided region, and calculates the contact position based on the second time and the first time. If there are multiple moving objects around the vehicle, the control unit 32 calculates the contact position for each moving object. Then, the control unit 32 generates a risk map based on the moving object map and the predicted contact position (S3). The risk map is a map that shows the degree of risk, which is the degree to which the vehicle and other vehicles come into contact in each divided region obtained by dividing the area around the vehicle into predetermined sizes. If there are multiple moving objects around the vehicle, the control unit 32 generates a risk map that shows the risk for each moving object in each divided region. The configuration and operation of each part included in the control unit 32 will be described in detail below.

[0033] The acquisition unit 321 acquires vehicle information, including the vehicle's speed, and predicted movement information, including predicted positions, which predict the positions of moving objects around the vehicle for each unit of time within a predetermined period. For example, the predetermined period is 5 seconds, and the unit time is 0.1 seconds. For example, the acquisition unit 321 acquires vehicle information from the vehicle sensor 10 and predicted movement information from the moving object recognition device 12 for each unit of time.

[0034] The acquisition unit 321 may acquire motion prediction information from the moving object recognition device 12, which further includes the speed limit of the road on which the vehicle is traveling, and a division value obtained by dividing the sum of the total length of the moving object and the length of one side of the divided area by the speed of the moving object. The moving object recognition device 12, for example, acquires an image generated by a camera included in the external sensor 11 and identifies the speed limit indicated by the regulatory sign included in the image. The moving object recognition device 12 calculates the division value by, for example, calculating the total length of the moving object and the speed of the moving object based on point cloud data acquired from a LiDAR sensor or millimeter-wave sensor included in the external sensor 11.

[0035] The first calculation unit 322 generates a moving object map for each unit of time based on the predicted position included in the movement prediction information acquired by the acquisition unit 321 (step S1 in Figure 2). For example, for each divided region obtained by dividing the area around the vehicle into predetermined sizes, the first calculation unit 322 calculates the probability of the moving object reaching the divided region after a unit of time, the first time the moving object reaches the divided region, and the time the moving object passes through the divided region. The first calculation unit 322 then generates a moving object map showing the probability of arrival, the first time, and the time of passage for each divided region, and stores the moving object map in the storage unit 31. In the following description, the probability of arrival, the first time, and the time of passage shown in the moving object map may be collectively referred to as "predicted arrival information."

[0036] Figure 3 is a diagram showing an example of a processing sequence for generating a moving object map. The processing sequence shown in Figure 3 is a processing sequence that executes the process of step S1 shown in Figure 2. The first calculation unit 322 moves the arrival prediction information from a unit time ago, contained in each divided region of the moving object map, by referring to the moving object map stored in the storage unit 31 (S10). Figure 4 is a diagram illustrating the overview of the operation of moving the arrival prediction information. Figure 4 shows the vehicle S and divided regions A01 to A16 at time T0, a unit time ago, and the vehicle S and divided regions A01 to A16 at the current time, time T1. In Figure 4, for the sake of simplicity, the divided region A in front of the vehicle S is shown among the divided regions A around the vehicle S.

[0037] The first calculation unit 322 determines, for example, that the vehicle S moved a distance D in the direction of travel during the unit time from time T0 to time T1, based on the control information calculated by the driving control device 21 and the vehicle information acquired by the acquisition unit 321. The first calculation unit 322 determines, for example, that the divided region A04 at time T0 corresponds to the divided region A12 at time T1, based on the direction and distance D of the vehicle S's movement, and moves the predicted arrival information from the unit time prior contained in the divided region A04 to the divided region A12. The first calculation unit 322 moves the predicted arrival information for other divided regions different from the divided region A04 in the same way as for the divided region A04.

[0038] Returning to Figure 3, the first calculation unit 322 acquires movement prediction information from the acquisition unit 321 (S11). The first calculation unit 322 selects one unselected division area A from among the division areas A around its own vehicle (S12), and executes a process to update the arrival prediction information included in the selected division area A (S13). Figure 5 is a diagram showing an example of a processing sequence for updating arrival prediction information. The processing sequence shown in Figure 5 is a processing sequence that executes the process of step S13 shown in Figure 3.

[0039] The first calculation unit 322 determines whether the moving object will reach the divided region A selected in step S12 based on the predicted position included in the movement prediction information acquired from the acquisition unit 321 (S130). For example, the first calculation unit 322 determines that the moving object will reach the divided region A if the predicted position is included in the divided region A, and determines that the moving object will not reach the divided region A if the predicted position is not included in the divided region A.

[0040] If the first calculation unit 322 determines that a moving object will reach its destination (YES in S130), it selects one of the one or more moving objects that it determined will reach its destination (S131). The first calculation unit 322 then calculates the probability of reaching the divided region A selected in step S12 and updates the value to that probability (S132). Subsequently, the first calculation unit 322 calculates the first time in the divided region A and updates the value to that first probability (S133), and calculates the time taken to pass through the divided region A and updates the value to that time (S134). Next, the process in which the first calculation unit 322 calculates the arrival probability, the first time, and the passage time when it is determined that the moving body has reached the divided area A will be described.

[0041] First, the process (S132) in which the first calculation unit 322 calculates the arrival probability will be described. In the following description, it is assumed that an index is assigned to each divided area A, and the event that the moving body reaches the divided area A with index i is m i = 1, and the event that the moving body does not reach the divided area A with index i is m i = 0. Then, the probability that the moving body reaches the divided area A with index i is represented as P(m i = 1) or P(m i ). Further, whether or not the predicted position at the current time reaches the divided area A is represented by the value indicated by z t . Specifically, when it is determined that the predicted position reaches the divided area A, z t indicates 1, and when it is determined that the predicted position does not reach the divided area A, z t indicates 0.

[0042] The first calculation unit 322 updates the arrival probability by calculating the arrival probability based on the arrival probability calculated at the time one unit time before the current time and the predicted position acquired at the current time. The first calculation unit 322 updates the arrival probability one unit time before, for example, by applying a binary Bayes filter. The probability that the moving body reaches the divided area A when z1 to z n regarding the grid with index i up to time t = t tn is obtained can be expressed as in Equation (1) using conditional probability.

Equation

[0043] Equation (1) can be expressed as Equations (2) and (3) when expressed in logarithmic odds after being transformed by Bayes' theorem.

number

[0044] As described above, by using log odds to represent the probability of arrival, the first calculation unit 322 can calculate the probability of arrival by adding the calculated value based on the predicted position acquired at the current time to the probability of arrival from a unit time ago. As a result, the first calculation unit 322 can reduce the error due to observation noise compared to calculating the probability of arrival for each unit time. Furthermore, by using log odds to represent the probability of arrival, the first calculation unit 322 can express the probability of arrival in the range of "-∞ to ∞" instead of "0 to 1". As a result, the first calculation unit 322 can suppress loss of precision and rounding errors when calculating the probability of arrival.

[0045] Next, the process (S133) by which the first calculation unit 322 calculates the first time will be explained. The first time is, for example, the time it takes for the moving object to reach each divided region A (for example, a time such as "2.0 seconds"). If the predicted position is included in the divided region A, the first calculation unit 322 calculates the first time based on the first time calculated at a time one unit time before the current time and the predicted position acquired at the current time. The first calculation unit 322 calculates the first time based on the first time one unit time before and the predicted position acquired at the current time, for example, by applying a Kalman filter.

[0046] Using the first time step τ(k) at time k, the state model at the first time step can be expressed as shown in equation (4), and the observation model at the first time step can be expressed as shown in equation (5).

number

[0047] TIFF2026074748000005.tif25170

number

[0048] Furthermore, the optimal Kalman gain based on the prior variance is given by equation (8), and the estimate of the first time (i.e., the first time) is given by equation (9).

number

number

[0049] The optimal Kalman gain g(k) is updated and converges each time the above process is performed, but the system represented by equations (4) and (5) is time-invariant. Therefore, the first calculation unit 322 calculates the converged value g shown in equation (11). optYou may have calculated this in advance.

number

[0050] Next, the process (S134) in which the first calculation unit 322 calculates the passage time will be explained. If the predicted position is included in the divided region A, the first calculation unit 322 calculates the passage time based on the passage time calculated at a time unit before the current time and the division value obtained at the current time. The division value is the division value included in the movement prediction information acquired by the acquisition unit 321. For example, the first calculation unit 322 applies a Kalman filter to calculate the passage time based on the passage time from a unit before the current time and the division value obtained at the current time.

[0051] Time of passage Δτ at time k k Using this, the state model of transit time can be expressed as shown in equation (12), and the observation model of transit time can be expressed as shown in equation (13).

number

[0052] TIFF2026074748000011.tif25170

number

[0053] Furthermore, the optimal Kalman gain based on prior variance is given by equation (16), and the estimated transit time (i.e., transit time) is given by equation (17).

number

number

[0054] The optimal Kalman gain g(k) is updated and converges each time the above process is performed, but the system represented by equations (12) and (13) is time-invariant. Therefore, the first calculation unit 322 calculates the converged value g shown in equation (19). opt You may have calculated this in advance.

number

[0055] The first calculation unit 322, if there are unselected moving objects (NO in S135), selects an unselected moving object (S131), calculates the probability of arrival, first time, and passage time of the moving object, and updates the calculated probability of arrival, first time, and passage time (S132-S134). If all moving objects are selected (YES in S135), the first calculation unit 322 terminates the process of updating the arrival prediction information.

[0056] If the first calculation unit 322 determines that the moving object will not reach its destination (NO in S130), it calculates the probability of reaching the divided region A selected in step S12 using equations (2) and (3), similar to step S132, and updates the probability of reaching the destination (S136). Subsequently, the first calculation unit 322 calculates the first time for the divided region A and updates the value to the first time, maintaining the passage time for the divided region A. If, for example, the predicted position is not included in the divided region A, the first calculation unit 322 calculates a subtracted value by subtracting a unit of time from the first time calculated at a time unit time prior to the current time, and updates the first time to the subtracted value (S137). If, for example, the predicted position is not included in the divided region, the first calculation unit 322 maintains the passage time calculated at a time unit time prior to the current time (S138). Then, the first calculation unit 322 finishes the process of updating the arrival prediction information.

[0057] Returning to Figure 3, the first calculation unit 322, if there are unselected division regions A (NO in S14), selects one unselected division region A (S12) and repeats the operation (S13) of updating the arrival prediction information contained in the division region A. If all division regions A have been selected (YES in S14), the first calculation unit 322 selects one division region A again (S15) and, if the arrival prediction information contained in the division region A is unnecessary, executes the operation of deleting the arrival prediction information (S16). "When arrival prediction information is unnecessary" means, for example, if the moving object passed through division region A at a time earlier than the current time, it is clear that the moving object will not reach division region A.

[0058] Figure 6 shows an example of a processing sequence for deleting arrival prediction information. The processing sequence shown in Figure 6 is a processing sequence that executes the process of step S16 shown in Figure 3. The first calculation unit 322 determines whether or not there is arrival prediction information for the divided region A selected in step S15 (S160). If there is no arrival prediction information (NO in S160), the first calculation unit 322 terminates the process of deleting the arrival prediction information. On the other hand, if there is arrival prediction information (YES in S160), the first calculation unit 322 selects one of the one or more moving objects that will reach the divided region A selected in step S15 (S161) and identifies the arrival prediction information corresponding to that moving object.

[0059] The first calculation unit 322 determines that if the first time included in the identified arrival prediction information is less than 0 (NO in S162), either the moving object arrived at a time earlier than the current time, or the arrival prediction information was incorrect, and deletes the arrival prediction information (S163). The first calculation unit 322 also deletes the arrival prediction information if the first time is 0 or greater (YES in S162), but the arrival probability included in the identified arrival prediction information is less than the probability threshold (NO in S164) (S163). The probability threshold is, for example, a value corresponding to 50% (for example, 0.5 in the range of 0 to 1), and is stored in the storage unit 31. The first calculation unit 322 does not delete the arrival prediction information if the first time is 0 or greater (YES in S162) and the arrival probability is greater than or equal to the probability threshold (YES in S164).

[0060] If there are any unselected moving objects among the moving objects that reach the divided region A selected in step S15 (NO in S165), the first calculation unit 322 selects one of the unselected moving objects (S161) and repeats the process from step S162 to step S164. On the other hand, if the first calculation unit 322 has selected all of the one or more moving objects that reach the divided region A (YES in S165), it terminates the process of deleting the arrival prediction information.

[0061] Returning to Figure 3, if there are still unselected divided regions A (NO in S17), the first calculation unit 322 selects one of the unselected divided regions A (S15), and if the arrival prediction information contained in that divided region A is unnecessary, it executes a process to delete the arrival prediction information (S16). On the other hand, if all divided regions A are selected again (YES in S17), the first calculation unit 322 terminates the process of generating the moving object map.

[0062] The second calculation unit 323, shown in Figure 1, calculates the position where the vehicle and the moving object come into contact based on the moving object map generated by the first calculation unit 322 and the vehicle information acquired by the acquisition unit 321 (step S2 in Figure 2). The second calculation unit 323 calculates the second time at which the vehicle arrives for each divided region A. Then, for example, the second calculation unit 323 calculates the position where the vehicle and the moving object come into contact based on the time difference between the second time at which the vehicle arrives for each divided region A and the first time for each divided region A included in the moving object map.

[0063] Figure 7 shows an example of a processing sequence for calculating the contact position. The processing sequence shown in Figure 7 is a processing sequence that executes the process of step S2 shown in Figure 2. The second calculation unit 323 obtains the mobile object map generated by the first calculation unit 322 at the time closest to the current time by referring to the storage unit 31 (S20). The second calculation unit 323 obtains the vehicle information obtained by the acquisition unit 321 (S21).

[0064] The second calculation unit 323 selects one of the divided regions A around the vehicle indicated by the moving object map acquired in S20 (S22). Based on the distance between each divided region A and the vehicle, the vehicle's speed included in the vehicle information, and the vehicle's maximum acceleration stored in the storage unit 31, the second calculation unit 323 calculates the second time te that the vehicle will reach for each divided region A (S23). The second time te is, for example, the time it takes for the vehicle to reach each divided region A (for example, a time such as "2.0 seconds"). The following describes the process by which the second calculation unit 323 calculates the second time te.

[0065] Figure 8 is a diagram illustrating the process of calculating the second time point te. Figure 8 shows position O, which indicates the center of the axle of the drive wheel of the vehicle S; position P, which indicates the intersection of the diagonals of the divided region A; and the trajectory L when the vehicle S moves to position P. The coordinates of position O are (0,0) and the coordinates of position P are (x,y). The trajectory L is the path taken by the vehicle S when it moves to the divided region A, calculated based on the vehicle model of the vehicle S, and the length of the trajectory L is the distance between the divided region A and the vehicle S. In this embodiment, for the sake of simplicity, the length of the trajectory L is taken as the arc length of a sector with radius r and angle θ. Note that in Figure 8, position O is shown when the rear wheels of the vehicle S are the drive wheels.

[0066] In Figure 8, the length of the trajectory L, the radius r, and the angle θ are expressed by equations (20) to (22).

number

[0067] Then, the second time te is calculated using the trajectory L calculated by the second calculation unit 323 and the speed limit v included in the movement prediction information. limit And the maximum acceleration a stored in the memory unit 31 max Using the vehicle information and the speed v0 of the vehicle S, the results are expressed as shown in equations (23) and (24). Equation (23) states that the speed of the vehicle S when it reaches position P is v limit The second time te is shown in the case where the speed of the vehicle S is not reached, and equation (24) is given by the time the vehicle S reaches position P if the speed of the vehicle S is v limit This shows the second time te when the value is reached.

number

[0068] The second calculation unit 323 calculates, for example, the first second time te calculated using equation (23) and the second second time te calculated using equation (24). Then, the second calculation unit 323 calculates the maximum acceleration a max The sum obtained by multiplying the first second time te by the speed v0 of the vehicle S is the speed limit v limit The second calculation unit 323 determines whether the following is true. The second calculation unit 323 determines whether the added value is the speed limit v limit If the following conditions are met, the first second time te is determined to be the second time te, and the added value is the speed limit v limit If it exceeds this value, the second time te is determined to be the second time te. The process by which the second calculation unit 323 calculates the second time te has been explained above.

[0069] Returning to Figure 7, the second calculation unit 323 selects one of the one or more moving objects that will reach the selected divided region A (S24), and identifies the arrival prediction information corresponding to that moving object. The second calculation unit 323 identifies the first time t included in the identified arrival prediction information. o The second time t was calculated as follows: e The time difference with the time threshold t lim Determine whether it is less than (S25). Time threshold t lim This is a threshold value calculated by the second calculation unit 323 for each divided region A, based on the sum of the length of one side of the divided region A and the total length of the vehicle S stored in the storage unit 31, divided by the speed v0 of the vehicle S, and the passage time included in the arrival prediction information.

[0070] The second calculation unit 323 uses, for example, equation (26) to determine the time threshold t lim Calculate Δt and perform the determination shown in equation (25) in step S25. o This is the transit time included in the arrival prediction information, v e Δt is the speed of the vehicle S, h is the total length of the vehicle S, l is the length of one side of the divided region A, and Δt m This is the margin (allowance time).

number

[0071] The second calculation unit 323 calculates the first time t o and the second time t e The time difference with the time threshold t lim If it is less than (YES in S25), the index corresponding to the divided region A selected in step S22 is stored in the storage unit 31 as the index of the divided region A in contact with the moving object and the vehicle S (S26). Meanwhile, the second calculation unit 323 calculates the first time t o and the second time t e The time difference with the time threshold t lim If the above conditions are met (NO in S25), the index corresponding to the partitioned area A selected in step S22 is not stored in the storage unit 31.

[0072] If there are one or more moving bodies that reach the division region A selected in step S22 that have not been selected (NO in S27), the second calculation unit 323 selects one of the unselected moving bodies (S24) and executes steps S25 and S26. If the second calculation unit 323 has selected all moving bodies that reach the division region A selected in step S22 (YES in S27), it proceeds to step S28.

[0073] If there are unselected division regions A (NO in S28), the second calculation unit 323 selects one of the unselected division regions A (S22) and repeats the process from step S23 to step S27. If all division regions A are selected (YES in S28), the second calculation unit 323 terminates the process of calculating the position where the vehicle S and the moving object come into contact.

[0074] The third calculation unit 324 shown in Figure 1 calculates the first time t o and the second time t eBased on this, the third calculation unit 324 calculates the risk level indicating the degree to which the vehicle S and the moving object will come into contact for each divided region A. Then, the third calculation unit 324 generates a risk level map showing the risk level for each divided region A (step S3 in Figure 2). The third calculation unit 324 generates, for example, a first time t o and the second time t e Based on the second calculation unit 323, which calculates the position where the vehicle S and the moving object come into contact, the third calculation unit 324 generates a risk map by updating the risk map generated a unit time earlier. The third calculation unit 324 stores the risk maps generated for each unit time in the storage unit 31, for example.

[0075] Figure 9 shows an example of a processing sequence for generating a risk map. The processing sequence shown in Figure 9 is a processing sequence that executes the process of step S3 shown in Figure 2. The third calculation unit 324 obtains a risk map from a unit time ago by referring to the storage unit 31 (S30). The third calculation unit 324 obtains an index corresponding to the divided area A where the vehicle S and the moving object come into contact, which has been stored in the storage unit 31 by the second calculation unit 323, by referring to the storage unit 31 (S31).

[0076] The third calculation unit 324 selects one index from the one or more acquired indices (S32) and identifies the division area A corresponding to that index and the division areas A surrounding that division area A (S33). The surrounding division areas A are, for example, division areas A that include the intersection of the diagonals of the division area A corresponding to the selected index, within a radius of 10 meters from the intersection of the diagonals of the division area A corresponding to the selected index. The third calculation unit 324 then performs a process (S34) to add an additional value to the risk level of the identified division area A from a unit time ago.

[0077] Figure 10 shows an example of a processing sequence for adding a value to the risk level. The processing sequence shown in Figure 10 is a processing sequence that executes the process of step S34 shown in Figure 9. The third calculation unit 324 selects one of the multiple divided regions A identified in step S33 (S340). The third calculation unit 324 calculates the added value J of the selected divided region A using formula (27) (S341).

number

[0078] The third calculation unit 324 adds the calculated sum J to the risk level from the unit time before (S342). The third calculation unit 324 calculates the risk level from the first time t o and the second time t e The time difference with the time threshold t lim In a divided region A less than , an additional value J corresponding to the probability p of reaching the divided region A is added to the risk level one unit time before the current time. That is, the third calculation unit 324 calculates an additional value J corresponding to the probability p of reaching the divided region A by substituting 0 for d in equation (27) in the divided region A corresponding to the index selected in step S32, and adds the additional value J to the risk level one unit time before the divided region A.

[0079] By operating as described above, the third calculation unit 324 can calculate the risk level using the risk level calculated a unit time earlier. As a result, even if the sum value J calculated at the current time includes errors due to observation noise from the vehicle sensor 10 and the external sensor 11, as well as modeling errors of the vehicle model, the third calculation unit 324 can reduce the error in calculating the risk level.

[0080] The third calculation unit 324 calculates that the time difference is equal to the time threshold t limIn the division region A between a division region A less than a certain range and a division region A within a predetermined range, the risk level of a unit time prior to the current time is calculated based on the time difference with respect to the time threshold t. lim The probability p of reaching a division region A less than a certain value and the sum J corresponding to the distance d from the division region A are added. The predetermined range is, for example, within a radius of 10m from the intersection of the diagonals of division region A, and the division region A within the predetermined range is the "surrounding division region A" identified in step S33.

[0081] In other words, the third calculation unit 324 calculates an added value J by substituting the distance to the division area A corresponding to the index selected in step S32 into d of equation (27) and substituting the probability p of reaching the division area A into p of equation (27). The third calculation unit 324 then adds the added value J to the risk level of each division area A in the surrounding area. By operating in this manner, the third calculation unit 324 can increase the risk level of the division area A surrounding the division area A that is estimated to be in contact with the vehicle S and the moving object, thereby reducing the error in calculating the risk level even if there is an error in the estimated position of the division area A.

[0082] If there are any unselected division regions A among the multiple division regions A identified in step S33 (NO in S343), the third calculation unit 324 selects one of the unselected division regions A (S340) and executes steps S341 and S342. If all of the division regions A identified in step S33 are selected (YES in S343), the third calculation unit 324 terminates the process of adding the added value J to the risk level.

[0083] Returning to Figure 9, the third calculation unit 324 selects one of the unselected indices (S32) if there are any unselected indices among the one or more indices acquired in step S31 (NO in S35). Then, the third calculation unit 324 repeats steps S32 to S34. If the third calculation unit 324 selects all of the one or more indices acquired in step S31 (YES in S35), it identifies the index corresponding to the divided area A around the vehicle S that has not had its risk level added (S36). The third calculation unit 324 identifies the divided area A corresponding to the identified index (S37) and calculates the risk level of the divided area A by subtracting a subtraction value from the risk level of the divided area A from a unit time ago (S38).

[0084] Figure 11 shows an example of a processing sequence for subtracting a subtraction value from the risk level. The processing sequence shown in Figure 11 is a processing sequence that executes the process of step S38 shown in Figure 9. The third calculation unit 324 selects one of the one or more divided regions A identified in step S37 (S380) and subtracts a predetermined value (subtraction value) from the risk level of the divided region A one unit time ago (S381). The predetermined value is, for example, a value corresponding to 3% (for example, 0.03 in the range of 0 to 1). That is, in divided regions A where the addition value J is not added, the third calculation unit 324 subtracts a predetermined value from the risk level one unit time ago from the current time.

[0085] If there are any unselected division areas A among the one or more division areas A identified in step S37 (NO in S382), the third calculation unit 324 selects one of the unselected division areas A (S380) and executes step S381. If the third calculation unit 324 has selected all of the division areas A identified in step S37 (YES in S382), it terminates the process of subtracting the subtraction value from the risk level. Then, the third calculation unit 324 terminates the process of generating the risk level map shown in Figure 9 and stores the generated risk level map in the storage unit 31.

[0086] <Variation> In the above description, an example configuration was given in which the first calculation unit 322 generates a moving object map and the third calculation unit 324 creates a risk map, but the system is not limited to this configuration. The first calculation unit 322 may store the calculated arrival probability, first time, and passage time in the storage unit 31 in association with the index of the divided area A, without creating a moving object map. The third calculation unit 324 may store the calculated risk in the storage unit 31 in association with the index of the divided area A, without creating a risk map.

[0087] <Effects of the surrounding environment recognition device 30> As explained above, the surrounding environment recognition device 30 includes an acquisition unit 321 that acquires vehicle information including the speed of the vehicle S, and movement prediction information including predicted positions which predict the positions where moving objects around the vehicle S will move for each unit of time over a predetermined period of time, on a unit-time basis, and for each divided region A obtained by dividing the area around the vehicle S into predetermined sizes based on the predicted position, it acquires the arrival probability p of a moving object reaching a divided region A after a unit of time, and the first time t of the moving object reaching a divided region A. o The first calculation unit 322 calculates the second time t that the vehicle S arrives at for each divided region A. e A second calculation unit 323 calculates the first time t o and the second time t e The system includes a third calculation unit 324 that calculates a risk level indicating the degree to which the vehicle S and the moving object come into contact for each divided region A, based on the above.

[0088] Then, the third calculation unit 324 calculates the first time t o and the second time t e The time difference with the time threshold t lim In a divided region A less than the current time, an additional value J corresponding to the probability p of reaching the divided region A is added to the risk level calculated a unit time before the current time. In a divided region A where the additional value J is not added, the third calculation unit 324 subtracts a predetermined value from the risk level calculated a unit time before the current time.

[0089] With the surrounding environment recognition device 30 configured in this way, the surrounding environment recognition device 30 can calculate the degree of danger based on the degree of danger from a unit time ago and the sum J corresponding to the calculated probability of arrival p. As a result, even if the calculated probability of arrival p includes errors due to observation noise from the vehicle sensor 10 and the external sensor 11, and modeling errors of the vehicle model, the surrounding environment recognition device 30 can reduce the error in calculating the degree of danger by using the degree of danger calculated from a unit time ago. Therefore, the surrounding environment recognition device 30 can improve the accuracy of the degree of danger.

[0090] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of its gist. For example, all or part of the apparatus can be configured by functionally or physically distributing and integrating in any unit. Furthermore, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effects of the new embodiments resulting from the combinations are combined with the effects of the original embodiments. [Explanation of Symbols]

[0091] 1. Surrounding Environment Recognition System 10 Vehicle sensors 11. External sensors 12 Mobile object recognition device 20 Route generation device 21 Driving control device 22 Actuators 30. Surrounding Environment Recognition Device 31 Storage section 32 Control Unit 321 Acquisition Department 322 First Calculation Unit 323 Second Calculation Unit 324 Third Calculation Unit

Claims

1. An acquisition unit acquires vehicle information, including the speed of the vehicle, and movement prediction information, including predicted position, which predicts the position of moving objects around the vehicle for each unit of time within a predetermined period, for each unit of time. Based on the predicted position, a first calculation unit calculates, for each divided region obtained by dividing the area around the vehicle into predetermined sizes, the probability that the moving body will reach the divided region after the unit time, and the first time the moving body will reach the divided region. A second calculation unit calculates the second time the vehicle arrives at each of the divided regions, It includes a third calculation unit that calculates a degree of risk indicating the degree to which the vehicle and the moving object come into contact for each divided region, based on the first time and the second time. The third calculation unit, in the divided region where the time difference between the first time and the second time is less than a threshold, adds an additional value corresponding to the probability of the divided region to the risk level from the unit time prior to the current time, and in the divided region where the additional value is not added, subtracts a predetermined value from the risk level from the unit time prior to the current time. A device for recognizing the surrounding environment.

2. The second calculation unit calculates the second time for each of the divided regions based on the distance between each divided region and the vehicle, the speed of the vehicle, and the maximum acceleration of the vehicle stored in the memory unit. The surrounding environment recognition device according to claim 1.

3. The third calculation unit adds, in the division region from the division region where the time difference is less than the threshold to the division region within a predetermined range, an additional value corresponding to the probability of the division region where the time difference is less than the threshold and the distance from the division region to the risk level from the current time to the risk level from the unit time prior to the current time. The surrounding environment recognition device according to claim 1.

4. The first calculation unit calculates the probability based on the probability calculated at a time unit time prior to the current time and the predicted position obtained at the current time. The surrounding environment recognition device according to claim 1.

5. The first calculation unit calculates a first time based on the first time calculated at a time prior to the current time by the unit time and the predicted position acquired at the current time, if the predicted position is included in the divided region, and calculates a subtracted value as the first time at the current time by subtracting the unit time from the first time calculated at a time prior to the current time by the unit time. The surrounding environment recognition device according to claim 1.

6. The first calculation unit further calculates the time it takes for the moving body to pass through each of the divided regions based on the predicted position. The surrounding environment recognition device according to claim 1.

7. The movement prediction information includes a division value obtained by dividing the sum of the total length of the moving body and the length of one side of the divided region by the speed of the moving body. The first calculation unit calculates the passage time based on the passage time calculated at a time prior to the current time by the unit time and the division value obtained at the current time if the predicted position is included in the divided area, and maintains the passage time calculated at a time prior to the current time by the unit time if the predicted position is not included in the divided area. The surrounding environment recognition device according to claim 6.

8. The second calculation unit calculates the threshold value for each divided region based on the length of one side of the divided region, the sum of the total length of the vehicle stored in the storage unit, and the speed of the vehicle, and the passage time. The surrounding environment recognition device according to claim 6.

9. The first calculation unit generates a moving object map showing the probability and first time for each of the divided regions, The third calculation unit generates a risk map showing the risk level for each of the divided regions. The surrounding environment recognition device according to claim 1.

10. A computer executes An acquisition step of acquiring vehicle information including the speed of the vehicle, and predicted movement information including predicted position, which predicts the position of moving objects around the vehicle for each unit of time within a predetermined period, for each unit of time, Based on the predicted position, a first calculation step is performed to calculate, for each divided region obtained by dividing the area surrounding the vehicle into predetermined sizes, the probability that the moving body will reach the divided region after the unit time, and the first time the moving body will reach the divided region. A second calculation step of calculating the second time the vehicle arrives at each of the divided regions, The system includes a third calculation step, which calculates a risk level indicating the degree to which the vehicle and the moving object come into contact for each divided region, based on the first and second time points. In the third calculation step, in the divided region where the time difference between the first time and the second time is less than a threshold, an additional value corresponding to the probability of the divided region is added to the risk level from the unit time prior to the current time, and in the divided region where the additional value is not added, a predetermined value is subtracted from the risk level from the unit time prior to the current time. Methods for recognizing the surrounding environment.

Images