Driver assistance systems

Abstract

The aim is to provide a driver assistance system that enables automatic tuning when fusing data from roadside sensors, thereby enabling the generation of highly reliable dynamic maps. [Solution] The driving support system comprises a dynamic map generation unit that generates a dynamic map using roadside information from a roadside device, a reliability assignment unit that assigns reliability to the dynamic map generated by comparing roadside information and mobile information from a mobile object, and a dynamic map generation parameter search unit that searches for parameters of multiple sensors involved from the detection of an obstacle by the roadside device's sensor to the generation of the dynamic map, so that the reliability value becomes a preset expected value, and transmits the optimized parameters to the dynamic map generation unit.

Description

Technical Field

[0001] The present disclosure relates to a driving support system.

Background Art

[0002] In recent years, active research and development has been conducted on technologies for realizing autonomous driving. By using roadside information from sensors installed in roadside units and information from sensors mounted on moving objects, it is possible to assist in the autonomous driving of moving objects (see, for example, Patent Document 1). In the driving support system described in Patent Document 1, driving support is normally performed by preferentially using vehicle-to-roadside information, and when the detection accuracy of the roadside device is below a specified value, vehicle-to-vehicle information is fused with the vehicle-to-roadside information to complement it, and driving support is performed. This can suppress a decrease in the driving support level due to the detection accuracy of the roadside device.

[0003] In addition, a technique for expanding the detection range by fusing data from sensors of roadside devices and data from sensors mounted on vehicles and overlapping their detection ranges is also known (see, for example, Patent Document 2). In Patent Document 2, when fusing data from sensors of roadside devices and data from sensors mounted on vehicles, reliability coefficients are assigned to both data. The reliability coefficient includes sensor parameters, the detection distance of the detection target, and the detection angle of the detection target, and a value considering the calibration parameters of the sensor is used.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] In Patent Document 1, paragraphs 0032 to 0034 describe that the detection accuracy of roadside devices decreases under certain constraints. It then states that when the detection accuracy falls below a specified value, vehicle-to-vehicle information is used to supplement it. However, even in this case, maintaining positional accuracy is difficult because the system uses positional information of objects obtained from roadside devices. Furthermore, it is necessary to regularly maintain roadside devices to keep detection accuracy constant regardless of constraints.

[0006] Furthermore, while the technology described in Patent Document 2 assigns confidence factors to both the data from roadside device sensors and the data from vehicles, as indicated in the constraints of Patent Document 1, the accuracy of each sensor included in the confidence factor fluctuates depending on the time of day, weather, etc. Therefore, taking into account the effects of changes over time, roadside device sensors in particular require regular maintenance. In addition, since the number of roadside devices installed is enormous, there is a challenge in that performing maintenance on them is a heavy workload.

[0007] Furthermore, in recent years, dynamic maps have been generated using data from various sensors and used for driver assistance. However, generating dynamic maps involves fusing data from multiple sensors, requiring various tunings. The optimal tuning values ​​also change depending on the time of day, weather, and the passage of time. Therefore, tuning must be performed regularly in accordance with the maintenance of sensors and other equipment. If tuning is not performed, the accuracy of object detection by sensors will decrease, and the reliability of the dynamic map data will also decline.

[0008] This disclosure provides technology to solve the above-mentioned problems and aims to provide a driver assistance system that can automatically tune parameters when fusing multiple sensor data from roadside units. [Means for solving the problem]

[0009] The driver assistance system disclosed herein is A roadside unit equipped with multiple sensors to detect obstacles. A mobile device that acquires information about its own position and orientation, A roadside information receiving unit that receives the obstacle detected by the roadside unit as roadside information, A mobile body information receiving unit receives information on the position and orientation of the aforementioned mobile body as mobile body information, A dynamic map generation unit generates a dynamic map using the roadside information received by the roadside information receiving unit, A reliability assigning unit that compares the roadside information received by the roadside information receiving unit with the mobile information received by the mobile information receiving unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit, The edge server includes a dynamic map generation parameter search unit which searches for the parameters of the sensor related to the process from detecting an obstacle with the sensor to generating a dynamic map, so that the reliability value assigned by the reliability assignment unit becomes a preset expected value, and transmits the optimized parameters to the dynamic map generation unit, The mobile device is configured to receive the dynamic map generated by the edge server. [Effects of the Invention]

[0010] According to the driver assistance system disclosed herein, it becomes possible to automatically tune the parameters when fusing data from roadside sensors, thereby reducing the maintenance burden on sensors and enabling the generation of highly reliable dynamic maps. [Brief explanation of the drawing]

[0011] [Figure 1] This is a block diagram showing the overall configuration of the driver assistance system according to Embodiment 1. [Figure 2] This is a functional block diagram showing the configuration of the roadside unit according to Embodiment 1. [Figure 3] This is a functional block diagram showing the configuration of the mobile body according to Embodiment 1. [Figure 4] This is a functional block diagram showing the configuration of the edge server in the driver assistance system according to Embodiment 1. [Figure 5] This is a diagram for explaining the process up to dynamic map generation according to Embodiment 1. [Figure 6] This is a diagram showing examples of various parameters related to generating a dynamic map according to Embodiment 1. [Figure 7] This is a diagram showing an example in which a roadside unit detects a plurality of moving objects. [Figure 8A] This is a diagram showing an example of assigning a reliability level to a dynamic map according to Embodiment 1. [Figure 8B] This is a diagram showing an example of assigning a reliability level to a dynamic map according to Embodiment 1. [Figure 9] This is a flowchart for explaining the operation of a driving support system according to Embodiment 1. [Figure 10] This is a functional block diagram showing the configuration of an edge server in a driving support system according to Embodiment 2. [Figure 11] This is a flowchart for explaining the operation of a driving support system according to Embodiment 2. [Figure 12A] This is a diagram showing an example of assigning a reliability level to a dynamic map according to Embodiment 2. [Figure 12B] This is a diagram showing an example of assigning a reliability level to a dynamic map according to Embodiment 2. [Figure 13] This is a functional block diagram showing the configuration of an edge server in a driving support system according to Embodiment 3. [Figure 14] This is a functional block diagram showing the configuration of an optimization processing task management unit of an edge server according to Embodiment 3. [Figure 15A] This is a flowchart for explaining the operation of an optimization processing task management unit according to Embodiment 3. [Figure 15B] This is a flowchart for explaining the operation of an optimization processing task management unit according to Embodiment 3. [Figure 15C] This is a flowchart for explaining the operation of an optimization processing task management unit according to Embodiment 3. [Figure 16]This is a functional block diagram showing the configuration of the optimization processing task management unit of another edge server in the driver assistance system according to Embodiment 3. [Figure 17] This is a functional block diagram showing the configuration of the optimization processing task management unit of another edge server in the driver assistance system according to Embodiment 3. [Figure 18] This is a functional block diagram showing the configuration of the optimization processing task management unit of another edge server in the driver assistance system according to Embodiment 3. [Figure 19] Figure 17 is a flowchart illustrating the operation of the driver assistance system. [Figure 20] This figure shows an example of the hardware configuration of an edge server according to Embodiments 1 to 3. [Figure 21] This figure shows an example of the hardware configuration of a mobile body according to Embodiments 1 to 3. [Modes for carrying out the invention]

[0012] Hereinafter, embodiments of the driver assistance system disclosed in this specification will be described with reference to the figures. In each figure, the same reference numerals indicate the same or corresponding parts. Therefore, detailed descriptions of them may be omitted to avoid redundancy.

[0013] In autonomous driving technology, the level of technology applicable to general vehicles is defined by the Society of Automotive Engineers (SAE) in the United States. For autonomous driving levels 0 to 3, the driver is obligated to monitor the vehicle, while for autonomous driving levels 4 and above, the vehicle can continue to drive autonomously without driver monitoring. The following explanation will focus on autonomous driving levels 3 and 4.

[0014] Embodiment 1. The driver assistance system according to Embodiment 1 will be described below with reference to the diagrams. <Overall configuration of the driver assistance system> Figure 1 is a block diagram showing the overall configuration of the driver assistance system according to Embodiment 1, Figure 2 is a functional block diagram showing the configuration of the roadside unit, Figure 3 is a functional block diagram showing the schematic configuration of the mobile unit, and Figure 4 is a functional block diagram showing the configuration of the edge server in detail. In Figure 1, the driver assistance system 1 is an infrastructure-linked automated driving system including a roadside unit (referred to as RSU in the figure; RSU: Road Side Unit) 100, an edge server 200, and a mobile unit 300. The roadside unit 100 is intended to detect at least the mobile unit 300 within its detection area. Furthermore, the driver assistance system 1 may also include a control system 400 that receives information from the edge server 200 and centrally manages multiple edge servers 200 within its jurisdiction. Here, the mobile unit 300 is, for example, an autonomous vehicle such as an automobile. The configuration and functions of each are described below. Alternatively, instead of the control system 400, there may be an administrator or management device that manages roadside equipment including the roadside unit.

[0015] <Configuration of roadside unit 100> The driver assistance system 1 is equipped with multiple roadside units 100. As shown in Figure 2, each roadside unit 100 is equipped with an image sensor 112 for detecting obstacles, a radio wave sensor 113, an optical sensor 114, an ultrasonic sensor 115, etc. These sensors detect obstacle information and road surface information within a certain field of view or area. The detected information is transmitted as roadside information from the communication module 116 to the edge server 200. As described above, the roadside information includes at least obstacle information, and the obstacle information includes information on the moving object 300.

[0016] The image sensor 112, like a forward-facing camera, captures images of obstacles and calculates the distance to the obstacles from the image data captured within a certain field of view. Furthermore, the image sensor 112 acquires information about the road surface on which the mobile body 300 travels. Examples of road surface information include information about the white lines on the road on which the mobile body 300 travels, intersection information, and sidewalk information.

[0017] The radio wave sensor 113, for example, is a millimeter-wave radar sensor (MMWR), which acquires positional information of obstacles within a certain field of view and directly calculates the velocity of obstacles within the field of view using the Doppler effect.

[0018] The optical sensor 114, as exemplified by LiDAR (Light detection and ranging sensors), irradiates an optical laser within a certain field of view and detects point cloud data obtained from the reflection of the laser from obstacles.

[0019] The ultrasonic sensor 115 can detect the location and distance of an obstacle by emitting ultrasonic waves and detecting the time it takes for the reflected waves to arrive from surrounding objects. Furthermore, it may also be equipped with other sensors.

[0020] It is not necessary for each roadside unit 100 to be equipped with all of these sensors, but it is desirable to arrange multiple roadside units 100 so that information from the same area can be detected by multiple types of sensors. The roadside information transmitted from each roadside unit 100 to the edge server 200 may be sensed data, or it may be transmitted as detection data that recognizes obstacles, etc.

[0021] <Configuration of Mobile Unit 300> The driver assistance system 1 targets multiple mobile bodies 300. As shown in Figure 3, each mobile body 300 is configured as follows: a self-position detection unit 301 that detects the position of the mobile body 300; a communication module 302 that receives signals from the edge server 200; an IMU (Inertial measurement unit) sensor 303 that detects the yaw rate of the mobile body 300; an automated driving module (referred to as the AD module in the figure; AD: Automated Driving) 310 that receives the detected position information of the mobile body 300, the received signals from the edge server 200, and the detected yaw rate of the mobile body 300 to control automatic driving; an electric power steering (hereinafter referred to as EPS; EPS: Electric Power Steering) 305, actuator 306, and brake 307 controlled by the output of the automated driving module 310; and a communication module 308 for transmitting mobile body information, including the detected position information and the attitude of the mobile body 300, and the calculation results from the automated driving module 310 to the edge server 200.

[0022] The self-position detection unit 301 is, for example, a GNSS (Global Navigation Satellite System) sensor. A GNSS antenna is connected to the GNSS sensor, and the GNSS antenna receives positioning signals from positioning satellites orbiting in the satellite orbit. By analyzing the received positioning signals, the self-position can be obtained from the phase center information of the GNSS antenna (latitude, longitude, altitude, and azimuth, etc.). When transmitting mobile information from the mobile body 300, the reliability based on the self-position estimation accuracy may also be transmitted at the same time.

[0023] The autonomous driving module 310 controls the automatic driving of the mobile body 300 based on position information detected by the self-position detection unit 301, the attitude (direction of movement) of the mobile body detected by the IMU sensor, and dynamic map information acquired from the edge server 200. Specifically, it generates a target path, determines the driving speed, generates control commands for the EPS 305, each actuator 306, and brake 307, and controls the automatic driving.

[0024] <Configuration of Edge Server 200> Figure 4 is a functional block diagram showing the configuration of the edge server 200 according to Embodiment 1. In Figure 4, the edge server 200 is assumed to be a multi-access edge computer (MEC). The edge server 200 includes a roadside information receiving unit 201 that receives roadside information detected by the roadside unit 100, a mobile body information receiving unit 202 that receives mobile body information including the position and orientation of the mobile body detected by the mobile body 300, a dynamic map generation unit 203 that generates a dynamic map based on the roadside information received by the roadside information receiving unit 201, a dynamic map reliability assigning unit 204 that assigns reliability to the dynamic map using the roadside information received by the roadside information receiving unit 201 and the mobile body information received by the mobile body information receiving unit 202, and a dynamic map transmission unit 205 that transmits the generated dynamic map to the mobile body 300. Furthermore, the system includes an optimization start determination unit 206 that adjusts the detection parameters of the sensors mounted on the roadside unit 100 based on the reliability of the dynamic map assigned by the dynamic map reliability assignment unit 204 and the signal from the trigger input unit 209, and determines whether to start an optimization operation to improve the reliability of the dynamic map; a dynamic map generation parameter search unit 207 that searches for detection parameters and processing parameters for each sensor mounted on the roadside unit 100 to improve the reliability of the dynamic map when the optimization start determination unit 206 determines to start optimization; and an optimization information transmission unit 208 that outputs the optimized parameters and the information of the dynamic map updated by those parameters as optimization information.

[0025] A dynamic map is a map that includes not only geographically traversable routes and fixed features, but also superimposed information on mobile objects, obstacle information calculated by the edge server 200, and information that exists as rules but is not actually visible. Such dynamic maps are equipped with the latest road environments and are important for route generation for autonomous driving of mobile objects.

[0026] The dynamic map generation unit 203 fuses (fusions) roadside information, including obstacle and road surface information, acquired by the edge server 200 from multiple sensors of the roadside unit 100, and generates a dynamic map. The dynamic map confidence rating unit 204 assigns a confidence rating to the dynamic map generated by the dynamic map generation unit 203.

[0027] <Method for generating dynamic maps and assigning confidence levels to dynamic maps> Next, we will explain, using diagrams, how to generate dynamic maps and assign a confidence level R to them. Figure 5 illustrates the process of generating a dynamic map and assigning a confidence level R to the dynamic map based on roadside information detected by the roadside unit 100 and mobile body information, including the position and orientation of the mobile body detected by the mobile body 300. Here, we describe the case where the roadside information from the roadside unit 100 includes data acquired by at least LiDAR and a camera. Figure 6 shows examples of the parameters involved in generating the dynamic map, Figure 7 shows an example where the roadside unit 100 detects multiple vehicles, which are mobile bodies 300A and 300B, and Figures 8A and 8B show examples of assigning a confidence level to the dynamic map.

[0028] In Figure 5, the roadside information receiving unit 201 of the edge server 200 acquires point cloud data if the sensor mounted on the roadside unit 100 is a LiDAR, and acquires imaging data if the sensor is a camera, and transmits it to the dynamic map generation unit 203. The dynamic map generation unit 203 performs preprocessing on the data acquired from each type of sensor and detects obstacles. Next, the information of obstacles detected from the various sensors is merged. The dynamic map generated by reflecting the merged obstacle information is output to the mobile unit 300.

[0029] To detect a single obstacle from data from multiple sensors of different types, it is necessary to adjust the parameters (variables) related to the information from each sensor. For example, as shown in Figure 6, when acquiring roadside information from LiDAR, the detection accuracy is affected by spatial and temporal scan density, as well as signal strength. Furthermore, when preprocessing roadside information from LiDAR, the detection accuracy is affected by the algorithm used to remove the ground background from the point cloud, the settings of various thresholds used in that algorithm, and the settings of each filter and point cloud cluster.

[0030] When acquiring roadside information from a camera, detection accuracy is affected by camera calibration, image resolution, ISO sensitivity during imaging, and exposure time. Furthermore, when preprocessing roadside information from a camera, the detection accuracy is affected by factors such as the contrast correction value or color conversion correction value used to correct the acquired image, and the settings in the recognition algorithm used to detect obstacles from the image. Furthermore, the road surface information included in the roadside information is used for ground removal and contrast correction, etc.

[0031] When fusing obstacle information detected from roadside data obtained by LiDAR and obstacle information detected from roadside data obtained by cameras, adjustments are necessary, such as appropriately setting the weights assigned to each sensor that acquired information, the matching threshold for determining a single obstacle from data from different types of sensors, and the noise covariance value when fusing the data. When making adjustments, it is necessary to tune not only the settings of multiple parameters individually, but also the combination of multiple parameters. These parameter settings and tuning affect the detection accuracy and reliability. Algorithms such as Nearest Neighbor and Multiple Hypothesis Tracking may be used as methods for linking data of the same object from multiple types of sensor information. A threshold for whether or not to fuse the data may be set based on the detection accuracy detected by each sensor, such as the mean, standard deviation, and covariance of the errors.

[0032] Furthermore, when generating dynamic maps, detection accuracy is affected by parameters such as the threshold for whether to merge or individually detect multiple obstacles, the confidence threshold for each detection result of merged obstacles, and the amount of movement and rotation when transforming the coordinates of detected obstacles.

[0033] Thus, in order to generate a dynamic map, it is important to adjust each parameter at each stage, and by adjusting them, it is possible to identify and detect a single obstacle from multiple types of sensors. Note that while Figure 6 shows the parameters for LiDAR and camera sensors, the edge server 200 also holds parameter information for other sensors.

[0034] Next, we will explain the reliability R of the dynamic map generated from roadside information from the roadside unit 100's sensors. As shown in Figure 7, the sensors mounted on the roadside unit 100 detect the moving objects 300A and 300B as obstacles within the area X indicated by the dotted line. Here, examples of point cloud data and image data are shown. In addition, the moving object information, including the position and orientation of each vehicle acquired by the moving objects 300A and 300B, is transmitted to the moving object information receiving unit 202 of the edge server 200. In Figure 5, the position information and attitude information of the mobile bodies 300A and 300B received by the mobile body information receiving unit 202 are output to the dynamic map reliability assignment unit 204, where they are pre-processed and merged. The position information and attitude information of the mobile bodies 300A and 300B are compared with the detection results of the mobile bodies 300A and 300B as obstacles detected from the roadside information from the roadside unit 100's sensors, and a reliability R is assigned to the dynamic map.

[0035] Figures 8A and 8B show the detection results of the moving bodies 300A and 300B based on roadside information from the roadside unit 100's sensor, overlaid with the detection results from the moving bodies 300A and 300B themselves. In each figure, the dashed lines represent the detection results from the roadside unit 100's sensor, the solid lines represent the detection results from the moving bodies 300A and 300B themselves, and the arrows indicate the direction of movement.

[0036] Comparing Figure 8A and Figure 8B, it can be seen that Figure 8B shows a higher degree of agreement between the detection results of the moving objects 300A and 300B based on roadside information from the roadside unit 100's sensor and the detection results of the moving objects 300A and 300B themselves. In other words, Figure 8B is assigned a higher level of confidence. At least the positional accuracy and direction of movement of the moving objects are taken into consideration when assigning this level of confidence R.

[0037] Furthermore, the reliability score R is related to the type of sensor and the parameters of each process performed up to the dynamic map generation shown in Figures 5 and 6, and is characterized by being assigned a high value under the following conditions. (1) Data consistency If objects with the same characteristics are consistently output across multiple steps in generating the dynamic map, the reliability is high. (2) Propagation of sensor reliability Location information, primarily derived from point clouds, and type information, primarily derived from images, are considered highly reliable. (3) Propagation of algorithm confidence The results produced by rule-based algorithms are generally considered more reliable. (4) Infrequent correction and extrapolation The fewer times corrections and extrapolations are applied during the processing, the higher the reliability.

[0038] Furthermore, the position and attitude information of the mobile bodies 300A and 300B are compared with the detection results of the mobile bodies 300A and 300B as obstacles detected from the roadside information from the roadside unit 100's sensors, and a confidence level R is assigned to the dynamic map. Therefore, the accuracy of the estimation of the self-position information from the mobile bodies also affects the confidence level. For this reason, the accuracy of the estimation of the self-position information included in the mobile body information may be considered when assigning the confidence level R to the dynamic map.

[0039] If the confidence level R of the assigned dynamic map is equal to or greater than the preset threshold Thr1, the generated dynamic map is transmitted to the mobile device from the dynamic map transmission unit 205. At this time, the confidence level R of the assigned dynamic map may also be transmitted.

[0040] <Optimization of Parameters> In the dynamic map reliability assignment unit 204, the assigned reliability R is transmitted to the optimization start determination unit 206. When the assigned reliability R is smaller than a preset threshold Thr1, it is determined to start the optimization of the parameters of each process executed until the dynamic map is generated. At this time, the dynamic map generated by the dynamic map generation unit 203 is not transmitted to the moving object 300. In addition, since the parameters optimized previously may also vary over time, it is necessary to perform optimization periodically. Therefore, a signal from a timer may be input to the trigger input unit 209 to start the execution of parameter optimization. The trigger is not limited to periodic execution by a timer, and the execution of parameter optimization may be started by an external input signal such as an input signal from an operator. Note that the optimization of parameters is tuning for generating a dynamic map with high reliability.

[0041] For example, the detection results of the moving objects 300A and 300B based on the roadside information from the sensors of the roadside unit 100 are compared with the detection results by the moving objects 300A and 300B themselves. Assume that in FIG. 8A, the reliability Ra is assigned, in FIG. 8B, the reliability Rb (Rb>Ra) is assigned, and Ra < Thr1 ≤ Rb. In the case of FIG. 8B, the dynamic map is transmitted to the moving object 300, but in the case of FIG. 8A, it is determined to start the optimization of the parameters.

[0042] When it is determined by the optimization start determination unit 206 to start the optimization of the parameters, the dynamic map generation parameter search unit 207 varies the parameters of each sensor related to the generation of the dynamic map shown in FIG. 6, and searches for the values and combinations of the parameters so that the reliability R of the resulting dynamic map is at least equal to or higher than the threshold Thr1 and becomes the highest value or an expected value approaching it.

[0043] Here, there are multiple parameters, each with a range of variation in value, and many combinations are possible. Therefore, conditions that result in a predetermined expected reliability are extracted using methods such as parameter sweeping. The extracted conditions are used as optimization parameters, and dynamic map information updated with the optimized parameters is generated. Furthermore, the optimization method used to extract conditions where the confidence level reaches a predetermined expected value through parameter exploration is not limited to parameter sweeping. Optimization may also be performed using random variation exploration, exploration using mathematical models that mimic biological evolution or natural selection, exploration using machine learning models for inference, or a combination thereof.

[0044] If, as a result of parameter searching, the confidence level achieved by the optimized parameters reaches or exceeds the expected value, the dynamic map generation parameter search unit 207 transmits the updated dynamic map information and the updated confidence level (expected value) to the mobile body 300 via the dynamic map transmission unit 205. The values ​​of each optimized parameter are also transmitted to the dynamic map generation unit 203 and used in the dynamic map generation calculation in the next step (calculation cycle). Furthermore, the updated dynamic map information, updated confidence level, and optimized parameters are transmitted as optimization information from the optimization information transmission unit 208 to the control system 400.

[0045] If the parameter search is completed within a certain time or the confidence score R does not converge to the expected value, the dynamic map generation parameter search unit 207 terminates the parameter search. Then, the optimization information transmission unit 208 transmits to the control system 400 the situation where the confidence score R of the dynamic map is smaller than the threshold Thr1 and cannot be improved. Furthermore, the information transmitted to the control system may also include dynamic map information, confidence levels, and parameter values ​​prior to the start of parameter search, regardless of whether the situation is optimized or cannot be improved.

[0046] The control system 400 can plan and execute maintenance for the roadside equipment 100 based on information from the optimization information transmission unit 208.

[0047] <Operation of Driving Assistance System 1> The operation of the driver assistance system 1 will be explained using Figure 9. Figure 9 is a flowchart illustrating the operation of the driver assistance system according to Embodiment 1. First, in step S101, roadside information is acquired. Information including obstacle information and road surface information detected by each sensor of the roadside unit 100 is transmitted as roadside information and received by the roadside information receiving unit 201 of the edge server 200.

[0048] In step S102, mobile object information is acquired. Information including the position and orientation of the mobile object detected by the mobile object 300 is transmitted as mobile object information and received by the mobile object information receiving unit 202 of the edge server 200.

[0049] In step S103, the dynamic map generation unit 203 generates a dynamic map using roadside information. In step S104, the dynamic map reliability assignment unit 204 assigns a reliability level R to the dynamic map generated using roadside information and mobile object information.

[0050] In step S105, the optimization start determination unit 206 determines whether the assigned confidence level R is equal to or greater than the threshold Thr1. If the confidence level R is equal to or greater than the threshold Thr1, the dynamic map generated in step S103 is transmitted from the dynamic map transmission unit 205 to the mobile device 300 (step S106).

[0051] In step S105, if it is determined that the confidence level R is less than the threshold Thr1, in step S107, the dynamic map generation parameter search unit 207 optimizes the parameters related to each sensor until a dynamic map is generated. In step S107, if the parameters are optimized and the confidence level R converges to the expected value within a predetermined time t1, the dynamic map updated with the optimized parameters is transmitted to the mobile unit 300. The optimized parameters are also transmitted to the dynamic map generation unit 203 so that they can be used for generating the dynamic map in the next step. Furthermore, the optimized parameters and their combination, the updated confidence level, and the updated dynamic map are transmitted from the optimization information transmission unit 208 to the control system 400 (step S109).

[0052] In step S107, if the confidence level R does not converge within a predetermined time t1, the parameter search is terminated (No in step S108). Then, the optimization information transmission unit 208 transmits to the control system 400 the situation where the confidence level R of the dynamic map is smaller than the threshold Thr1 and cannot be improved (step S109). Steps S101 through S109 are executed repeatedly.

[0053] On the other hand, if a signal is input to the trigger input unit 209 due to a timer or operator input (step S111), in step S112, the dynamic map generation parameter search unit 207 optimizes the parameters related to each sensor until a dynamic map is generated. At this time, the parameter optimization is performed by referring to the dynamic map and its reliability generated in the previous calculation cycle.

[0054] In step S112, if the parameters are optimized and the confidence level R converges to the expected value within a predetermined time t1, the optimized parameters are transmitted to the dynamic map generation unit 203 so that they can be used in the generation of the dynamic map in the next step. Furthermore, the optimized parameters, their combination, and the updated confidence level are transmitted from the optimization information transmission unit 208 to the control system 400 (step S109).

[0055] In step S112, if the confidence level R does not converge within a predetermined time t1, the parameter search is terminated (No in step S113). Then, the optimization information transmission unit 208 transmits the situation that the confidence level R of the dynamic map cannot be improved to the control system 400 (step S109).

[0056] In step S106, the mobile body 300, having received the dynamic map via the communication module 302, uses that information to enable the automatic driving module 310 to control the EPS 305, actuator 306, brake 307, etc., thereby supporting automatic driving.

[0057] In step S109, the control system 400 plans and executes a maintenance plan for the roadside equipment 100 based on the information from the optimization information transmission unit 208. If the results indicate that the reliability R cannot be improved by optimizing the parameters, then actions such as accelerating the maintenance plan for roadside unit 100, conducting an immediate inspection, or analyzing the cause will be planned. Even if the reliability R can be improved through parameter optimization, preparations should be made to determine the optimal maintenance timing by tracking the changes in reliability R, the updated parameters, and their combinations.

[0058] In this embodiment 1, the operations of the roadside unit 100, edge server 200, and mobile unit 300 are synchronized at a reference time, for example, by communication from the control system 400. Therefore, synchronized information can be used for merging roadside information from multiple roadside units 100, merging information from multiple sensors of the roadside unit 100, and matching roadside information with mobile unit information for reliability assessment.

[0059] As described above, the driving assistance system of Embodiment 1 includes a roadside unit having multiple sensors to detect obstacles, a mobile unit acquiring its own position and orientation information, and an edge server. The edge server includes a roadside information receiving unit that receives obstacles detected by the roadside unit as roadside information, a mobile unit information receiving unit that receives position and orientation information of the mobile unit as mobile unit information, a dynamic map generation unit that generates a dynamic map using the roadside information, a confidence level assignment unit that compares the roadside information and the mobile unit information and assigns a confidence level to the dynamic map, and a dynamic map generation parameter search unit that searches for sensor parameters related to the period from obstacle detection by the sensors to dynamic map generation so that the confidence level assigned by the confidence level assignment unit becomes a preset expected value, and transmits the optimized parameters to the dynamic map generation unit. The mobile unit is configured to receive the dynamic map generated by the edge server. With this configuration, the dynamic map generation parameter search unit optimizes the parameters related to tuning during sensor data fusion so that the confidence level can be maintained at a constant value and provides them to the dynamic map generation unit. Therefore, the tuned parameters are used in the next dynamic map generation, enabling the generation of a highly reliable dynamic map. In other words, the edge server can automatically tune the parameters when fusing data from roadside sensors, reducing the maintenance burden on sensors and enabling the generation of highly reliable dynamic maps. This makes it possible to provide highly reliable dynamic maps to mobile vehicles, improving the reliability of driver assistance systems.

[0060] Furthermore, the driver assistance system of Embodiment 1 further includes a control system that centrally manages multiple edge servers. The edge servers transmit the dynamic map, the reliability, and the parameter values ​​to the control system along with the execution results before execution by the dynamic map generation parameter search unit. This allows the tuning status of the roadside machine parameters to be collected and reflected in the maintenance plan. As a result, the maintenance burden can be reduced and the efficiency of maintenance work can be improved.

[0061] Embodiment 2. The driver assistance system according to Embodiment 2 will be described below with reference to diagrams. The overall configuration of the driver assistance system is the same as that of the driver assistance system 1 in Figures 1 and 4 of Embodiment 1. The only difference from Embodiment 1 is the configuration of the edge server, so the explanation will focus on the differences.

[0062] <Configuration of Edge Server 200A> Figure 10 is a functional block diagram showing the configuration of the edge server 200 of the driver support system 1A according to Embodiment 2, and Figure 11 is a flowchart explaining the operation of the driver support system according to Embodiment 2. In Figure 10, the difference from Figure 4 of Embodiment 1 is that the edge server 200A is further equipped with a received information time series storage unit 210. As shown in Figures 10 and 11, the received information time-series storage unit 210 stores roadside information detected by the roadside device 100 received by the roadside information receiving unit 201 and mobile information received by the mobile information receiving unit 202 in a time-series (step S201), and transmits the time-series information to the dynamic map reliability assignment unit 204.

[0063] The dynamic map reliability assignment unit 204 compares the position information and attitude information of the moving object 300 included in the moving object information with the detection results of the moving object 300 as an obstacle included in the roadside information. In this embodiment 2, the comparison is performed using the time-series information stored in the received information time-series storage unit 210, and a reliability R is assigned to the dynamic map (step S104).

[0064] <Dynamic map confidence level R> Next, we will explain, using a diagram, how to assign a confidence level R to a dynamic map using time-series information. Figures 12A and 12B show the trajectory of the mobile body 300, which is a time-series arrangement of roadside information from the roadside unit 100's sensors, overlaid with the trajectory of the mobile body 300's own position information, also arranged in time. In each figure, the dashed line represents the trajectory based on roadside information from the roadside unit 100's sensors, and the solid line represents the trajectory based on the mobile body 300's own position information. The trajectory is, for example, the trajectory of the mobile body's center of gravity, but is not limited to the center of gravity as long as it is a point within the mobile body. Note that the trajectories shown in Figures 12A and 12B are also treated as dynamic map information.

[0065] In Figure 12A, the trajectory based on the information received from the mobile unit 300 and the trajectory based on the information received from the roadside unit 100 are similar trajectories. On the other hand, in Figure 12B, the trajectory based on the information received from the roadside unit 100 branches off midway and does not align with the trajectory based on the information received from the mobile body 300. This indicates that the obstacle detected by the roadside unit 100 cannot be recognized as the same mobile body 300, meaning that the information is linked to other objects. Furthermore, although discontinuous, some trajectories follow the trajectory based on the information received from the mobile body 300 from a certain point onward. Comparing Figure 12A and Figure 12B, we can see that Figure 12A shows a higher degree of trajectory agreement. In other words, Figure 12A is assigned a higher level of confidence.

[0066] As shown in the two examples in Figures 12A and 12B, the dynamic map reliability assignment unit 204 assigns a reliability R using the difference and continuity between the trajectory based on the information received from the mobile unit 300 and the trajectory based on the information received from the roadside unit 100. In this case, the confidence level R has the characteristic of being assigned a high value under the following conditions. (1) Proximity of distance: The closer the distance between trajectories, the higher the reliability. In other words, the more accurate the location information. (2) Similarity of shape: The smaller the difference in slope between each point in the trajectory, the higher the reliability. In other words, it is effective for matching data using moving objects traveling along curved paths. (3) Continuity: The greater the continuity of the trajectory, the higher the reliability. In other words, it is recognized as the same object and can be tracked for a long period of time.

[0067] Even when a confidence level R of the dynamic map is assigned using time-series information, in step S105 of Figure 11, it is determined whether the confidence level R is equal to or greater than a preset threshold Thr2. If the confidence level R is less than the threshold Thr2, the parameters of the roadside unit 100 sensor related to dynamic map generation are optimized, as described in Embodiment 1. The other steps are the same as in Figure 9 of Embodiment 1, so their explanation is omitted.

[0068] As described above, the driving support system of Embodiment 2 provides the same effects as Embodiment 1. Furthermore, the edge server further has a received information time-series storage unit that stores roadside information received by the roadside information receiving unit and mobile information received by the mobile information receiving unit in time series, and the reliability assignment unit compares the time-series roadside information and time-series mobile information stored in the received information time-series storage unit and assigns a reliability to the dynamic map generated by the dynamic map generation unit, thereby enabling the comparison of moving trajectories. This makes it possible to assign a highly accurate reliability by evaluating the difference and continuity of trajectories, especially in driving routes with curves.

[0069] Furthermore, even if the operations of the roadside unit, edge server, and mobile unit 300 are not synchronized with a reference time, as in Embodiment 1, the processes of assigning reliability, searching for parameters, and optimization can still be performed.

[0070] Embodiment 3. The following describes the driver assistance system according to Embodiment 3 with reference to diagrams. The overall configuration of the driver assistance system is the same as that of driver assistance system 1 in Figures 1 and 4. The only difference from Embodiments 1 and 2 is the configuration of the edge server, so the explanation will focus on these differences.

[0071] <Configuration of Edge Server 200B> Figure 13 is a functional block diagram showing the configuration of the edge server 200 of the driver assistance system 1B according to Embodiment 3. In Figure 13, the difference from Figure 10 of Embodiment 2 is that the edge server 200B is equipped with an optimization processing task management unit 220. Figure 14 is a diagram for explaining the configuration of the optimization processing task management unit 220, and adds dashed lines to Figure 13 to show each functional unit constituting the optimization processing task management unit 220 and the input / output signals to each functional unit.

[0072] <Configuration and operation of the optimization processing task management unit 220> The edge server 200B performs dynamic map generation processing to transmit dynamic maps to the mobile unit 300, while the optimization processing task management unit 220 is responsible for task management to perform optimization processing of the parameters related to the sensors of the roadside unit 100 in parallel with that processing. In other words, in addition to the second embodiment, a separate parameter optimization processing task is executed in the background.

[0073] The optimization processing task management unit 220 includes, as functional units, a verification dynamic map generation unit 221, a verification reliability assignment unit 222, a verification dynamic map generation parameter search unit 223, an optimization processing effectiveness determination unit 224, and an optimization result parameter update unit 225. The operation of each of these functional units will be explained using the flowcharts shown in Figures 15A, 15B, and 15C. Note that the parameter optimization processing task performed by the optimization processing task management unit 220 is executed in the background, and the main operation of the driver assistance system 1B, namely acquiring roadside information and mobile object information, generating a dynamic map, and transmitting the generated dynamic map and reliability to the mobile object and control system, is as shown in Figure 11. Therefore, in the following explanation, it will be assumed that the flowcharts in Figure 11 are executed in parallel, and the explanation will be omitted.

[0074] In Figure 15A, when a dynamic map is generated by the dynamic map generation unit 203 and a confidence level R is assigned by the dynamic map confidence level assignment unit 204, the optimization processing task management unit 220 acquires this information (step S301). The verification dynamic map generation parameter search unit 223 of the optimization processing task management unit 220 performs optimization of parameters related to multiple sensors until the dynamic map is generated (step S302). Parameter optimization is performed until the confidence level R converges or a preset time t3 has elapsed (step S303).

[0075] Next, in step S304, the optimization processing task is executed. In Figure 15B, first, the optimization processing task management unit 220 acquires the latest roadside information and time-series information of the roadside information and mobile object information (step S41). The verification dynamic map generation unit 221 of the optimization processing task management unit 220 generates a verification dynamic map using the latest roadside information acquired in step S41 and the optimized parameters and their combinations obtained in step S303 (step S42).

[0076] Next, the verification confidence assignment unit 222 assigns a verification confidence level Rv to the verification dynamic map using the time-series information acquired in step S41 (step S43). The dynamic map generation parameter search unit 223 of the optimization processing task management unit 220 performs optimization of the parameters related to each sensor until a dynamic map is generated (step S44). Parameter optimization is performed until the verification confidence level Rv converges or a preset time t3 has elapsed (step S45).

[0077] The operations from step S41 to step S45 are repeated until the number of times m is exceeded. Furthermore, from the second time onward, in the generation of the verification dynamic map in step S42, the verification map generation unit 21 uses the parameters and their combinations that were optimized in step S45 for the confidence level Rv. If the operations from step S41 to step S45 exceed a preset number of times m, the last obtained verification confidence score Rv, the parameters at that time, and their combinations are output to the optimization process effectiveness determination unit 224 (step S46).

[0078] Next, in step S305, a determination process is performed to determine whether or not the results of the optimization processing task in step S304 are valid. In Figure 15C, the optimization process effectiveness determination unit 224 obtains the confidence level R of the latest dynamic map from the dynamic map confidence level assignment unit 204 (step S51), and in step S46, compares it with the verification confidence level Rv obtained from the verification dynamic map generation parameter search unit 223 (step S52).

[0079] In step S52, the confidence level R and the verification confidence level Rv are compared, and it is determined whether the result of the optimization task in step S304 is valid, for example, by the following method. (1) If the verification confidence level Rv is greater than the confidence level R by a predetermined confidence improvement rate ΔR (Rv-R>ΔR), the result of step S304, which is the optimization processing task, is determined to be valid. (2) If the number of repetitions m in step S304 is small, and the validity determination in (1) above is performed consecutively for a predetermined number of times, the result of step S304, which is the optimization processing task, is determined to be valid. (2) If the number of repetitions m in step S304 is small, and the validity determination in (1) above occurs within a predetermined time period with a predetermined probability, the result of step S304, which is the optimization processing task, is determined to be valid. (4) Determine using any combination of (1) to (3) above.

[0080] In step S52, if it is determined that the result of the optimization processing task in step S304 is valid, the determination result, the verification confidence score Rv, and the parameters and their combinations at that time are output to the optimization result parameter update unit 225 (step S53).

[0081] In Figure 15A, if the optimization result parameter update unit 225 determines that the acquired determination result is valid, it transmits the verification confidence score Rv, the parameters at that time, and their combinations to the dynamic map generation unit 203, and the dynamic map generation unit 203 generates a dynamic map based on the updated parameters and their combinations (step S306).

[0082] The optimization result parameter update unit 225 may transmit the verification confidence score Rv, the parameters at that time, and their combinations to the dynamic map generation unit 203 when an authentication signal from the trigger input unit 209 is received by an operator or the like. Alternatively, the optimization result parameter update unit 225 may transmit the acquired judgment result, the verification confidence score Rv, the parameters at that time, and their combinations to the control system 400 via the optimization information transmission / reception unit 208B. If the judgment result is valid, permission may be obtained from the control system 400 to transmit the verification confidence score Rv, the parameters at that time, and their combinations to the dynamic map generation unit 203, and then transmitted to the dynamic map generation unit 203.

[0083] As described above, if the optimization processing effectiveness determination unit 224 determines that the result of step S304, which is the optimization processing task, is effective, then the dynamic map generation unit 203 generates a dynamic map using the parameters and their combinations at that time, thereby realizing a highly accurate dynamic map.

[0084] Figures 13 and 14 show an example in which the optimization processing task management unit 220 is added to the edge server 200A in Figure 11, and this example has been described. However, as shown in Figure 16, a configuration in which the optimization processing task management unit 220 is added to the edge server 200 in Figure 4 of Embodiment 1 is also possible.

[0085] Furthermore, in Figures 14 and 16, the edge servers 200B and 200C do not necessarily have to be equipped with an optimization start determination unit 206 and a dynamic map generation parameter search unit 207. Figures 17 and 18 are functional block diagrams showing the configuration of edge servers 200D and 200E in another driver assistance system according to Embodiment 3. Figures 17 and 18 show examples that do not include the optimization start determination unit 206 and the dynamic map generation parameter search unit 207, respectively, compared to Figures 14 and 16. Even in such examples, the optimization processing tasks of the optimization processing task management unit 220 described above optimize the parameters related to the sensors, and the optimized parameters and their combinations are reflected in the generation of the dynamic map.

[0086] Figure 19 is a flowchart illustrating the operation of the driver assistance system shown in Figure 17. The difference from Figure 11 is that instead of steps S105 and S107 to S113 in Figure 11, the processing in the optimization processing task management unit 220 from steps S301 to S306 is executed. After the processing in the optimization processing task management unit 220 is executed, the optimized parameters and combinations are sent to step S103, which generates a dynamic map, and are reflected therein. Furthermore, the operation of the driver assistance system corresponding to Figure 18 is the same as that of the system in Figure 18, except that it does not include step S201.

[0087] As described above, the driver assistance system of Embodiment 3 has the same effects as Embodiment 1. Furthermore, the driver assistance system of Embodiment 3 includes an edge server having multiple sensors, a roadside unit that detects obstacles, a mobile unit that acquires information on its own position and orientation, a roadside information receiving unit that receives obstacles detected by the roadside unit as roadside information, a mobile unit information receiving unit that receives information on the position and orientation of the mobile unit as mobile unit information, a dynamic map generation unit that generates a dynamic map using the roadside information, a reliability assignment unit that compares the roadside information and the mobile unit information and assigns a reliability level to the dynamic map generated by the dynamic map generation unit, and an optimization processing task management unit. The optimization processing task management unit includes: a dynamic map generation parameter search unit that searches for sensor parameters related to the process from obstacle detection by the sensor to dynamic map generation, and optimizes the parameters so that the reliability value assigned by the reliability assignment unit becomes a preset expected value; a verification dynamic map generation unit that generates a verification dynamic map using the parameters optimized by the dynamic map generation parameter search unit and the latest roadside information received by the roadside information receiving unit; a verification reliability assignment unit that compares the latest roadside information and the latest mobile object information and assigns a verification reliability to the verification dynamic map generated by the verification dynamic map generation unit; an optimization processing effectiveness determination unit that compares the reliability assigned to the latest dynamic map generated by the dynamic map generation unit by comparing the latest roadside information and the mobile object information with the verification reliability assigned by the verification reliability assignment unit and determines whether the parameter optimization in the dynamic map generation parameter search unit is effective; and an optimization result parameter update unit that, if the determination result of the optimization processing effectiveness determination unit is effective, transmits the optimized parameters to the dynamic map generation unit. The mobile object also receives the dynamic map generated by the edge server. By running parameter optimization as a separate block repeatedly in the background, overall system controllability is improved, and the reliability of the dynamic map is further enhanced.

[0088] Furthermore, the edge server includes a time-series received information storage unit that stores roadside information received by the roadside information receiving unit and mobile information received by the mobile information receiving unit in time series. The reliability assignment unit compares the time-series roadside information and time-series mobile information stored in the time-series received information storage unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit. The verification reliability assignment unit compares the time-series roadside information and time-series mobile information stored in the time-series received information storage unit and assigns a verification reliability level to the verification dynamic map generated by the verification dynamic map generation unit. Thus, the same effects as in Embodiment 2 are further achieved.

[0089] Figure 20 shows an example of the hardware configuration of the edge servers 200, 200A, 200B, and 200C in the embodiments 1 to 3 described above, which include an arithmetic processing circuit 1001, a storage device 1002, and a communication circuit 1003. Furthermore, Figure 21 shows an example of the hardware configuration of the mobile units 300, 300A, and 300B in the embodiments 1 to 3 described above, which include an arithmetic processing circuit 1001, a storage device 1002, a communication circuit 1003, and an input / output circuit 1004. Although not shown in the diagram, the memory device includes auxiliary memory such as ROM (Read Only Memory) which stores programs that execute the functions of each functional unit, and RAM (Random Access Memory) which stores data representing the execution results of each functional unit, which are the calculation results of the programs. It may also include auxiliary memory such as a hard disk. The arithmetic processing circuit 1001 executes the program input from the memory device 1002. In this case, the program is input to the arithmetic processing circuit 1001 from the auxiliary memory device via a volatile memory device. The arithmetic processing circuit 1001 may also output data such as calculation results to the volatile memory device of the memory device 1002, or it may save the data to the auxiliary memory device via the volatile memory device.

[0090] The communication circuit 1003 can use a communication module compliant with, for example, LTE (Long Term Evolution), 4G (4th Generation), or 5G (5th Generation). Indoors, wireless or wired LAN (Local Area Network), Wi-Fi (registered trademark), and Bluetooth (registered trademark) can be used.

[0091] The arithmetic processing circuit 1001 may be a processor such as a CPU (Central Processing Unit) or a DSP (Digital Signal Processor). Alternatively, dedicated hardware may be used for the arithmetic processing circuit 1001. If the arithmetic processing circuit 1001 is dedicated hardware, it may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0092] Furthermore, each component of the edge servers 200, 200A, 200B, and 200C, and the functional units they possess, can realize the aforementioned functions through hardware, software, or a combination thereof. For example, the arithmetic processing circuits of the edge servers implement their functions as dedicated hardware, while other functions are implemented through software. Similarly, in the mobile units 300, 300A, and 300B, each component and the functional parts they possess can realize their respective functions through a dedicated hardware arithmetic processing circuit, while other functions are realized through software. Thus, the aforementioned functions can be realized through hardware, software, or a combination thereof.

[0093] Furthermore, the mobile units 300, 300A, and 300B that constitute the driving support systems 1, 1A, and 1B in embodiments 1 to 3 are not limited to automobiles. For example, if the mobile unit is a transport vehicle or towing vehicle that travels within a factory, warehouse, or between warehouses, the roadside unit may be located within or near the warehouse. For example, the mobile unit may be a robot within a factory, and the roadside unit may be appropriately located within the factory.

[0094] While this disclosure describes various exemplary embodiments and examples, the various features, aspects, and functions described in one or more embodiments are not limited to the application of a particular embodiment, but are applicable individually or in various combinations to the embodiments. Accordingly, countless variations not illustrated are conceivable within the scope of the art disclosed in this specification. These include, for example, modifying, adding or omitting at least one component, or even extracting at least one component and combining it with components of other embodiments.

[0095] The various aspects of this disclosure are summarized below as an appendix.

[0096] (Note 1) A roadside unit equipped with multiple sensors to detect obstacles. A mobile device that acquires information about its own position and orientation, A roadside information receiving unit that receives the obstacle detected by the roadside unit as roadside information, A mobile body information receiving unit receives information on the position and orientation of the aforementioned mobile body as mobile body information, A dynamic map generation unit generates a dynamic map using the roadside information received by the roadside information receiving unit, A reliability assigning unit that compares the roadside information received by the roadside information receiving unit with the mobile information received by the mobile information receiving unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit, The edge server includes a dynamic map generation parameter search unit which searches for the parameters of the sensor related to the process from detecting an obstacle with the sensor to generating a dynamic map, so that the reliability value assigned by the reliability assignment unit becomes a preset expected value, and transmits the optimized parameters to the dynamic map generation unit, The mobile device is a driver assistance system that receives the dynamic map generated by the edge server. (Note 2) The edge server has an optimization start determination unit that determines whether to perform optimization of the parameters, The driver assistance system as described in Appendix 1, wherein the optimization start determination unit determines to perform the optimization of the parameters when the reliability value assigned by the reliability assignment unit is smaller than a preset threshold, or when a start trigger signal is input. (Note 3) The aforementioned trigger signal is a periodic signal from a timer or an external input signal including an operator, as described in Appendix 2 of the driving support system. (Note 4) The driving support system according to any one of the appendices 1 to 3, wherein the dynamic map generation parameter search unit terminates the optimization of the parameters if the reliability does not converge to a preset expected value within a preset time. (Note 5) The driving support system according to any one of the appendices 1 to 4, wherein the moving object information includes estimation accuracy for estimating its own position, and the reliability assigning unit assigns reliability using the transition accuracy of the self-position information of the moving object information. (Note 6) The driving assistance system according to any one of Appendix 1 to 5, wherein the dynamic map generation parameter search unit uses at least one of the following as an optimization method for extracting conditions under which the confidence level converges to a preset expected value: search by parameter sweep, search by random variation, search using a mathematical model that mimics biological evolution or natural selection, and search by inference using a machine learning model. (Note 7) The system further includes a control system that centrally manages multiple edge servers. The edge server is a driver assistance system according to any one of the appendices 1 to 6, which transmits the results performed by the dynamic map generation parameter search unit to the control system. (Note 8) The driver assistance system described in Appendix 7, wherein the edge server transmits the dynamic map, the confidence level, and the parameter values ​​to the control system along with the execution results before execution by the dynamic map generation parameter search unit. (Note 9) The edge server has a received information time-series storage unit that stores the roadside information received by the roadside information receiving unit and the mobile information received by the mobile information receiving unit in time series. The driving support system according to any one of the appendices 1 to 8, wherein the reliability assigning unit compares the time-series roadside information and the time-series mobile information stored in the received information time-series storage unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit. (Note 10) A roadside unit equipped with multiple sensors to detect obstacles. A mobile device that acquires information about its own position and orientation, A roadside information receiving unit that receives the obstacle detected by the roadside unit as roadside information, A mobile body information receiving unit receives information on the position and orientation of the aforementioned mobile body as mobile body information, A dynamic map generation unit generates a dynamic map using the roadside information received by the roadside information receiving unit, A reliability assigning unit that compares the roadside information received by the roadside information receiving unit with the mobile information received by the mobile information receiving unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit, An operation support system comprising an edge server having an optimization processing task management unit, The optimization processing task management unit, A dynamic map generation parameter search unit searches for the parameters of the sensor related to the process from obstacle detection by the sensor to dynamic map generation, and optimizes the parameters so that the reliability value assigned by the reliability assignment unit becomes a preset expected value. A verification dynamic map generation unit generates a verification dynamic map using the parameters optimized by the dynamic map generation parameter search unit and the latest roadside information received by the roadside information receiving unit. A verification reliability assigning unit that compares the latest roadside information and the latest mobile object information and assigns a verification reliability to the verification dynamic map generated by the verification dynamic map generation unit, An optimization process effectiveness determination unit compares the reliability assigned to the dynamic map generated by the dynamic map generation unit by comparing the latest roadside information and the latest mobile object information with the verification reliability assigned by the verification reliability assignment unit to determine whether the optimization of the parameters in the dynamic map generation parameter search unit is effective. The system includes an optimization result parameter update unit that, if the determination result of the optimization process effectiveness determination unit is valid, transmits the optimized parameters to the dynamic map generation unit. The mobile device is a driver assistance system that receives the dynamic map generated by the edge server. (Note 11) The driving support system as described in Appendix 10, wherein the dynamic map generation parameter search unit terminates the optimization of the parameters when a preset time has elapsed or the verification confidence level has converged to a preset expected value. (Note 12) The driving support system according to Appendix 10 or 11, wherein the moving object information includes an estimation accuracy for estimating its own position, the reliability assigning unit assigns the reliability using the transition accuracy of the self-position information of the moving object information, and the verification reliability assigning unit assigns the verification reliability using the transition accuracy of the self-position information of the moving object information. (Note 13) The driving support system according to any one of Appendix 10 to 12, wherein the dynamic map generation parameter search unit uses at least one of the following as an optimization method for extracting conditions under which the confidence level converges to a preset expected value: search by parameter sweep, search by random variation, search using a mathematical model that mimics biological evolution or natural selection, and search by inference using a machine learning model. (Note 14) The driver assistance system according to any one of the appendices 10 to 13, wherein the optimization processing effectiveness determination unit compares the latest roadside information and the latest moving object information to determine whether the optimization processing is effective by combining at least one of the following conditions: if the value of the verification reliability is greater than the reliability by a preset value; if the determination that the value of the verification reliability is greater than the reliability by a preset value occurs a preset number of times in a row; and if the determination that the value of the verification reliability is greater than the reliability by a preset value occurs within a preset time period with a preset probability. (Note 15) The system further includes a control system that centrally manages multiple edge servers. The edge server is a driver support system according to any one of the appendices 10 to 14, which transmits the result determined by the optimization processing effectiveness determination unit to the control system. (Note 16) The driver assistance system according to Appendix 15, wherein the edge server transmits the optimized parameters to the dynamic map generation unit after receiving a permission signal from the control system if the determination result of the optimization processing effectiveness determination unit is valid. (Note 17) The edge server has a received information time-series storage unit that stores the roadside information received by the roadside information receiving unit and the mobile information received by the mobile information receiving unit in time series. The reliability assignment unit compares the time-series roadside information and the time-series mobile information stored in the received information time-series storage unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit. The driver assistance system according to any one of claims 10 to 16, wherein the verification reliability assigning unit compares the time-series roadside information and the time-series mobile information stored in the received information time-series storage unit and assigns a verification reliability to the verification dynamic map generated by the verification dynamic map generation unit. [Explanation of Symbols]

[0097] 1,1A,1B: Driving assistance system, 100: Roadside unit, 112: Image sensor, 113: Radio wave sensor, 114: Optical sensor, 115: Ultrasonic sensor, 116: Communication module, 200,200A,200B,200C,200D,200E: Edge server, 201: Roadside information receiving unit, 202: Mobile object information receiving unit, 203: Dynamic map generation unit, 204: Dynamic map reliability assignment unit, 205: Dynamic map transmission unit, 206: Optimization start determination unit, 207: Dynamic map generation parameter search unit, 208: Optimization information transmission unit, 208B,208C,208D,208E: Optimization information transmission / reception unit, 209: Trigger input unit, 210: Received information time series storage unit, 220: Optimization processing task management unit, 221: Dynamic map generation unit for verification, 222: Confidence level assignment unit for verification, 223: Dynamic map generation parameter search unit for verification, 224: Optimization processing effectiveness determination unit, 225: Optimization result parameter update unit, 300, 300A, 300B: Moving body, 301: Self-position detection unit, 302: Communication module, 303: IMU sensor, 305: Electric power steering, 306: Actuator, 307: Brake, 308: Communication module, 310: Automated driving module, 400: Control system, 1001: Calculation processing circuit, 1002: Memory device, 1003: Communication circuit, 1004: Input / output circuit.

Claims

1. A roadside unit equipped with multiple sensors to detect obstacles. A mobile device that acquires information about its own position and orientation, A roadside information receiving unit that receives the obstacle detected by the roadside unit as roadside information, A mobile body information receiving unit receives information on the position and orientation of the aforementioned mobile body as mobile body information, A dynamic map generation unit generates a dynamic map using the roadside information received by the roadside information receiving unit, A reliability assigning unit that compares the roadside information received by the roadside information receiving unit with the mobile information received by the mobile information receiving unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit, The edge server includes a dynamic map generation parameter search unit which searches for the parameters of the sensor related to the process from detecting an obstacle with the sensor to generating a dynamic map, so that the reliability value assigned by the reliability assignment unit becomes a preset expected value, and transmits the optimized parameters to the dynamic map generation unit, The mobile device is a driver assistance system that receives the dynamic map generated by the edge server.

2. The edge server has an optimization start determination unit that determines whether to perform optimization of the parameters, The driver assistance system according to claim 1, wherein the optimization start determination unit determines to perform the optimization of the parameters when the reliability value assigned by the reliability assignment unit is smaller than a preset threshold, or when a start trigger signal is input.

3. The driving support system according to claim 2, wherein the trigger signal is a periodic signal from a timer or an external input signal including an operator.

4. The driving support system according to any one of claims 1 to 3, wherein the dynamic map generation parameter search unit terminates the optimization of the parameters if the reliability does not converge to a preset expected value within a preset time.

5. The driving support system according to any one of claims 1 to 3, wherein the moving object information includes an estimation accuracy for estimating its own position, and the reliability assigning unit assigns a reliability using the transition accuracy of the self-position information of the moving object information.

6. The driving support system according to any one of claims 1 to 3, wherein the dynamic map generation parameter search unit uses at least one of the following as an optimization method for extracting conditions under which the confidence level converges to a preset expected value: a search by parameter sweep, a search by random variation, a search using a mathematical model that mimics biological evolution or natural selection, and a search by inference using a machine learning model.

7. The system further includes a control system that centrally manages multiple edge servers, The driver assistance system according to any one of claims 1 to 3, wherein the edge server transmits the results performed by the dynamic map generation parameter search unit to the control system.

8. The driver assistance system according to claim 7, wherein the edge server transmits the dynamic map, the confidence level, and the parameter values ​​to the control system along with the execution results before execution by the dynamic map generation parameter search unit.

9. The edge server has a received information time-series storage unit that stores the roadside information received by the roadside information receiving unit and the mobile information received by the mobile information receiving unit in time series. The driving support system according to any one of claims 1 to 3, wherein the reliability assigning unit compares the time-series roadside information and the time-series mobile information stored in the received information time-series storage unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit.

10. A roadside unit equipped with multiple sensors to detect obstacles. A mobile device that acquires information about its own position and orientation, A roadside information receiving unit that receives the obstacle detected by the roadside unit as roadside information, A mobile body information receiving unit receives information on the position and orientation of the aforementioned mobile body as mobile body information, A dynamic map generation unit generates a dynamic map using the roadside information received by the roadside information receiving unit, A reliability assigning unit that compares the roadside information received by the roadside information receiving unit with the mobile information received by the mobile information receiving unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit, An operation support system comprising an edge server having an optimization processing task management unit, The aforementioned optimization processing task management unit, A dynamic map generation parameter search unit searches for the parameters of the sensor related to the process from obstacle detection by the sensor to dynamic map generation, and optimizes the parameters so that the reliability value assigned by the reliability assignment unit becomes a preset expected value. A verification dynamic map generation unit generates a verification dynamic map using the parameters optimized by the dynamic map generation parameter search unit and the latest roadside information received by the roadside information receiving unit. A verification reliability assigning unit that compares the latest roadside information and the latest mobile object information and assigns a verification reliability to the verification dynamic map generated by the verification dynamic map generation unit, An optimization process effectiveness determination unit compares the reliability assigned to the dynamic map generated by the dynamic map generation unit by comparing the latest roadside information and the latest mobile object information with the verification reliability assigned by the verification reliability assignment unit to determine whether the optimization of the parameters in the dynamic map generation parameter search unit is effective. The system includes an optimization result parameter update unit that, if the determination result of the optimization process effectiveness determination unit is valid, transmits the optimized parameters to the dynamic map generation unit. The mobile device is a driver assistance system that receives the dynamic map generated by the edge server.

11. The driving support system according to claim 10, wherein the dynamic map generation parameter search unit terminates the optimization of the parameters when a preset time has elapsed or the verification confidence level has converged to a preset expected value.

12. The driving support system according to claim 10 or 11, wherein the moving object information includes an estimation accuracy for estimating its own position, the reliability assigning unit assigns the reliability using the transition accuracy of the self-position information of the moving object information, and the verification reliability assigning unit assigns the verification reliability using the transition accuracy of the self-position information of the moving object information.

13. The driving support system according to claim 10 or 11, wherein the dynamic map generation parameter search unit uses at least one of the following as an optimization method for extracting conditions under which the confidence level converges to a preset expected value: a search by parameter sweep, a search by random variation, a search using a mathematical model that mimics biological evolution or natural selection, and a search by inference using a machine learning model.

14. The driver assistance system according to claim 10 or 11, wherein the optimization processing effectiveness determination unit compares the latest roadside information and the latest moving object information to determine whether the optimization processing is effective by combining at least one of the following conditions: if the value of the verification reliability is greater than the reliability by a preset value; if the determination that the value of the verification reliability is greater than the reliability by a preset value occurs a preset number of times in a row; and if the determination that the value of the verification reliability is greater than the reliability by a preset value occurs within a preset time period with a preset probability.

15. The system further includes a control system that centrally manages multiple edge servers, The driver assistance system according to claim 10 or 11, wherein the edge server transmits the result determined by the optimization processing effectiveness determination unit to the control system.

16. The driver assistance system according to claim 15, wherein the edge server transmits the optimized parameters to the dynamic map generation unit after receiving a permission signal from the control system if the determination result of the optimization processing effectiveness determination unit is valid.

17. The edge server has a received information time-series storage unit that stores the roadside information received by the roadside information receiving unit and the mobile information received by the mobile information receiving unit in time series. The reliability assignment unit compares the time-series roadside information and the time-series mobile information stored in the received information time-series storage unit and assigns a reliability level to the dynamic map generated by the dynamic map generation unit. The driver assistance system according to claim 10 or 11, wherein the verification reliability assigning unit compares the time-series roadside information and the time-series mobile information stored in the received information time-series storage unit and assigns a verification reliability to the verification dynamic map generated by the verification dynamic map generation unit.

Images