Autonomous driving system

Abstract

To provide an autonomous driving system capable of smoothly resuming autonomous driving with less user intervention.SOLUTION: The autonomous driving system includes a vehicle that has a target route for autonomous driving and acquires the location of the vehicle, azimuth, and measurement reliability. The vehicle includes an automatic driving restart determination unit that switches from a stopped state to automatic driving. The automatic driving restart determination unit is configured so as to, when the measurement reliability is equal to or greater than the first threshold, allow the vehicle to resume autonomous driving when the absolute value of the first location deviation between the target driving path and the location of the vehicle is equal to or less than a second threshold value set in advance, and the absolute value of the angular deviation between the azimuth of the target driving path and the azimuth of the vehicle is equal to or less than a third threshold value set in advance.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an automated driving system. [Background technology]

[0002] In recent years, the mobility sector has seen the introduction of cutting-edge technologies such as artificial intelligence (AI), as well as information and communication technologies such as big data and 5G (5th generation mobile communications system), and advances in these technologies have led to vigorous development of automated transportation for people and goods. This is expected to solve a variety of social issues, such as resolving the previously difficult last-mile problem, alleviating driver shortages in the logistics sector, and easing traffic congestion.

[0003] For example, in the case of vehicle operations within a factory, drivers have traditionally used trucks or towing carts to transport goods from one point to another. However, in order to improve factory utilization rates, it is necessary to keep the transportation within the factory running at all times. However, this requires solutions such as increasing the number of vehicles that can be towed or extending the transportation time slots, which increases equipment costs and labor costs due to the need for additional drivers. Therefore, to reduce these burdens, automated or unmanned transportation vehicles are required.

[0004] In autonomous driving technology, the technical levels applicable to general vehicles are defined by the Society of Automotive Engineers (SAE) in the United States. At autonomous driving levels 0 to 3, the driver is required to monitor the vehicle, while at autonomous driving levels 4 and above, the vehicle can continue to drive autonomously without the driver having to monitor it.

[0005] To achieve an autonomous driving system of Level 4 or higher, many vehicles will need to be equipped with a range of sensors to ensure safe driving, including a front camera to monitor the area in front of the vehicle, millimeter-wave radar sensors (MMWR) to monitor obstacles in front and behind the vehicle, light detection and ranging sensors (LiDAR) to measure the distance to surrounding obstacles using point cloud information, and ultrasonic sensors to measure the distance to obstacles.

[0006] However, even with such a sensor cluster, blind spots still exist due to the sensor's limited field of view (FoV). Furthermore, installing a sensor cluster on a vehicle increases the cost of the sensors. Furthermore, as the number of autonomous vehicles increases, if there is no prioritization of the vehicles, it becomes impossible to determine which autonomous vehicle should take priority, resulting in both vehicles stopping. To solve these problems, autonomous driving systems that arbitrate between autonomous vehicles are being developed by utilizing road-side units (RSUs), multi-access edge computers (MECs), 5G technology, and other technologies.

[0007] Furthermore, there are also satellite-based autonomous driving systems that use positioning information from artificial satellites. In these systems, vehicles receive positioning information on their position and heading angle from a satellite positioning system (GNSS: Global Navigation Satellite System), including quasi-zenith satellites, and travel along a pre-designated route, minimizing deviations between their own position and the route they are travelling on. The provision of such systems has made autonomous driving possible outdoors, and is enabling the efficient transportation of people and goods.

[0008] It is also an important issue to smoothly resume automatic driving of a stopped vehicle, and some methods for doing so have been disclosed (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Japanese Patent Publication No. 2021-163268 Summary of the Invention [Problem to be solved by the invention]

[0010] In Patent Document 1, the system determines whether to resume driving when the positional and angular deviations of the vehicle relative to the driving route are below a threshold. However, in the system disclosed in Patent Document 1, for example, in a use case where the main switch is turned from off to on, the information received from GNSS is reset, and the vehicle position and vehicle azimuth angle become undefined. Therefore, the vehicle position and vehicle azimuth angle must be measured again. The vehicle position can be measured by waiting for a certain period of time without moving the vehicle in an environment where signals from GNSS can be received. However, since the vehicle azimuth angle cannot be measured without moving the vehicle, resuming autonomous driving in this state carries the risk of driving in a direction that deviates from the driving route depending on the vehicle's attitude angle. Therefore, manual driving intervention is required until the vehicle azimuth angle can be measured, resulting in a system with a high level of user intervention.

[0011] The present disclosure discloses technology for solving the above-mentioned problems, and aims to provide an autonomous driving system that requires little user intervention and allows for smooth resumption of autonomous driving. [Means for solving the problem]

[0012] The autonomous driving system of the present disclosure includes a vehicle having a target driving route for autonomous driving, and acquires the position, azimuth angle, and measurement reliability of the vehicle, The vehicle is equipped with an automatic driving resumption determination unit that switches from a stopped state to automatic driving, When the measurement reliability is equal to or greater than a predetermined first threshold, an absolute value of a first position deviation between the target driving path and the position of the vehicle is equal to or less than a predetermined second threshold value; and When the absolute value of the angular deviation between the azimuth angle of the target driving route and the azimuth angle of the vehicle is equal to or less than a third threshold value set in advance, the resumption of automatic driving of the vehicle is permitted. [Effects of the Invention]

[0013] According to the present disclosure, it is possible to provide an autonomous driving system that requires little user intervention and allows for smooth resumption of autonomous driving. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a block diagram showing a configuration of an automatic driving system according to a first embodiment. [Figure 2A] FIG. 2 is a diagram illustrating an example of a hardware configuration of a multi-access edge computer (MEC) according to the first embodiment. [Figure 2B] FIG. 2 is a diagram illustrating an example of a hardware configuration of a roadside unit (RSU) according to the first embodiment. [Figure 2C] FIG. 1 is a diagram illustrating an example of a hardware configuration of a vehicle that performs automatic driving according to a first embodiment. [Figure 3] 2 is a functional block diagram showing a configuration of a roadside unit (RSU) according to the first embodiment. FIG. [Figure 4A] 1 is a functional block diagram showing the configuration of an autonomously driven vehicle according to a first embodiment. [Figure 4B] FIG. 2 is a functional block diagram showing the configuration of an automatic driving resumption determination unit of an AD module mounted on a vehicle. [Figure 5] FIG. 2 is a diagram showing transitions of vehicle states of an autonomously driven vehicle according to the first embodiment. [Figure 6A] 4 is a flowchart showing an operation for restarting autonomous driving of a vehicle in the autonomous driving system according to the first embodiment. [Figure 6B] 4 is a flowchart showing an operation for restarting autonomous driving of a vehicle in the autonomous driving system according to the first embodiment. [Figure 6C] 4 is a flowchart showing an operation for restarting autonomous driving of a vehicle in the autonomous driving system according to the first embodiment. [Figure 7] FIG. 2 is a functional block diagram showing the configuration of a vehicle azimuth angle determination unit included in a multi-access edge computer (MEC). [Figure 8A] FIG. 2 is a diagram showing an example of setting reliability in the autonomous driving system according to the first embodiment. [Figure 8B] FIG. 10 is a diagram showing another example of reliability setting in the autonomous driving system according to the first embodiment. [Figure 9] FIG. 10 is a diagram showing an example of a list of azimuth angles of stopped vehicles and their reliability levels stored in a multi-access edge computer (MEC). [Figure 10] 2 is a diagram for explaining the operation of restarting autonomous driving of a vehicle in the autonomous driving system according to the first embodiment. FIG. [Figure 11A] FIG. 10 is a diagram for explaining a vehicle travel route in the autonomous driving system according to the second embodiment. [Figure 11B] FIG. 10 is a diagram for explaining a vehicle travel route in the autonomous driving system according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0015] Hereinafter, an embodiment of an automated driving system according to the present disclosure will be described with reference to the drawings. Note that in each drawing, the same reference numerals indicate the same or corresponding parts. Therefore, detailed descriptions thereof may be omitted to avoid duplication.

[0016] Embodiment 1 The autonomous driving system according to the first embodiment will be described below with reference to the drawings. <Overall configuration of the autonomous driving system> Fig. 1 is a block diagram showing the configuration of an autonomous driving system according to embodiment 1. In Fig. 1, the autonomous driving system is an infrastructure-cooperative autonomous driving system including a roadside unit group 100 including roadside units (hereinafter referred to as RSUs) 110_1, . . . , 110_n, a multi-access edge computer (hereinafter referred to as MEC) 200, and vehicles 300A and 300B equipped with an autonomous driving (AD) module (hereinafter referred to as AD module). When referring to RSUs collectively, they are referred to as RSU 110. Note that, although the case where there are two vehicles performing autonomous driving will be described, the number of vehicles is not limited to two.

[0017] Communication between the RSU 110 and the MEC 200 is typically performed over a wireless network such as LTE (Long Term Evolution) or 5G, or some other dedicated network. Similarly, communication between the MEC 200 and the vehicles 300A and 300B is also performed over a high-speed, low-latency wireless network such as LTE or 5G, or some other dedicated network.

[0018] FIG. 2A illustrates an example of the hardware configuration of the MEC 200 according to the first embodiment. FIG. 2B illustrates an example of the hardware configuration of the RSU 110. FIG. 2C illustrates an example of the hardware configuration of the autonomously driven vehicle 300 according to the first embodiment. The MEC 200, the RSU 110, and the vehicle 300 each include an arithmetic processing circuit 1001, a storage device 1002 including a read-only memory (ROM) storing a program for executing the functions of each functional unit and a random access memory (RAM) for storing data representing the execution results of each functional unit, which are the results of calculations performed by the program, and an input / output circuit 1003. The arithmetic processing circuit 1001 includes a processor configured using a central processing unit (CPU). This processor may be configured using a digital signal processor (DSP) or a logic circuit. Dedicated hardware may be applied to the arithmetic processing circuit 1001. When the arithmetic processing circuit 1001 is dedicated hardware, the arithmetic processing circuit 1001 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.

[0019] In FIG. 2A , the detection results of the RSU 110 and the like are input to the input / output circuit 1003 of the MEC 200, and the MEC 200 exchanges data with the vehicle 300 via the input / output circuit 1003. 2B, in the RSU 110, detection results from various sensors, which will be described later, an image sensor 112, a radio wave sensor 113, and an optical sensor 114 are input to an input / output circuit 1003, and calculation results and the like are output to the MEC 200. In FIG. 2C, detection results of each detection unit described later, namely the host vehicle position detection unit 301, the obstacle sensor 303, and the IMU (Inertial Measurement Unit) sensor 304, and information from the MEC 200 are input to the vehicle 300 via the input / output circuit 1003. The calculation result is output to an Electric Power Steering (hereinafter referred to as EPS) 305, an actuator 306, and a brake 307 described later, and autonomous driving is executed.

[0020] <Configuration of RSU110> FIG. 3 shows the configuration of a roadside unit group 100 including a plurality of RSU110_1, 110_2, ···, 110_n, and is a functional block diagram showing the configuration of each RSU110. Each RSU110 is equipped with an image sensor 112 for detecting obstacles, a radio wave sensor 113, and an optical sensor 114. A CPU and software for executing a detection operation are stored in each sensor, and a memory for holding sensor data is also included. Each RSU110 detects obstacle information and a driving road surface within a region of a certain field of view.

[0021] In the image sensor 112, for example, as represented by a front monitoring camera, an obstacle is imaged, and the distance to the obstacle is calculated from the image data captured within a certain field of view. Further, the image sensor 112 acquires information on the driving road environment in which the vehicle 300 travels. As an example of the driving road environment, white line information on the driving road on which the vehicle 300 travels is acquired.

[0022] Also, in the radio wave sensor 113, for example, as represented by MMWR, in addition to the position information of an obstacle within a certain field of view, the speed of the obstacle within the field of view is directly calculated by the Doppler effect.

[0023] Furthermore, in the optical sensor 114, for example, as represented by LiDAR, an optical laser is irradiated within a certain field of view, and point cloud data obtained by the reflection of the laser from an obstacle is detected.

[0024] In the RSU 110, sensor signals output from the image sensor 112, radio wave sensor 113, and optical sensor 114 are input to a sensor information calculation unit 115, which processes the sensor signal outputs collectively. The sensor information calculation unit 115 performs filtering to remove noise information from the detection signals input from each sensor, and collectively processes information from multiple sensors, thereby enabling more reliable determination and identification of the presence or absence of obstacles in the surrounding environment where distances can be measured within the field of view of the RSU 110. For example, sensor fusion processing is performed to process information from different sensors to increase the reliability of the obstacle position accuracy, or obstacles are identified based on reinforcement learning such as deep learning.

[0025] Here, the processed data after filtering does not necessarily need to be processed in bulk and transmitted to the MEC 200. By processing all data from multiple RSUs 110_1, 110_2, ..., 110_n in bulk by the MEC 200, an environment can be created in which measurement signals that may be attenuated by multiple filters are also processed in bulk.

[0026] Through the above processing, the sensor information calculation unit 115 acquires obstacle information based on the multiple sensor information output from the RSUs 110, including location information such as the latitude and longitude of the obstacle, the obstacle's speed, the obstacle's height, and information identifying the obstacle as a vehicle, pedestrian, or motorcycle. The sensor information calculation unit 115 also transmits the point cloud information and image data acquired from the RSUs 110 to the MEC 200 via the communication module transmission unit 116. The sensor information calculation unit 115, which performs the above processing, is provided in all RSUs 110 and transmits the obstacle information calculated individually by each of the RSUs 110_1, 110_2, . . . , 110_n to the MEC 200. At this time, the measurement reliability when each sensor detected the information is also transmitted to the MEC 200. If any RSU 110 experiences an abnormal condition, the sensor information calculation unit 115 transmits abnormal condition information to the MEC 200.

[0027] Note that in the RSU 110, the method for detecting obstacles by sensors and the identification means are not limited to the above examples. In addition, the sensor information can also be combined with other sensors as needed, for example, other sensors that provide traffic flow detection and weather detection.

[0028] <Configuration of MEC 200> In FIG. 1, the functional blocks constituting the MEC 200 are shown. The MEC 200 receives obstacle information and the abnormal state information of each RSU 110 from a plurality of RSU 110s. In addition, the hardware configuration is as shown in FIG. 2A and includes an arithmetic processing circuit 1001 and a storage device 1002 that stores data and processing instructions.

[0029] The MEC 200 includes a communication module receiving unit 201 and a communication module transmitting unit 204 for high-speed communication with the RSU 110. The communication module receiving unit 201 receives obstacle information including the presence or absence of an obstacle, the position of the obstacle, and the detection time, and the measurement reliability from each RSU 110, and stores the obstacle information and the measurement reliability. In addition, the MEC 200 also receives information regarding the presence or absence of abnormalities in the sensors provided in each RSU 110 from the RSU 110. In addition, the communication module receiving unit 201 receives position information such as the latitude and longitude of the vehicle 300, the vehicle azimuth angle, and the measurement reliability from the vehicle 300.

[0030] The MEC 200 includes a dynamic map generation unit 202 that superimposes the received obstacle information on a digital map. The dynamic map generation unit 202 represents the obstacle information output from the communication module receiving unit 201 on the map. In the present embodiment, the vehicle position information of each of the vehicles 300A and 300B and the obstacle information acquired from the RSU !10 are superimposed on the dynamic map, and the dynamic map is output to the communication module transmitting unit 204.

[0031] Furthermore, the MEC200 defines an autonomous vehicle in an ignition-off state (hereinafter referred to as IG-OFF) as a stopped vehicle, and includes a vehicle azimuth determination unit 203 that detects the stopped vehicle and estimates the azimuth angle of the stopped vehicle. The vehicle azimuth determination unit 203 receives the point cloud information and image data detected by the roadside device group 100, and the obstacle information and measurement reliability obtained in the detection area, via the communication module receiving unit 201, and estimates the azimuth angle of the stopped vehicle by multiple means based on these. Also, a list (described later) of the vehicle azimuth angles and reliabilities estimated by each means is stored, and the vehicle azimuth determination unit 203 transmits the one with the highest reliability in the reliability list to the communication module transmitting unit 204.

[0032] <Configuration of Vehicle 300> FIG. 4A is a functional block diagram showing the configuration of a vehicle 300 that performs autonomous driving, and FIG. 4B is a functional block diagram showing the configuration of an autonomous driving restart determination unit of an AD module mounted on the vehicle 300. In FIG. 4A, the vehicle 300 includes a self-vehicle position detection unit 301 that detects the position of the vehicle 300, a communication module receiving unit 302 that receives a signal from the MEC200, an obstacle sensor 303 mounted on the vehicle that detects the distance to an obstacle, an IMU sensor 304 that detects the yaw rate of the vehicle 300, and the detected position information of the vehicle 300, the received signal from the MEC200, the detected distance to the obstacle, and the detected yaw rate of the vehicle 300 are input together with their respective measurement reliabilities, and an AD module 310 having an autonomous driving restart determination unit 3100 that determines whether to restart the autonomous driving of the vehicle, an EPS 305, an actuator 306, and a brake 307 controlled by the output of the AD module 310, and a communication module transmitting unit 308 for transmitting the calculation result in the AD module 310 to the MEC200.

[0033] <Configuration of AD Module 310> Using FIG. 4B, the details of the AD module 310 having an autonomous driving restart determination unit 3100 that determines whether to restart the autonomous driving of the vehicle will be described. The automatic driving resumption determination unit 3100 includes a driving state management unit 3111 that manages the driving state of the vehicle 300, an IG state detection unit 3112 that detects the on / off state of the IG switch, a destination detection unit 3121 that detects information about a target point to be reached by the vehicle 300, a target driving route planning unit 3122 that receives as input the destination information output from the destination detection unit 3121 and the host vehicle position output from the host vehicle position detection unit 301 and plans a target driving route along which the vehicle 300 will travel, and a vehicle position detection unit 3122 that detects the vehicle position of the vehicle 300 when the IG switch is off. the vehicle azimuth angle and measurement reliability; an estimated vehicle azimuth angle detection unit 3114 that detects the azimuth angle of a stopped vehicle estimated by MEC200 via the communication module receiving unit 302; an automatic driving judgment unit 3110 that receives signals from the driving state management unit 3111, the IG state detection unit 3112, the positioning state memory unit 3113, the estimated vehicle azimuth angle detection unit 3114 and the vehicle position detection unit 301 and allows the resumption of automatic driving; a driving start instruction detection unit 3123 that detects a driving start instruction sent from outside the vehicle by a control unit or an HMI (Human Machine Interface) or the like via the communication module receiving unit 302; and an automatic driving control unit 3120 that receives signals from the automatic driving judgment unit 3110, the target driving route planning unit 3122 and the driving start instruction detection unit 3123.

[0034] The automatic driving control unit 3120 calculates the target operation amount, target running amount, and target braking amount, and the EPS 305 controls the vehicle attitude angle so that it follows the target operation amount, the actuator 306 controls the vehicle speed so that it follows the target running amount, and the brake 307 controls the braking amount so that it follows the target braking amount.

[0035] The target steering amount may be, for example, a target steering angle, a target steering angle, or a target torque, and is calculated by the automatic driving control unit 3120. The EPS 305 controls the attitude angle of the vehicle by feedback control, feedforward control, or a combination of these so as to follow the target operation amount. The controller that controls the EPS 305 may be provided in the automatic driving control unit 3120 or in the EPS 305. An example of an EPS 305 equipped with a controller is an electric power steering with a steering angle sensor. By equipping a vehicle with an electric power steering controller that controls the rotation of this electric power steering with a steering angle sensor, a configuration can be realized that performs tracking control on a target steering angle, a target angular velocity, or their derivative values.

[0036] Furthermore, the target driving amount may be, for example, a target vehicle speed, a target acceleration, a target jerk, a target driving torque, etc., and is calculated by the automatic driving control unit 3120. The actuator 306 controls the vehicle speed by feedback control, feedforward control, or a combination of these so as to follow the target operation amount. The controller that controls the actuator 306 may be provided in the automatic driving control unit 3120 or in the actuator 306. As an example of an actuator 306 equipped with a controller, by equipping a vehicle with a configuration of a drive electric motor equipped with a rotation sensor and an inverter that controls the rotation of the drive motor, a configuration can be realized that performs follow-up control to a target longitudinal distance, a target vehicle speed, or their derivative values.

[0037] The target braking amount may be, for example, a brake pressure, which is calculated by the automatic driving control unit 3120. The brake 307 controls the braking amount by feedback control, feedforward control, or a combination of these so as to follow the target braking amount. The controller that controls the brake 307 may be provided in the automatic driving control unit 3120 or in the brake 307. As an example of the brake 307 equipped with a controller, a configuration for controlling the brake pressure to follow a target brake pressure can be realized by providing a vehicle with a hydraulic brake and a control unit for controlling the pressure of the hydraulic brake.

[0038] Furthermore, the AD module 310 of the vehicle 300 has a function of estimating the azimuth angle of stopped vehicles around the road on which the vehicle 300 is traveling, based on outputs from the communication module receiving unit 302, the obstacle sensor 303, the IMU sensor 304, and the vehicle position detection unit 301. The vehicle position detection unit 301 detects the position and azimuth of the vehicle using positioning information from, for example, the GNSS, and also acquires the reliability of the positioning.

[0039] The stopped vehicle detection unit 3131 detects stopped vehicles around the roadway based on the point cloud data and image data related to stopped vehicles received from the MEC 200 via the communication module receiving unit 302 and the distance to the stopped vehicle detected by the obstacle sensor 303. The stopped vehicle direction calculation unit 3134 estimates and calculates the direction (posture) of the stopped vehicle. The vehicle azimuth angle detection unit 3132 acquires the vehicle azimuth angle and the reliability of positioning from the vehicle position detection unit 301 . The first azimuth angle calculation unit 3135 of the stopped vehicle calculates the azimuth angle of the stopped vehicle based on the azimuth angle of the host vehicle output from the azimuth angle detection unit 3132 of the host vehicle and the reliability at the time of positioning. The calculated azimuth angle of the stopped vehicle is transmitted to the MEC 200 via the communication module transmission unit 308 together with the position and reliability of the stopped vehicle.

[0040] Furthermore, the yaw angle calculation unit 3133 of the host vehicle integrates the yaw rate of the host vehicle output from the IMU sensor 304 to calculate the attitude angle (azimuth angle) of the vehicle. Furthermore, the second azimuth angle calculation unit 3136 of the stopped vehicle calculates the position coordinate Σ(X, Y, Θ) of the vehicle 300, with an arbitrary coordinate as the origin, using the integrated value of the travel distance acquired from a travel distance sensor (not shown) equipped on the vehicle. The means for calculating the position coordinate Σ(X, Y, Θ) is a technique called wheel odometry, which can calculate relative position coordinates from arbitrary coordinates using wheel speed or encoder sensors on the tires of wheeled robots and automobiles. The origin of the position coordinate Σ is the vehicle position and azimuth angle estimated from the point when the positioning reliability falls below a certain threshold. Therefore, the position coordinate Σa(Xa, Ya, Θa) can be calculated by adding the position coordinate to the reliability information, absolute vehicle position information, and vehicle attitude angle output from the host vehicle position detection unit 301. This allows the position on the dynamic map detected by autonomous vehicles traveling near the stopped vehicle and the azimuth angle of the stopped vehicle to be calculated and transmitted to the MEC 200 via the communication module transmission unit 308.

[0041] <Vehicle state transition> 5 shows the transition of the vehicle state of the autonomously driven vehicle 300. This transition of the vehicle state indicates the vehicle state detected by the driving state management unit 3111. There are five vehicle state transitions: IG-OFF state 501, manual driving state 502, autonomous driving state 503, minimum risk maneuver state 504 (hereinafter referred to as MRM state 504), and emergency stop state 505. The driving state management unit 3111 detects these states and manages the driving state of the vehicle 300.

[0042] The IG-OFF state 501 is assumed to be a switch provided in the vehicle, such as an ignition switch or a main switch. The main switch has indications such as "OFF," "manual," and "automatic," and in the first embodiment, the driving management state is set according to the state of these switches. When the user switches the main switch from "OFF" to "manual," the state transitions to the manual driving state 502. Conversely, when the main switch is switched from "manual" to "OFF," the state transitions from the manual driving state 502 to the IG-OFF state 501.

[0043] When the user switches the main switch from "manual" to "auto," the state transitions to automatic driving state 503. In automatic driving state 503, if there is a target route output from target driving route planning unit 3122, a driving start instruction output from driving start instruction detection unit 3123, and automatic driving permission from automatic driving determination unit 3110, the vehicle is permitted to drive, and vehicle 300 can perform automatic driving.

[0044] An example of when the vehicle 300 cannot continue autonomous driving during autonomous driving is when the reliability of the positioning state drops for a certain period of time. The vehicle position detection unit 301 detects the vehicle's position and vehicle azimuth angle and stores them together with the positioning reliability. If the positioning reliability remains below a threshold for a certain period of time, the vehicle transitions from the autonomous driving state 503 to the MRM state 504, and the vehicle 300 stops. However, if the drop in the positioning reliability is temporary and the positioning reliability recovers to above the threshold within a certain period of time, the vehicle transitions again from the MRM state 504 to the autonomous driving state 503, and the vehicle 300 resumes autonomous driving.

[0045] The emergency stop state 505 is a state in which the vehicle 300 is unable to continue traveling due to a hardware abnormality in the vehicle or the emergency stop switch inside the vehicle being pressed. Also, if the positioning reliability drops for a certain period of time or more, it is determined that traveling cannot be continued, and the state transitions from the MRM state 504 to the emergency stop state 505. Furthermore, if a vehicle abnormality is detected during automatic traveling in the automatic traveling state 503, it is determined that automatic traveling cannot be resumed immediately, and the state transitions to the emergency stop state 505.

[0046] <Procedure for restarting automatic driving> Next, the operation of restarting autonomous driving of vehicle 300 in the autonomous driving system according to embodiment 1 will be described with reference to the flowcharts shown in FIGS. 6A, 6B, and 6C.

[0047] First, in step S001, it is determined whether the IG-OFF state 501 has been switched to the manual driving state 502 or the automatic driving state 503. If the determination result in step S001 is NO, it is determined that the manual driving state 502 or the automatic driving state 503 is continuing, and the process proceeds to step S002, where the vehicle position information is obtained from the vehicle position detection unit 301. In step S003, the vehicle azimuth angle information is acquired from the vehicle position detection unit 301, and in step S004, the positioning reliability information is acquired from the vehicle position detection unit 301.

[0048] In step S005, a decision is made to resume automatic driving based on the target driving route planned by the target driving route planning unit 3122 and the vehicle position information, azimuth angle information, and positioning reliability acquired from the vehicle position detection unit 301.

[0049] The decision to resume autonomous driving is made based on whether the following three criteria are met. (1) The first restart determination criterion is to determine whether the positioning reliability information is equal to or greater than a preset first threshold. If the positioning reliability information is equal to or greater than the first threshold, the first restart determination flag is output as 1. If the reliability is less than the first threshold, the first restart determination flag is output as 0 (step S0051).

[0050] (2) The second restart judgment criterion is to determine whether the absolute value of the positional deviation (lateral deviation) between the target driving path and the vehicle position is equal to or less than a preset second threshold. If the positional deviation is equal to or less than the second threshold, the second restart judgment criterion is met, and the second restart judgment flag is output as 1. If the positional deviation exceeds the second threshold, the second restart judgment criterion is not met, and the second restart judgment flag is output as 0 (step S0052).

[0051] (3) The third restart determination criterion is to determine whether the absolute value of the angle deviation between the azimuth angle of the target driving route and the azimuth angle of the vehicle is equal to or less than a preset third threshold. If the angle deviation is equal to or less than the third threshold, the third restart determination criterion is met, and the second restart determination flag is output as 1. If the angle deviation exceeds the third threshold, the second restart determination flag is output as 0 (step S0053).

[0052] If the first to third restart determination flags are all 1, the traveling restart permission flag is output as 1. On the other hand, if any one of the first to third restart determination flags is 0, the traveling restart permission flag is output as 0 (step S0054).

[0053] By providing such criteria for determining whether to resume autonomous driving, even if a vehicle has stopped due to unstable positioning conditions and a temporary drop in positioning reliability information, it can automatically resume driving by waiting for the positioning reliability information to recover, realizing a resumption decision function that reduces manual intervention in driving. This makes it possible to provide a highly reliable autonomous driving system that can further reduce the possibility of deviation from the driving route.

[0054] If the determination result in step S001 is YES, the process proceeds to step S101, where the current vehicle position information is obtained from the vehicle position detection unit 301. In step S102, the positioning status is read from the positioning status storage unit 3113. The vehicle position, vehicle azimuth angle, and positioning reliability information stored immediately before IG-OFF are stored in the positioning status storage unit 3113. In this sequence, the information saved immediately before IG-OFF is recorded as storage unit host vehicle position information, storage unit host vehicle azimuth angle information, and storage unit positioning reliability information, and this information is acquired in step S102.

[0055] In step S103, a position deviation α between the current vehicle position information acquired from vehicle position detection unit 301 and the stored vehicle position information read from positioning state storage unit 3113 is compared, and it is confirmed that the absolute value of the position deviation α is equal to or less than threshold value β. This makes it possible to confirm whether the vehicle position when vehicle 300 is IG-OFF has changed from the vehicle position when IG-ON. If the absolute value of the position deviation α is equal to or less than threshold value β, it is determined that the vehicle has not moved before and after the IG switch was turned ON / OFF, and the process proceeds to step S105. In step S105, the read storage unit vehicle position information is set as a first azimuth, and in step S106, the read storage unit positioning reliability information is acquired as a first reliability.

[0056] On the other hand, if the determination result in step S103 is NO, that is, if it is determined that the vehicle has moved before and after the IG switch has been turned ON / OFF, the process proceeds to step S104. In step S104, 0 is substituted as the first reliability. This is a process to prevent the storage unit host vehicle azimuth from being selected when selecting an azimuth in step S108, which will be described later.

[0057] In step S107, the vehicle 300 compares the azimuth angle of the stopped vehicle stored in the MEC 200 with the reliability. The MEC 200 stores a list of at least one azimuth angle of the stopped vehicle calculated based on the point cloud data or image data from the RSU 110. In step S107, the vehicle 300 requests the MEC 200 to compare the azimuth angle of the vehicle 300. From the multiple azimuth angles and the reliability corresponding to each azimuth angle stored in the MEC 200, the azimuth angle with the highest reliability is acquired as the second azimuth angle. Furthermore, if no azimuth angle is stored in the MEC 200, a reliability of 0 is acquired.

[0058] In step S108, the first azimuth angle read from the positioning state storage unit 3113 of the vehicle 300 in step S102 and the first reliability which is the reliability defined in steps S105 and S106 are compared with the second azimuth angle and the second reliability which is the reliability thereof acquired from the MEC200 in step S107. The azimuth angle with the highest reliability among these values is selected as the vehicle azimuth angle, and step S005 which is the automatic driving start determination step is executed. As described above, when the determination result in step S103 is NO, since the first reliability is set to 0 in step S104, the second azimuth angle will be selected in step S108. [[ID=I]]

[0059] After the azimuth angle is selected in step S108, the determination of starting automatic driving is made in step S005 described above. When passing through step S108, that is, when the state of the vehicle 300 is switched from the IG-OFF state in step S001, the following determination is made. In the first restart determination criterion, it is determined whether the reliability selected in step S108 is greater than or equal to the third threshold (step S0051). In the second restart determination criterion, it is determined whether the lateral position deviation between the vehicle position acquired in step S101 and the target driving route is less than or equal to the first threshold (step S0052). In the third restart determination criterion, it is determined whether the angular deviation between the azimuth angle of the driving route and the azimuth angle selected in step S108 is less than or equal to the second threshold (step S0053). Based on the first to third restart determination criteria, the determination of restarting automatic driving is completed (step S0054).

[0060] Note that when the first reliability acquired in step S106, that is, the positioning reliability information of the storage unit, is greater than or equal to the first threshold used in step S0051, step S108 may be skipped.

[0061] <Configuration of the vehicle azimuth angle determination unit 203 of the MEC200> FIG. 7 shows a functional block diagram of the vehicle azimuth angle determination unit 203 included in the MEC200. The vehicle azimuth angle determination unit 203 includes a point cloud detection unit 210 that detects point cloud information from each RSU110 via the communication module receiving unit 201, a first vehicle direction calculation unit 211 that calculates the direction of the stopped vehicle based on the detected point cloud information, a first azimuth angle calculation unit 212 that calculates the azimuth angle of a feature by comparing feature information present within the field of view of the RSU110 that transmits the point cloud information with a dynamic map, and a first vehicle azimuth angle calculation unit 213 that calculates the azimuth angle of the stopped vehicle using the calculation results of the first vehicle direction calculation unit 211 and the first azimuth angle calculation unit 212.

[0062] The first vehicle direction calculation unit 211 calculates the direction of the stopped vehicle, i.e., the vehicle attitude angle, from the detected point cloud information. Typically, vehicles are identified from point cloud information obtained by an optical sensor (e.g., LiDAR), and the presence or absence of reflection points allows for the distinction between the vehicle and other spaces. Therefore, by detecting the point cloud information, the point cloud information output from the RSU 110 is a vehicle, and its direction can also be determined. Specifically, the first vehicle direction calculation unit 211 estimates the first vehicle direction relative to the stopped vehicle based on the point cloud information output from the point cloud detection unit 210. One estimation method involves detecting vehicle feature points (e.g., vehicle edges and contours) from the point cloud data and estimating the vehicle attitude from the arrangement and relationship of these feature points. Another method involves clustering the point cloud data to separate the vehicle point clouds and analyzing these clusters to estimate the vehicle shape and attitude. Another method involves inputting point cloud data and estimating the vehicle attitude using a neural network suitable for point cloud processing.

[0063] The first azimuth angle calculation unit 212 identifies registered feature information from the point cloud data detected by the point cloud detection unit 210, and measures the orientation of the feature using at least one or more features. For example, there are stationary objects such as buildings and traffic lights, and these stationary objects are registered on a dynamic map, and the point cloud information is associated with the feature. In other words, the azimuth angle of the feature registered on the dynamic map can be used as the feature reference azimuth angle.

[0064] In the first vehicle azimuth angle calculation unit 213, the feature-based azimuth angle calculated by the first azimuth angle calculation unit 212 and the first vehicle direction estimated by the first vehicle direction calculation unit 211 are both on a dynamic map, and based on the relative angle between the first estimated vehicle direction 1 and the feature-based azimuth angle, the vehicle azimuth angle on the dynamic map is set as a feature-based estimated vehicle azimuth angle, and a feature-based reliability calculated from the reliability of the point cloud information is calculated.

[0065] Here, the feature-based estimated vehicle azimuth is calculated using point cloud information output from the RSU 110. The optical sensor of the RSU 110 has a map of detection accuracy according to the detection distance, and generally, detection accuracy deteriorates when detecting an obstacle that is farther away. For this reason, it has a correction term that lowers the feature-based reliability when detecting an obstacle that is farther away.

[0066] The vehicle azimuth angle determination unit 203 also includes an image detection unit 220 that detects image information from each RSU 110 via the communication module receiving unit 201, a second vehicle direction calculation unit 221 that calculates the direction of the stopped vehicle based on the detected image information, a second azimuth angle calculation unit 222 that calculates the azimuth angle of the road from the image information, and a second vehicle azimuth angle calculation unit 223 that calculates the azimuth angle of the stopped vehicle using the calculation results of the second vehicle direction calculation unit 221 and the second azimuth angle calculation unit 222.

[0067] The image detection unit 220 detects the image data captured by the RSU 110 and the reliability of each pixel, and outputs the detected data to a second vehicle direction calculation unit 221 and a second azimuth angle calculation unit 222 . The second vehicle direction calculation unit 221 has a means for estimating the direction of the stopped vehicle from image data, and estimates and outputs the second vehicle direction. Methods for estimating the posture of a stopped vehicle include a method of detecting the stopped vehicle from an image using an object detection algorithm such as YOLO (You Only Look Once) or SSD (Single Shot multibox Detector) and estimating the posture from the detection results, a method of detecting feature points (e.g., corners and edges) of the vehicle in the image and estimating the posture of the vehicle from these feature points, and a method of inputting image data and directly estimating the posture of the vehicle using a neural network. These methods use a convolutional neural network (CNN) or a pose estimation network.

[0068] The second azimuth angle calculation unit 222, which calculates the azimuth angle from the road, has a white line detection means that detects white lines that exist on the road, and is able to associate the direction of the white lines calculated by the white line detection means with the road on the dynamic map, and calculates a white line reference azimuth angle. Other methods for detecting white lines include a method that converts the image into grayscale, sets a threshold value to convert the white line pixels to white (255) and the rest of the background to black (0), and uses binarization processing to identify areas where the white line is bright; a method that uses an edge detection algorithm to detect edges in the image and extract the position of the white line; and a Hough transform method that detects straight lines from an edge image.

[0069] In the second vehicle azimuth angle calculation unit 223, the estimated second vehicle direction output by the second vehicle direction calculation unit 221 and the white line-reference azimuth angle output by the second azimuth angle calculation unit 222 are both on a dynamic map, and these relative angles can be calculated on the dynamic map. That is, the white line-reference estimated vehicle azimuth angle can be calculated by adding the estimated second vehicle direction to the white line-reference azimuth angle. At this time, the corresponding reliability is also calculated.

[0070] Here, the white line-based estimated vehicle azimuth calculated by the second azimuth angle calculation unit 222 is calculated based on image information output from the RSU 110. The image sensor included in the RSU 110 has a map of detection accuracy according to the noise content rate of sunlight and the like as disturbance information, and generally, detection accuracy deteriorates when detecting an obstacle that is farther away. For this reason, it has a correction term that lowers the feature-based reliability when an obstacle that is farther away is detected.

[0071] Note that, in the case where the first azimuth angle calculation unit 212 that calculates the feature-reference azimuth angle does not have a specific feature to be registered on the dynamic map and is therefore unable to calculate the feature-reference azimuth angle, the azimuth angle of the stopped vehicle may be calculated using the white line-reference azimuth angle calculated by the white line detection means as input to the first vehicle azimuth angle calculation unit 213. Similarly, on a road without white lines, the second vehicle azimuth angle calculation unit 223 may calculate the azimuth angle of the stopped vehicle using the feature-reference azimuth angle from the first azimuth angle calculation unit 212.

[0072] Furthermore, vehicle azimuth angle determination unit 203 includes vehicle azimuth angle selection unit 230. Vehicle azimuth angle selection unit 230 receives as input the azimuth angle of the stopped vehicle based on the estimated first vehicle direction output from first vehicle azimuth angle calculation unit 213 and its reliability, the azimuth angle of the stopped vehicle based on the estimated second vehicle direction output from second vehicle azimuth angle calculation unit 223 and its reliability, and the azimuth angle of the stopped vehicle estimated by each autonomous vehicle and its reliability. In other words, it includes a list of multiple vehicle azimuth angles for stopped vehicles and their associated reliability.

[0073] The vehicle azimuth angle selector 230 has a reliability list in which multiple vehicle azimuth angles are associated with stopped vehicles stored in the MEC 200, and selects and outputs the vehicle azimuth angle with the highest reliability from the list when a reference request (S106) is received from the autonomously driven vehicle via the communication module receiver 201. That is, based on the reliability list created by inputting the vehicle azimuth angles estimated from the first vehicle azimuth angle calculator 213 and the second vehicle azimuth angle calculator 223 and from other vehicles, it selects the azimuth angle with the highest reliability and outputs it to the communication module transmitter 204.

[0074] Here, the "reliability" will be explained using an example. (1) Positioning reliability of positioning information from GNSS The reliability of the positioning information is calculated using a Kalman filter. The latitude and longitude detected by GNSS are used as state variables to calculate the position error in real time. First, the area A of the two-dimensional error ellipse is calculated using the eigenvalues ​​of the covariance matrix P. The covariance matrix P is a matrix that represents the uncertainty of the error, and in the two-dimensional case is expressed by the following equation (1).

[0075]

number

[0076] The area A of the error ellipse is calculated using the following equation (2) in relation to the eigenvalues ​​(λ1, λ2) of the covariance matrix P: A=π*sprt(λ1*λ2) (2) The area A of this error ellipse indicates the spread of the error, with a larger area indicating lower reliability and a smaller area indicating higher reliability. In other words, if the error ellipse is small, the measurement result can be trusted to be close to the true value, but if the error ellipse is large, the reliability of the measurement result will be low.

[0077] Fig. 8A shows an example in which the reliability is set using the area A of this error ellipse. In Fig. 8A, A0 is the area of ​​the ellipse measured in advance, and the reliability can be calculated based on a numerical value normalized to this ellipse area.

[0078] (2) Reliability of azimuth angle The reliability of the azimuth angle can be calculated in the same way as the reliability of the latitude and longitude positioning information. That is, for the azimuth angle and its rate of change, the angular velocity, is used to calculate the area Ad of the error ellipse, with the angular velocity being a state variable. Using the area Ad0 of the error ellipse of the angular velocity measured in advance, Ad / Ad0 is calculated and associated with the reliability in the same way as in Figure 8A.

[0079] (3) Reliability of object detection using images (cameras) The confidence level of an image-based object detection can usually be obtained automatically when using an object recognition library or framework. For example, in the YOLO algorithm, each detection is associated with a confidence score that indicates the certainty that the object really exists. This is usually expressed in a range from 0 to 1, with a higher score indicating that the model is more confident in its detection. Figure 8B shows an example of setting the confidence level of an image-based object detection using the YOLO confidence score.

[0080] (4) Reliability of LiDAR measurement data The reliability of the position information (X, Y, Z) detected by LiDAR can also be calculated using, for example, a Kalman filter. In this case, a three-dimensional error ellipse is calculated to evaluate the reliability in three-dimensional space. A three-dimensional error ellipse indicates the uncertainty of the position information, taking into account errors along the X, Y, and Z coordinate axes. This allows the reliability of the position to be expressed in three dimensions. For example, the position coordinates (Xt, Yt, Zt) are used as state variables, a 3x3 covariance matrix is ​​defined, and the volume S of the error ellipse is calculated. S0 is the volume of the error ellipse measured in advance, and the reliability is assigned based on the normalized value S / S0 of this ellipse volume. The assignment to reliability is similar to that shown in Figure 8A.

[0081] The measurement reliability when acquiring information from other sensors not shown in the example is set in the same manner as above. It is necessary to make it possible to compare reliability obtained using different methods. This can be done by weighting the method and normalizing it. Furthermore, when the vehicle position and azimuth angle are calculated using multiple pieces of information, the reliability can be calculated by weighting or averaging the reliability of the information used in the calculation, or by using the value with the smaller reliability. Here, reliability such as the reliability of positioning information from GNSS, the reliability of object detection using images (camera), and the reliability of measurement data from LiDAR are explained as "measurement reliability" or simply "reliability."

[0082] 9 is a diagram showing an example of a list including the azimuth angles and reliability of stopped vehicles provided in the MEC 200. For example, when a reference request is received from an autonomous vehicle whose vehicle position is (x_a, y_a), the vehicle position information in the first to third rows of the list in FIG. 9 is information about the stopped vehicle (x_a, y_a), and the vehicle azimuth angle θ_a1, which is assigned the highest reliability of 5, is output to the communication module transmitter 204. When the vehicle position is (x_b, y_b), a selection is made from the information in the next two rows.

[0083] <Example of a stopped vehicle resuming autonomous driving> FIG. 10 is an explanatory diagram of resuming autonomous driving. Vehicles 300A and 300B in the figure have similar AD modules and travel along the target driving route indicated by the dotted line in the figure. At this time, vehicle 300A is stopped at the position indicated in the figure and is in the IG-OFF state. Vehicle 300 is present within the sensor fields of view of RSU 110_1 and RSU 110_2 (the triangular area indicated by the dashed line). RSU 110_1 and RSU 110_2 also detect feature 700 and white lines, respectively, and calculate their respective azimuth angles and reliability. Vehicle 300B is traveling along the target driving route, and when turning right at an intersection, vehicle 300B detects stopped vehicle 300A as an obstacle.

[0084] Vehicle 300B also includes the AD module shown in FIG. 4B, and when traveling vehicle 300B turns right at an intersection, it detects vehicle 300A with its engine off using obstacle sensor 303. A stopped vehicle direction calculation unit 3134 of vehicle 300B estimates the direction of the vehicle based on the obstacle information output from obstacle sensor 303. For example, an optical sensor provided in vehicle 300B can estimate the direction of vehicle 300A from reflection points from an obstacle. The direction of vehicle 300A may also be calculated using an image sensor mounted on vehicle 300B.

[0085] The host vehicle azimuth angle detection unit 3132 of the vehicle 300B acquires the azimuth angle of the vehicle 300B from the host vehicle position detection unit 301 of the vehicle 300B. The first azimuth angle calculation unit 3135 of the stopped vehicle calculates the azimuth angle and reliability of the stopped vehicle 300A based on the vehicle direction output from the stopped vehicle direction calculation unit 3134 and the host vehicle azimuth angle of the vehicle 300B, and outputs the calculated azimuth angle and reliability to the communication module transmission unit 308.

[0086] The host vehicle yaw angle calculation unit 3133 of the vehicle 300B calculates the yaw angle of the vehicle 300B obtained by integrating the yaw rate information output from the IMU sensor 304 of the vehicle 300B and the travel distance from an integrated travel distance sensor (not shown), and calculates the position coordinate Σ from an arbitrary point. Then, the host vehicle yaw angle calculation unit 3133 calculates the vehicle position and vehicle azimuth angle of the vehicle 300A on the dynamic map and outputs them to the communication module transmission unit 308. These operations are the same as those explained using FIG. 4B.

[0087] The MEC 200 calculates the vehicle position, azimuth angle, and reliability of the stopped vehicle 300B that is within the detection range of the RSU 110_1 and the RSU 110_2 from the point cloud information and image information received from the RSU 110_1 and the RSU 110_2. Furthermore, the vehicle 300B transmits the vehicle position, azimuth angle, and reliability of the stopped vehicle 300A that the vehicle 300B detected as an obstacle, and the MEC 200 creates a list of the azimuth angle and reliability corresponding to the vehicle position of the stopped vehicle as shown in FIG. When the stopped vehicle 300A resumes autonomous driving from the IG-OFF state, the process proceeds sequentially from step S101 to step S107 described above, and in step S107, the azimuth angle and reliability of the stopped vehicle 300A are checked against the MEC 200. The MEC 200 selects the azimuth angle with the highest reliability from the azimuth angles and reliability corresponding to the vehicle position of the stopped vehicle 300A using the vehicle azimuth angle selection unit 230, and transmits the selected azimuth angle to the vehicle 300A via the communication module transmission unit 204.

[0088] If the vehicle azimuth angle cannot be measured using the above method, i.e., if autonomous driving cannot be resumed based on the first to third resumption judgment criteria in step S005, it is also effective to update the GNSS positioning information by driving the vehicle a short distance. By driving a short distance, new position information can be obtained and a more accurate azimuth angle can be estimated. Only when the RSU and obstacle sensors installed in the vehicle monitor that there are no obstacles in the surrounding area, is autonomous driving of a short distance permitted and the vehicle azimuth angle measured.

[0089] As described above, according to the present embodiment, an autonomous driving system includes a vehicle having a target driving route for autonomous driving, and acquires the vehicle's position, azimuth angle, and measurement reliability, wherein the vehicle is equipped with an autonomous driving resumption determination unit that switches from a stopped state to autonomous driving, and is configured to permit the vehicle to resume autonomous driving when the measurement reliability is equal to or greater than a predetermined first threshold, the absolute value of a first position deviation between the target driving route and the vehicle position is equal to or less than a predetermined second threshold, and the absolute value of an angle deviation between the azimuth angle of the target driving route and the azimuth angle of the vehicle is equal to or less than a predetermined third threshold. With this configuration, when the autonomous vehicle transitions from a stopped state to a driving state, the autonomous driving system determines whether to resume autonomous driving using the vehicle's position, azimuth angle, and measurement reliability acquired within the autonomous driving system, thereby enabling smooth resumption of autonomous driving with little user intervention and based on the azimuth angle supported by highly reliable information.

[0090] The autonomous vehicle also includes a vehicle position detection unit that detects the vehicle's position and azimuth angle and acquires the measurement reliability, and a positioning status storage unit that stores the vehicle's position, azimuth angle, and reliability acquired by the vehicle position detection unit when autonomous driving ends. The vehicle's autonomous driving resumption determination unit calculates a second position deviation between the vehicle's position detected by the vehicle position detection unit before starting driving and the vehicle's position at the end of the previous autonomous driving session stored in the positioning status storage unit. If the second position deviation is equal to or less than a predetermined fourth threshold, the autonomous driving resumption determination unit determines whether to resume autonomous driving of the vehicle using the vehicle's azimuth angle and measurement reliability at the end of the previous autonomous driving session stored in the positioning status storage unit. This configuration makes it possible to determine whether to resume autonomous driving using information including the azimuth angle stored in the positioning status storage unit, even if the positioning information from the vehicle position detection unit becomes uncertain, for example, immediately after transitioning from a stopped state to a driving state. This enables a smooth resumption of autonomous driving with less user intervention using highly reliable vehicle position and azimuth angle information.

[0091] In addition, the autonomous driving system according to this embodiment further includes at least one RSU that detects obstacles within the field of view and the road surface on which the vehicle is traveling, and an MEC that outputs obstacle information to the vehicle, including the position, azimuth angle, and reliability of the obstacle, acquired from the RSU. The autonomous driving resumption determination unit of the vehicle is configured to compare the measurement reliability at the time of acquiring the vehicle's position detected by the vehicle position detection unit before starting driving with the reliability of the vehicle's position included in the obstacle information acquired from the MEC, and determine whether to resume autonomous driving of the vehicle using the vehicle's position and azimuth angle associated with the one with the greater reliability. Furthermore, the autonomous driving system according to this embodiment includes a plurality of autonomously driving vehicles, each of which is configured to have an obstacle sensor that detects obstacles around the vehicle, a stopped vehicle direction calculation unit that detects other vehicles that have stopped in relation to the obstacles detected by the obstacle sensor and estimates the direction of the stopped vehicle, and calculates the azimuth angle of the stopped vehicle using the azimuth angle of the vehicle and the direction of the stopped vehicle calculated by the stopped vehicle direction calculation unit, and transmits the azimuth angle to the MEC.

[0092] In this way, when an autonomous vehicle transitions from a stopped state to a driving state, it can use obstacle information acquired from the RSU and stopped vehicle information acquired from other vehicles, thereby obtaining an azimuth angle with higher measurement reliability and improving redundancy. Based on the azimuth angle with the highest reliability from these multiple combinations of vehicle position, azimuth angle, and reliability, autonomous driving can be resumed smoothly with less user intervention.

[0093] Embodiment 2 The autonomous driving system according to the second embodiment will be described below with reference to the drawings. 11A and 11B are diagrams for explaining a travel route of a vehicle in an autonomous driving system according to embodiment 2. Fig. 11A shows a view from above, and Fig. 11B shows a view from the side. In this embodiment 2, an example will be described in which the target travel route is not a virtual numerical value, and vehicle 300C travels by following electromagnetic induction wire 80 laid on the road surface.

[0094] 11A and 11B, the vehicle 300C is equipped with a guide sensor 309 that detects a fluctuating magnetic field from the electromagnetic induction line 80. It is desirable to install multiple guide sensors 309 on the bottom surface of the vehicle 300C. Furthermore, the vehicle 300C can estimate the vehicle's position and azimuth angle by being equipped with a dynamic map that includes the electromagnetic induction line. In this case, the reliability can be calculated, for example, from the noise level of the guide sensor 309. Furthermore, if the RSU 110 is installed along the electromagnetic induction line 80, information from the RSU can also be used as in the first embodiment.

[0095] As described above, according to the second embodiment, the same effects as those of the first embodiment can be achieved. Furthermore, since the vehicle travels on an electromagnetic induction line, the position and azimuth of the vehicle can be estimated without using a vehicle position detection unit. Alternatively, by using the vehicle position detection unit in combination with the first embodiment, the reliability of the vehicle position and azimuth can be improved.

[0096] Although the first and second embodiments have been described using vehicles as an example, the application of the autonomous driving system is not limited to automobiles, and it can be applied to various other mobile bodies. For example, it can be applied to an autonomous driving system including a mobile body that automatically travels along a target route, such as an in-building mobile robot that inspects the inside of a building, a production line inspection robot, and a personal mobility vehicle. In this case, the obstacle information etc. acquired by the RU100 can be information from an obstacle information detector installed, for example, inside a building, on a production line, or within the range in which the personal mobility vehicle operates.

[0097] While the present disclosure describes various exemplary embodiments and examples, the various features, aspects, and functions described in one or more embodiments are not limited to application to a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless variations not exemplified are conceivable within the scope of the technology disclosed in this specification, including, for example, cases where at least one component is modified, added, or omitted, and cases where at least one component is extracted and combined with components of another embodiment.

[0098] Various aspects of the present disclosure are summarized below as appendices.

[0099] (Appendix 1) An automated driving system including a vehicle having a target driving route that is driven automatically, and acquiring a position, an azimuth angle, and a measurement reliability of the vehicle, The vehicle is equipped with an automatic driving resumption determination unit that switches from a stopped state to automatic driving, When the measurement reliability is equal to or greater than a predetermined first threshold, an absolute value of a first position deviation between the target driving path and the position of the vehicle is equal to or less than a predetermined second threshold value; and An autonomous driving system that allows the vehicle to resume autonomous driving when the absolute value of the angular deviation between the azimuth angle of the target driving route and the azimuth angle of the vehicle is less than or equal to a preset third threshold value. (Appendix 2) The vehicle is a vehicle position detection unit that detects the position and azimuth angle of the vehicle and acquires the measurement reliability; a positioning state storage unit that stores the position, azimuth angle, and measurement reliability of the vehicle acquired by the vehicle position detection unit when the autonomous driving ends, The automatic driving resumption determination unit of the vehicle A second position deviation between the position of the vehicle detected by the vehicle position detection unit before starting driving and the position of the vehicle at the time of the end of the previous automatic driving stored in the positioning state storage unit is calculated, and if the second position deviation is equal to or less than a fourth threshold value set in advance, 2. The autonomous driving system according to claim 1, wherein the system determines whether to resume autonomous driving of the vehicle using the azimuth angle and measurement reliability of the vehicle at the time of the previous end of autonomous driving, which are stored in the positioning state memory unit. (Appendix 3) The automated driving system further includes at least one roadside device that detects obstacles within a field of view and a road surface that the vehicle is traveling on, and an edge computer that outputs obstacle information, including a position, an azimuth angle, and a measurement reliability of the obstacle, acquired from the roadside device to the vehicle; The automatic driving resumption determination unit of the vehicle An autonomous driving system as described in Appendix 2, which compares the measurement reliability of the vehicle's position detected by the vehicle position detection unit before starting driving with the measurement reliability of the vehicle's position included in the obstacle information obtained from the edge computer, and determines whether to resume autonomous driving of the vehicle using the vehicle's position and azimuth angle associated with the larger measurement reliability. (Appendix 4) the roadside device is equipped with an optical sensor, and the obstacle information includes reflection points from obstacles within a field of view as point cloud data; The edge computer includes a first vehicle direction calculation unit that calculates the direction of the stopped vehicle based on the obstacle information, a first azimuth angle calculation unit that calculates the azimuth angle of the feature based on reflection points from the feature fixed within the field of view among the obstacle information, and a first vehicle azimuth angle calculation unit that calculates the azimuth angle of the stopped vehicle and measurement reliability using the direction of the stopped vehicle calculated by the first vehicle direction calculation unit and the azimuth angle of the feature calculated by the first azimuth angle calculation unit. 1. The automated driving system of claim 3, (Appendix 5) 5. The autonomous driving system according to claim 4, wherein the measurement reliability of the obstacle information acquired from the roadside device decreases as the detection distance to the obstacle increases. (Appendix 6) the roadside device is equipped with an image sensor, and the obstacle information includes information about white lines on a road surface; The autonomous driving system described in Appendix 3, wherein the edge computer includes: a second vehicle direction calculation unit that calculates the direction of a stopped vehicle based on the obstacle information from the image sensor; a second azimuth angle calculation unit that calculates a white line reference azimuth angle, which is the direction of the road, from the white line information included in the obstacle information; and a second vehicle azimuth angle calculation unit that calculates the azimuth angle of the stopped vehicle and measurement reliability using the direction of the stopped vehicle calculated by the second vehicle direction calculation unit and the white line reference azimuth angle calculated by the second azimuth angle calculation unit. (Appendix 7) 7. The autonomous driving system according to claim 6, wherein the measurement reliability of the obstacle information acquired from the roadside device decreases as the detection distance to the obstacle increases. (Appendix 8) the measurement reliability is reduced when a noise content rate of the image acquired from the image sensor of the roadside device is equal to or greater than a preset noise threshold; (Appendix 9) A plurality of the vehicles are autonomously driven, The vehicle is Obstacle sensors that detect obstacles around the vehicle, a stopped vehicle direction calculation unit that detects another vehicle that has stopped in front of the obstacle detected by the obstacle sensor as a stopped vehicle and estimates the direction of the stopped vehicle; a first azimuth angle calculation unit for a stopped vehicle that calculates an azimuth angle of the stopped vehicle using the azimuth angle and reliability of the vehicle acquired by the vehicle position detection unit and the estimated direction of the stopped vehicle, An autonomous driving system according to any one of appendices 3 to 8, wherein the azimuth angle of the stopped vehicle calculated by the first azimuth angle calculation unit is transmitted to the edge computer together with its position and measurement reliability. (Appendix 10) A plurality of the vehicles are autonomously driven, The vehicle is Obstacle sensors that detect obstacles around the vehicle, a sensor for detecting a yaw rate of the vehicle; a stopped vehicle direction calculation unit that detects another vehicle that has stopped in front of the obstacle detected by the obstacle sensor as a stopped vehicle and estimates the direction of the stopped vehicle; a yaw angle calculation unit that calculates an azimuth angle of the vehicle from a yaw angle obtained by integrating the detected yaw rate; a second azimuth angle calculation unit for the stopped vehicle that calculates an azimuth angle of the stopped vehicle using the azimuth angle of the host vehicle calculated by the yaw angle calculation unit and the estimated direction of the stopped vehicle, An autonomous driving system according to any one of appendices 3 to 8, wherein the azimuth angle of the stopped vehicle calculated by the second azimuth angle calculation unit is transmitted to the edge computer together with its position and measurement reliability. (Appendix 11) The vehicle is When the measurement reliability of the vehicle position and azimuth angle detected by the vehicle position detection unit becomes equal to or less than a preset threshold value, The autonomous driving system described in Appendix 10, wherein the reliability at that time and the position of the vehicle calculated by accumulating the travel distance acquired by a travel distance sensor equipped on the vehicle are transmitted to the edge computer. (Appendix 12) The automatic driving resumption determination unit of the vehicle An autonomous driving system according to any one of appendices 9 to 11, which compares the magnitudes of the measurement reliability when the vehicle's position detected by the vehicle position detection unit before starting driving, the measurement reliability of the vehicle's position included in the obstacle information obtained from the edge computer, and the measurement reliability of the stopped vehicle obtained from the edge computer, and determines whether to resume autonomous driving of the vehicle using the vehicle's position and azimuth angle associated with the largest measurement reliability. (Appendix 13) The vehicle is A vehicle that follows and travels on an electromagnetic induction line laid on the road surface. An autonomous driving system according to any one of appendices 1 to 3, comprising a guide sensor that detects a fluctuating magnetic field from the electromagnetic induction line and map information including the electromagnetic induction line. [Explanation of symbols]

[0100] 80: electromagnetic induction line, 100: roadside unit group, 110, 110_1, 110_2, 110_n: RSU (roadside unit), 112: image sensor, 113: radio wave sensor, 114: optical sensor, 115: sensor information calculation unit, 116: communication module transmitter, 200: MEC (multi-access edge computer), 201: communication module receiver, 202: dynamic map generation unit, 203: vehicle azimuth angle determination unit, 204: communication module transmitter, 210: point cloud detection unit, 211: first vehicle direction calculation unit, 212: first azimuth angle calculation unit, 213: first vehicle azimuth angle calculation unit, 220: image detection unit, 221: second vehicle direction calculation unit, 222: second azimuth angle calculation unit, 223: second vehicle azimuth angle calculation unit, 230: Vehicle azimuth angle selection unit, 300, 300A, 300B, 300C: Vehicle, 301: Vehicle position detection unit, 302: Communication module reception unit, 303: Obstacle sensor, 304: IMU sensor, 305: EPS (electric power steering), 306: Actuator, 307: Brake, 308: Communication module transmission unit, 309: Guide sensor, 310: AD module (autonomous driving module), 3100: Autonomous driving resumption determination unit, 3110: Autonomous driving determination unit, 3111: Driving state management unit, 3112: IG state detection unit (ignition state detection unit), 3113: Positioning state storage unit, 3114: Estimated vehicle azimuth angle detection unit, 3120: Autonomous driving control unit, 3121: Destination detection unit, 3122: Target driving path planning unit, 3123: driving start instruction detection unit, 3131: stopped vehicle detection unit, 3132: azimuth angle detection unit for host vehicle, 3133: yaw angle calculation unit for host vehicle, 3134: direction calculation unit for stopped vehicle, 3135: first azimuth angle calculation unit for stopped vehicle, 3136: second azimuth angle calculation unit for stopped vehicle, 501: IG-OFF (ignition-off) state, 502: manual driving state, 503: automatic driving state, 504: MRM (minimum risk maneuver) state, 505: emergency stop state, 700: feature, 1001: calculation processing circuit, 1002: memory device, 1003: input / output circuit.

Claims

1. An autonomous driving system including a vehicle having a target driving route for autonomous driving, and acquiring a position, an azimuth angle, and a measurement reliability of the vehicle, The vehicle includes an automatic driving resumption determination unit that switches from a stopped state to automatic driving, When the measurement reliability is equal to or greater than a first threshold value set in advance, an absolute value of a first position deviation between the target driving route and the position of the vehicle is equal to or smaller than a second threshold value set in advance; and An autonomous driving system that permits the vehicle to resume autonomous driving when an absolute value of an angular deviation between an azimuth angle of the target driving route and an azimuth angle of the vehicle is equal to or less than a preset third threshold value.

2. The vehicle is a vehicle position detection unit that detects a position and an azimuth angle of a vehicle and obtains a measurement reliability; a positioning state storage unit that stores the vehicle position, the azimuth angle, and the measurement reliability acquired by the vehicle position detection unit when the autonomous driving ends, The automatic driving resumption determination unit of the vehicle, A second position deviation between the position of the vehicle detected by the vehicle position detection unit before starting driving and the position of the vehicle at the time of the end of the previous automatic driving stored in the positioning state storage unit is calculated, and if the second position deviation is equal to or smaller than a fourth threshold value set in advance, The autonomous driving system according to claim 1 , wherein the system determines whether to resume autonomous driving of the vehicle using the azimuth angle and measurement reliability of the vehicle at the time when the previous autonomous driving ended, which are stored in the positioning state storage unit.

3. The autonomous driving system further includes at least one roadside device that detects an obstacle within a field of view and a road surface that the vehicle is traveling on, and an edge computer that outputs obstacle information including a position, an azimuth angle, and a measurement reliability of the obstacle acquired from the roadside device to the vehicle, The automatic driving resumption determination unit of the vehicle, 3. The autonomous driving system according to claim 2, further comprising: a vehicle position detection unit that detects a vehicle position before the vehicle starts traveling; a vehicle position detection unit that detects a vehicle position before the vehicle starts traveling; a vehicle position detection unit that detects a vehicle position before the vehicle starts traveling; a vehicle position detection unit that detects a vehicle position before the vehicle starts traveling;

4. The roadside device is equipped with an optical sensor, and the obstacle information includes reflection points from obstacles within a field of view as point cloud data, The edge computer includes a first vehicle direction calculation unit that calculates a direction of a stopped vehicle based on the obstacle information, a first azimuth angle calculation unit that calculates an azimuth angle of a feature based on a reflection point from a feature fixed within the field of view among the obstacle information, and a first vehicle azimuth angle calculation unit that calculates an azimuth angle of the stopped vehicle and a measurement reliability using the direction of the stopped vehicle calculated by the first vehicle direction calculation unit and the azimuth angle of the feature calculated by the first azimuth angle calculation unit. The automated driving system according to claim 3 .

5. The autonomous driving system according to claim 4 , wherein the measurement reliability of the obstacle information acquired from the roadside device is decreased as the detection distance to the obstacle increases.

6. The roadside device includes an image sensor, and the obstacle information includes information about white lines on a road surface.

4. The autonomous driving system according to claim 3, wherein the edge computer comprises: a second vehicle direction calculation unit that calculates a direction of a stopped vehicle based on the obstacle information from the image sensor; a second azimuth angle calculation unit that calculates a white line reference azimuth angle, which is a road direction, from the white line information included in the obstacle information; and a second vehicle azimuth angle calculation unit that calculates an azimuth angle of the stopped vehicle and a measurement reliability using the direction of the stopped vehicle calculated by the second vehicle direction calculation unit and the white line reference azimuth angle calculated by the second azimuth angle calculation unit.

7. The autonomous driving system according to claim 6 , wherein the measurement reliability of the obstacle information acquired from the roadside device is decreased as the detection distance to the obstacle increases.

8. The autonomous driving system according to claim 7, wherein the measurement reliability is reduced when a noise content rate of the image acquired from the image sensor of the roadside device is equal to or greater than a preset noise threshold value.

9. A plurality of the vehicles are autonomously driven, The vehicle is An obstacle sensor that detects obstacles around the vehicle, a stopped vehicle direction calculation unit that detects another vehicle that has stopped in front of the obstacle detected by the obstacle sensor as a stopped vehicle and estimates a direction of the stopped vehicle; a first azimuth angle calculation unit for a stopped vehicle that calculates an azimuth angle of the stopped vehicle using the azimuth angle and reliability of the vehicle acquired by the vehicle position detection unit and the estimated direction of the stopped vehicle, The autonomous driving system according to claim 3 , wherein the azimuth angle of the stopped vehicle calculated by the first azimuth angle calculation unit is transmitted to the edge computer together with the position and measurement reliability.

10. A plurality of the vehicles are autonomously driven, The vehicle is An obstacle sensor that detects obstacles around the vehicle, a sensor for detecting a yaw rate of the vehicle; a stopped vehicle direction calculation unit that detects another vehicle that has stopped in front of the obstacle detected by the obstacle sensor as a stopped vehicle and estimates a direction of the stopped vehicle; a yaw angle calculation unit that calculates an azimuth angle of the vehicle from a yaw angle obtained by integrating the detected yaw rate; a second azimuth angle calculation unit for a stopped vehicle that calculates an azimuth angle of the stopped vehicle using the azimuth angle of the host vehicle calculated by the yaw angle calculation unit and the estimated direction of the stopped vehicle, The autonomous driving system according to claim 3 , wherein the azimuth angle of the stopped vehicle calculated by the second azimuth angle calculation unit is transmitted to the edge computer together with the position and measurement reliability.

11. The vehicle is When the reliability of the vehicle position and azimuth angle detected by the vehicle position detection unit becomes equal to or lower than a preset threshold value, The autonomous driving system according to claim 10, wherein the reliability at that time and the position of the vehicle calculated by integrating a travel distance acquired by a travel distance sensor equipped in the vehicle are transmitted to the edge computer.

12. The automatic driving resumption determination unit of the vehicle An autonomous driving system according to any one of claims 9 to 11, which compares the magnitudes of the measurement reliability at the time of acquiring the vehicle's position detected by the vehicle position detection unit before starting driving, the measurement reliability of the vehicle's position contained in the obstacle information acquired from the edge computer, and the measurement reliability of the stopped vehicle acquired from the edge computer, and determines whether to resume autonomous driving of the vehicle using the vehicle's position and azimuth angle associated with the largest measurement reliability.

13. The vehicle is A vehicle that travels along an electromagnetic induction line laid on the road surface. The autonomous driving system according to claim 1 , further comprising a guide sensor that detects a fluctuating magnetic field from the electromagnetic induction line and map information including the electromagnetic induction line.

Images