Emergency escape perception and decision system of unmanned vehicle based on vehicle-road cooperation

Abstract

The application discloses an emergency escape perception and decision system for unmanned vehicles based on vehicle-road cooperation, and relates to the technical field of unmanned driving.The application has the advantages of millimeter wave radar sensors, camera sensors and laser radar sensors;it is more accurate and timely to detect various emergency conditions on the road;the state of road participants is detected through sensors, and ultra-low delay broadcasting is performed on surrounding vehicles through 5G technology, which improves the environmental perception capability of the vehicle auxiliary driving system and future automatic driving vehicles, so that each vehicle can have a high-altitude view of the road and environment around the vehicle.Real-time output of all traffic participant key feature information can make up for the perception blind area of the intelligent automatic driving vehicle at the vehicle end, and complete traffic guidance of different objects.The solution has obvious advantages under the condition that a single vehicle sensor cannot perceive the environment due to serious environmental conditions such as large-curvature bends, high-speed heavy fog, rainy and snowy weather, etc.

Description

Technical Field

[0001] This invention belongs to the field of autonomous driving technology, and in particular relates to an emergency avoidance perception and decision-making system for autonomous vehicles based on vehicle-road cooperation. Background Technology

[0002] Autonomous vehicles or vehicles with driver assistance features will perceive the surrounding environment while driving to achieve driver assistance functions such as adaptive cruise control, automatic emergency braking, and lane departure assist.

[0003] Generally, obstacle perception in autonomous driving systems is achieved through machine vision and radar, with visual perception taking precedence over radar perception; that is, visual perception is primary, and radar perception is secondary. If the perception system misjudges an obstacle, other systems, including automatic emergency braking and automatic emergency steering, may not react. Radar perception is not suitable for detecting stationary obstacles during autonomous driving, and the recognition and judgment capabilities of cameras are severely reduced in rainy, snowy, or foggy weather.

[0004] For example, Chinese patent CN112631313B discloses a control method, device, and system for an unmanned driving device. When an anomaly is detected in the unmanned driving device, the anomaly type is obtained; the anomaly type is sent to a target roadside unit, which combines the anomaly type with environmental data near the unmanned driving device to generate control information. The target roadside unit is located near the driving range of the unmanned driving device. The system receives the control information returned by the target roadside unit and controls the unmanned driving device accordingly. Another example is Chinese patent CN113848956A, which discloses an unmanned vehicle system and method. Sensors within the unmanned vehicle module detect the vehicle's position, driving direction, and obstacle information in the surrounding environment in real time. The controller uses the information detected by the sensors to calculate control commands.

[0005] Research has found that fog patches on highways are one of the main causes of highway traffic accidents, especially chain-reaction collisions. Because they occur over short distances, often within a radius of only a few hundred meters, traditional highway weather monitoring systems are not easy to detect them, and even if they do, they cannot notify vehicles.

[0006] Therefore, an emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation is proposed. Summary of the Invention

[0007] The purpose of this invention is to provide an emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation. Through multiple fusion sensors on the roadside, it can accurately perceive the road conditions, the position, speed and direction of participating entities such as vehicles and pedestrians, identify some abnormal traffic events on the road, and implement vehicle-road perception technology.

[0008] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0009] This invention is an emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation, comprising:

[0010] The roadside perception module is used to collect environmental information and divide the environmental information into a primary viewpoint data group and a subsequent viewpoint data group. The roadside perception module includes a millimeter-wave radar sensor, a camera sensor, and a lidar sensor.

[0011] The edge data processing module analyzes environmental information, obtains the types of surrounding events, and acquires the distance parameters between obstacles and the current vehicle in the environmental information.

[0012] The vehicle-side decision execution module makes vehicle control decisions based on environmental information sent by the roadside perception module, combined with the vehicle's own dynamics and kinematics information.

[0013] Furthermore, it also includes a data communication module. Each roadside sensing module is equipped with a data communication module. The communication range of the data communication module covers three adjacent roadside sensing modules and satisfies R1-R2≥r, where R1 is the radius of the communication range coverage, R2 is the straight-line distance between two adjacent roadside sensing modules, and r is a preset distance representing the driver's reaction distance. The intersection of the edge of the communication range coverage area and the road is used as the travel point.

[0014] Furthermore, the roadside sensing module classifies the collected environmental information in the following ways:

[0015] S0: The millimeter-wave radar sensor, camera sensor, and lidar sensor all upload the collected environmental information to the processor every preset time T.

[0016] S1: Information grouping: The processor divides the environmental information received from the roadside perception module within the time period T into a group, removes redundant and erroneous information groups, and marks the remaining environmental information group as the view data group.

[0017] S2: If it occurs within the preset time T3 If there is a redundant information group, then the environmental information collected by the camera sensor is retained as the viewpoint data group; where [ξ] represents taking the integer part of ξ, This is the default value;

[0018] S3: After the vehicle starts, according to the vehicle route navigation, obtain the nearest travel point to the vehicle in the direction of travel, and use it as entry point one;

[0019] S4: Obtain the roadside perception module closest to entry point 1 as the first sensing module, and obtain the first-class view data group from the moment the vehicle enters the coverage area of ​​the data communication module until the distance between the vehicle and the first sensing module is R3.

[0020] Among them, 0 <R3<(R1-R2);

[0021] Obtain the next travel point in the direction of travel, and obtain the view data group between the entry point and the travel point as the continuation view data group;

[0022] Continue to acquire the next travel point, and use the view data group between every two adjacent travel points as the continuation view data group.

[0023] Furthermore, the method for acquiring the viewpoint data group is as follows:

[0024] The processor labels the information received from the millimeter-wave radar sensor, camera sensor, and lidar sensor as Gij, where i = 1, 2, 3, j = 1, 2, 3, ..., n; and G1n, G2n, and G3n represent the information uploaded by the millimeter-wave radar sensor, camera sensor, and lidar sensor for the nth time, respectively.

[0025] Using the time when the millimeter-wave radar sensor first uploads environmental information as the reference time, starting from the reference time, a time period is extracted every time T, and the environmental information within the same time period is divided into a group of environmental information.

[0026] Eliminate redundant and erroneous information groups:

[0027] Select any set of environmental information and obtain the number of pieces of information transmitted by the millimeter-wave radar sensor, camera sensor, and lidar sensor in the environmental information; if the number of pieces of information transmitted by any one of the millimeter-wave radar sensor, camera sensor, and lidar sensor is greater than 1 or equal to 0, then discard the set of environmental information.

[0028] Among the filtered environmental information, randomly select two sets of adjacent environmental information in terms of time, and judge the time intervals between the environmental information uploaded by two millimeter-wave radar sensors, two camera sensors, and two lidar sensors in the two sets of environmental information. If the time interval between the environmental information uploaded by two millimeter-wave radar sensors or two camera sensors or two lidar sensors is greater than T*β, mark the corresponding sensor as an abnormal sensor. If the number of times that the time interval between two adjacent sets of environmental information uploaded by the abnormal sensor is greater than T*β within the preset time T2 is greater than 3 times, after the data packets transmitted by the abnormal sensor are confirmed by the administrator, they are either excluded or retained;

[0029] Mark the remaining environmental information groups as perspective data groups;

[0030] Among them, β≥1, and β is a preset value.

[0031] Furthermore, when the edge data processing module analyzes the environmental information to obtain the surrounding event types, the following algorithm is executed:

[0032] F001: Obtain the environmental information in the first type of perspective data group as the analysis data;

[0033] F002: Retrieve the environmental information collected by the camera sensor from the analysis data, intercept the picture information collected by the camera every preset time T4. If the positions of the same object on the road in three consecutive pictures do not change, mark the corresponding object as a suspected obstacle;

[0034] F003: Along the driving direction of the road, if the vehicle distance between the foremost suspected obstacle measured by the millimeter-wave radar sensor and the next adjacent obstacle behind it is <L1, it is determined as a traffic accident event;

[0035] If the vehicle distance between the foremost suspected obstacle measured by the millimeter-wave radar sensor and the next adjacent obstacle behind it is ≥L1, and the vehicle distance between any two other adjacent suspected obstacles in the same lane is ≥L1, it is determined as a traffic jam event; L1 is a preset value;

[0036] Along the driving direction of the road, if there are pedestrians in the environmental information collected by the camera sensor, it is determined as a pedestrian crossing event;

[0037] If a vehicle presses on the lane dividing line for more than the preset time T5 in the environmental information collected by the camera sensor, it is determined as a lane departure event.

[0038] Furthermore, when the edge data processing module analyzes the environmental information, it also includes:

[0039] Z001: Obtain the environmental information in the subsequent type of perspective data group as the analysis data;

[0040] Z002: Retrieve each set of environmental information collected by the road-end perception module from the analysis data. According to the time sequence, if abnormal events exist in all of the three adjacent sets of environmental information collected by the millimeter-wave radar sensor, the camera sensor, and the lidar sensor, directly determine the abnormal event as a traffic accident event / traffic jam event / pedestrian crossing event / lane departure event according to step F003;

[0041] Z003: If the determination of abnormal events in the environmental information collected by the millimeter-wave radar sensor, the camera sensor, and the lidar sensor in the three adjacent sets of environmental information is inconsistent, then:

[0042] Arbitrarily select one of the millimeter-wave radar sensor, the camera sensor, and the lidar sensor and mark it as the reference parameter sensor; if the abnormal events collected by the reference parameter sensor in the three adjacent sets of environmental information are consistent, directly determine the abnormal event as a traffic accident event / traffic jam event / pedestrian crossing event / lane departure event according to step F003;

[0043] Otherwise, mark the corresponding environmental information as fuzzy information. If fuzzy information is continuously received 3 times within the preset time T6, determine the abnormal event as a traffic emergency abnormal event.

[0044] Furthermore, the millimeter-wave radar sensor is also used to detect the speed of vehicles on the road, and the lidar sensor obtains the distance and azimuth angle between the vehicle and the target object.

[0045] Furthermore, when the distance between the vehicle and the target object is less than the preset distance W, the edge data processing module obtains the current vehicle speed Vw, [[ID=E19]]

[0046] Calculate the estimated time:

[0047] If the estimated time Tm < Td1, send an emergency braking notice to the vehicle;

[0048] If Td1 ≤ Tm < Td2, for the generated traffic accident event / traffic jam event / lane departure event, send a detour notice to the vehicle; for the pedestrian crossing event, send a deceleration and waiting notice to the vehicle;

[0049] where Vmin and Vmax are the minimum driving speed and the maximum driving speed of the vehicle between the distance of 2W and W from the target object respectively. If the previous driving distance is less than W, use the minimum driving speed and the maximum driving speed during driving as Vmin and Vmax; η is a preset value, and 0 < η < 1, and Td1 and Td2 are preset times.

[0050] The present invention has the following beneficial effects:

[0051] This invention uses sensors to monitor the status of road users and broadcasts the information to surrounding vehicles with ultra-low latency via C-V2X or 5G technology, allowing them to know their environment. This enhances the environmental perception capabilities of in-vehicle driver assistance systems and future autonomous vehicles, enabling each car to have a high-altitude view of the road and environment around it.

[0052] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a schematic diagram of the emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Please see Figure 1 As shown, this invention is an emergency avoidance perception and decision-making system for autonomous vehicles based on vehicle-road cooperation. It addresses the problems in existing technologies, such as limited field of view of vehicle-side sensors, poor perception in adverse weather conditions, and insufficient time for autonomous vehicles to react and avoid risks. During operation, the system receives environmental information from at least one roadside broadcast unit. This environmental information refers to the surrounding road conditions of the current vehicle, which is within the monitoring range of at least one roadside perception module. Specifically, this system includes:

[0057] The roadside perception module is used to collect environmental information and divide it into primary viewpoint data groups and subsequent viewpoint data groups. The roadside perception module includes millimeter-wave radar sensors, camera sensors, and lidar sensors. Through the multi-fusion sensors on the roadside, it can accurately perceive the road conditions, the position, speed, and direction of participating entities such as vehicles and pedestrians, identify some abnormal traffic events on the road, and implement vehicle-road perception technology. Alternatively, it can simultaneously collect data from millimeter-wave radar sensors, camera sensors, and lidar sensors through a multi-task parallel data acquisition platform to achieve multi-sensor data synchronization.

[0058] The edge data processing module analyzes environmental information, obtains the types of surrounding events, and acquires distance parameters between obstacles and the current vehicle. It then transmits these event types to the vehicle via C-V2X or 5G technology, broadcasting the information to surrounding vehicles with ultra-low latency, allowing them to understand their environment. This significantly enhances the environmental perception capabilities of in-vehicle driver assistance systems and future autonomous vehicles, enabling each car to have a high-altitude view of the surrounding road and environment. The embedded edge data processing module performs traffic event detection and multi-sensor recognition result fusion at the roadside, and then broadcasts the environmental information via the 5G data communication module.

[0059] The vehicle-side decision execution module makes vehicle control decisions based on environmental information sent by the roadside perception module, combined with the vehicle's own dynamics and kinematics information.

[0060] It can solve many of the current problems faced by single-vehicle intelligent driving. In particular, it can obtain more comprehensive road information from the perspective of the roadside overhead view when traffic obstacles suddenly appear in heavy rain or fog, helping the intelligent driving system to make decisions.

[0061] Furthermore, it also includes a data communication module, which adopts C-V2X or 5G technology. Each roadside sensing module is equipped with a data communication module. The communication range of the data communication module covers three adjacent roadside sensing modules and satisfies R1-R2≥r, where R1 is the radius of the communication range coverage, R2 is the straight-line distance between two adjacent roadside sensing modules, and r is a preset distance representing the driver's reaction distance. The intersection of the edge line of the communication range coverage area and the road is used as the travel point. When a car enters the coverage area of ​​the data communication module, the vehicle-side T-box, ADAS domain control, and other remote communication receiving modules receive edge environment information. The vehicle-side decision execution module makes a decision based on the received edge environment information. The decision information is sent to various actuators on the vehicle through vehicle-side communication to complete operations such as emergency braking or emergency steering.

[0062] As an embodiment of the present invention, preferably, the roadside sensing module classifies the collected environmental information in the following manner:

[0063] S0: The millimeter-wave radar sensor, camera sensor, and lidar sensor all upload the collected environmental information to the processor every preset time T.

[0064] S1: Information grouping: The processor divides the environmental information received from the roadside perception module within the time period T into a group, removes redundant and erroneous information groups, and marks the remaining environmental information group as the view data group.

[0065] S2: If it occurs within the preset time T3 If there is a redundant information group, then the environmental information collected by the camera sensor is retained as the viewpoint data group; where [ξ] represents taking the integer part of ξ, This is the default value;

[0066] S3: After the vehicle starts, according to the vehicle route navigation, obtain the nearest travel point to the vehicle in the direction of travel, and use it as entry point one;

[0067] S4: Obtain the roadside perception module closest to entry point 1 as the first sensing module, and obtain the first-class view data group from the moment the vehicle enters the coverage area of ​​the data communication module until the distance between the vehicle and the first sensing module is R3.

[0068] Among them, 0 <R3<(R1-R2);

[0069] Obtain the next travel point in the direction of travel, and obtain the view data group between the entry point and the travel point as the continuation view data group;

[0070] Continue to acquire the next travel point, and use the view data group between every two adjacent travel points as the continuation view data group.

[0071] As an embodiment of the present invention, preferably, the method for acquiring the viewpoint data group is as follows:

[0072] The processor labels the information received from the millimeter-wave radar sensor, camera sensor, and lidar sensor as Gij, where i = 1, 2, 3, j = 1, 2, 3, ..., n; and G1n, G2n, and G3n represent the information uploaded by the millimeter-wave radar sensor, camera sensor, and lidar sensor for the nth time, respectively.

[0073] Take the time when the millimeter-wave radar sensor first uploads environmental information as the reference time. Starting from the reference time, intercept a time period every time T. The environmental information within the same time period is divided into a group of environmental information;

[0074] Eliminate the redundant error information group:

[0075] Arbitrarily select a group of environmental information, and obtain the number of information transmitted by the millimeter-wave radar sensor, camera sensor, and lidar sensor in the environmental information. If the number of information transmitted by any one of the millimeter-wave radar sensor, camera sensor, and lidar sensor is greater than 1 or equal to 0, then eliminate this group of environmental information;

[0076] Among the remaining environmental information, arbitrarily select two adjacent groups of environmental information in time. Judge the time interval between the environmental information uploaded by two millimeter-wave radar sensors, two camera sensors, and two lidar sensors in the two groups of environmental information. If the time interval between the environmental information uploaded by two millimeter-wave radar sensors or two camera sensors or two lidar sensors is greater than T*β, then mark the corresponding sensor as an abnormal sensor. If within the preset time T2, the number of times that the time interval between two adjacent groups of environmental information uploaded by the abnormal sensor is greater than T*β is greater than 3 times, then after the data packets transmitted by the abnormal sensor are confirmed by the administrator, they are either eliminated or retained;

[0077] Mark the remaining environmental information groups as perspective data groups;

[0078] Among them, β≥1, and β is a preset value.

[0079] As an embodiment provided by the present invention, preferably, when the edge data processing module analyzes the environmental information to obtain the surrounding event type, the following algorithm is executed:

[0080] F001: Obtain the environmental information in the first type of perspective data group as the analysis data;

[0081] F002: Retrieve the environmental information collected by the camera sensor from the analysis data, and intercept the picture information collected by the camera every preset time T4. If the positions of the same object on the road in three consecutive pictures do not change, then mark the corresponding object as a suspected obstacle;

[0082] F003: Along the driving direction of the road, if the vehicle distance between the frontmost suspected obstacle measured by the millimeter-wave radar sensor and the next adjacent obstacle behind it < L1, then it is determined as a traffic accident event;

[0083] If the distance between the suspected obstacle at the front and the adjacent obstacle behind it is ≥L1 as measured by the millimeter-wave radar sensor, and the distance between any two other adjacent suspected obstacles in the same lane is ≥L1, then it is determined to be a traffic congestion event; L1 is a preset value.

[0084] If a pedestrian is detected in the environmental information collected by the camera sensor along the direction of travel on the road, it is determined as a pedestrian crossing event.

[0085] If the environmental information collected by the camera sensor shows that a vehicle has been pressing on the lane divider for more than a preset time T5, it is determined to be a lane departure event.

[0086] More preferably, the following aspects can be determined based on existing methods for judging other accident types: 1. The type of event occurring directly in front of the vehicle, such as a traffic accident, road construction blocking the lane, a vehicle breakdown blocking the lane, slow traffic due to heavy fog, or a rockfall. 2. Whether the vehicle's current lane and adjacent lanes are occupied. 3. The distance parameters between obstacles and the current vehicle in the environmental information, including lateral and longitudinal parameters for vehicle control. The vehicle-side decision execution module needs to combine the environmental information sent by the roadside perception module and the edge data processing module with its own vehicle dynamics and kinematic information to complete vehicle control decisions. The vehicle controller collects the following vehicle information: vehicle speed, wheel speed, vehicle yaw angle, lateral and longitudinal acceleration, steering wheel angle, etc.

[0087] As an embodiment of the present invention, preferably, the edge data processing module further includes the following when analyzing environmental information:

[0088] Z001: Obtain environmental information from the continuation class perspective data group as analysis data;

[0089] Z002: Retrieve each set of environmental information collected by the roadside perception module from the analysis data. According to the time sequence, if there are abnormal events in all three adjacent sets of environmental information collected by the millimeter-wave radar sensor, camera sensor, and lidar sensor, then directly determine the abnormal event as a traffic accident event / traffic congestion event / pedestrian crossing event / lane departure event according to step F003.

[0090] Z003: If the determination of abnormal events in the environmental information collected by the millimeter-wave radar sensor, camera sensor, and lidar sensor in three adjacent sets of environmental information is inconsistent, then:

[0091] Select one of the millimeter-wave radar sensor, camera sensor, and lidar sensor, and mark it as the reference parameter sensor; if the abnormal events collected by the reference parameter sensor in three adjacent groups of environmental information are the same, directly determine the abnormal event as a traffic accident event / traffic jam event / pedestrian crossing event / lane departure event according to step F003;

[0092] Otherwise, mark the corresponding environmental information as fuzzy information. If three fuzzy information are continuously received within the preset time T6, determine the abnormal event as a traffic emergency abnormal event.

[0093] As an embodiment provided by the present invention, preferably, the millimeter-wave radar sensor is further used to detect the speed of vehicles on the road, and the lidar sensor obtains the distance and azimuth angle between the vehicle and the target.

[0094] As an embodiment provided by the present invention, preferably, when the distance between the vehicle and the target is less than the preset distance W, the edge data processing module obtains the current vehicle speed Vw of the vehicle,

[0095] Calculate the estimated time:

[0096] If the estimated time Tm < Td1, send an emergency braking notice to the vehicle;

[0097] If Td1 ≤ Tm < Td2, for the generated traffic accident event / traffic jam event / lane departure event, send a detour notice to the vehicle; for the pedestrian crossing event, send a deceleration and waiting notice to the vehicle;

[0098] Where, Vmin and Vmax are respectively the minimum driving speed and the maximum driving speed of the vehicle between the distance of 2W and W from the target. If the previous driving distance is less than W, use the minimum driving speed and the maximum driving speed during driving as Vmin and Vmax; η is a preset value, and 0 < η < 1, and Td1 and Td2 are preset times.

[0099] An emergency avoidance perception and decision-making system for autonomous vehicles based on vehicle-road cooperation leverages the advantages of camera sensors in traffic event detection, lane detection, and vehicle and pedestrian classification; millimeter-wave radar sensors for stable speed detection in all weather conditions; and lidar sensors for target distance and azimuth detection, enabling more accurate and timely detection of various emergency situations on the road. By monitoring the status of road participants (people, vehicles, and non-motorized vehicles) through sensors, and broadcasting information to surrounding vehicles via C-V2X or 5G technology with ultra-low latency, the system informs them of their environment. This enhances the environmental perception capabilities of onboard driver assistance systems and future autonomous vehicles, allowing each vehicle to have a high-altitude view of the surrounding road and environment. A roadside perception unit, based on multi-sensor fusion perception, combines the performance advantages of millimeter-wave radar, camera, and lidar sensors to achieve accurate digital twins of the road, enabling real-time output of key feature information of all traffic participants, filling the perception blind spots of intelligent autonomous vehicles, and providing traffic guidance for different objects. The solution described in this patent has significant advantages in environments that severely affect the perception of a single vehicle sensor, such as sharp curves, high-speed fog, and rain or snow.

[0100] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0101] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An emergency hazard avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation, characterized in that, Including: A roadside perception module, which is used to collect environmental information and divide the environmental information into a first-class perspective data group and a subsequent-class perspective data group. The roadside perception module includes a millimeter-wave radar sensor, a camera sensor, and a lidar sensor; An edge data processing module, which analyzes the environmental information to obtain the types of surrounding events and obtains the distance parameter between the obstacle and the current vehicle in the environmental information; A vehicle-end decision-making and execution module, which completes vehicle control decisions based on the environmental information sent by the roadside perception module in combination with the dynamic and kinematic information of its own vehicle; When the distance between the vehicle and the target is less than the preset distance W, the edge data processing module obtains the current vehicle speed Vw of the vehicle. Calculate the estimated time: If the estimated time Tm < Td1, an emergency braking notice is sent to the vehicle; If the estimated time Td1 ≤ Tm < Td2, for the generated traffic accident event / traffic jam event / lane departure event, a detour notice is sent to the vehicle; for the pedestrian crossing event, a deceleration and waiting notice is sent to the vehicle; Where, Vmin and Vmax are the minimum driving speed and the maximum driving speed of the vehicle between the distance of 2W and W from the target respectively. If the previous driving distance is less than W, the minimum driving speed and the maximum driving speed during driving are used as Vmin and Vmax; η is a preset value, and 0 < η < 1, Td1 and Td2 are preset times.

2. The emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to claim 1, characterized in that, It further includes a data communication module. There is one data communication module at each roadside perception module. The communication range of the data communication module covers three adjacent roadside perception modules and satisfies R1 - R2 ≥ r. Where, R1 is the radius of the communication range coverage, R2 is the straight-line distance between two adjacent roadside perception modules, r is a preset distance, and the intersection of the side line of the communication range coverage area and the road is used as the progress point.

3. The emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to claim 2, characterized in that, The roadside perception module classifies the collected environmental information. The classification method is as follows: S0: The millimeter-wave radar sensor, the camera sensor, and the lidar sensor each upload the collected environmental information to the processor every preset time T; S1: Divide the information group: The processor divides the environmental information transmitted by the roadside perception module received within the T time period into a group, eliminates the redundant error information group, and marks the remaining environmental information group as the perspective data group; S2: If it occurs within the preset time T3 If there is a redundant information group, then the environmental information collected by the camera sensor is retained as the viewpoint data group; where [ξ] represents taking the integer part of ξ, This is the default value; S3: After the vehicle starts, according to the vehicle path navigation, obtain the progress point closest to the vehicle in the forward direction as the first entry point; S4: Obtain the roadside perception module closest to the first entry point as the first sensing module, and obtain the perspective data group from when the vehicle enters the coverage range of the data communication module to when the distance between the vehicle and the first sensing module is R3 as the first-class perspective data group; Where, 0 < R3 < (R1 - R2); Obtain the progress point next to the entry point in the forward direction, and obtain the perspective data group between the entry point and the progress point as the subsequent-class perspective data group; Continue to obtain the next progress point, and use the perspective data group between every two adjacent progress points as the subsequent-class perspective data group.

4. The emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to claim 1, characterized in that, The acquisition method of the perspective data group is as follows: The processor marks the information transmitted by the millimeter-wave radar sensor, camera sensor, and lidar sensor as Gij respectively, where i = 1, 2, 3 and j = 1, 2, 3, …, n; G1n, G2n, and G3n are used to represent the information uploaded by the millimeter-wave radar sensor, camera sensor, and lidar sensor for the nth time respectively. Taking the time when the millimeter-wave radar sensor first uploads the environmental information as the reference time, starting from the reference time, a time period is intercepted every time T, and the environmental information within the same time period is divided into a group of environmental information. Eliminate the error-prone information groups: Arbitrarily select a group of environmental information, and obtain the number of information transmitted by the millimeter-wave radar sensor, camera sensor, and lidar sensor in the environmental information; if the number of information transmitted by any one of the millimeter-wave radar sensor, camera sensor, and lidar sensor is greater than 1 or equal to 0, then eliminate this group of environmental information. Among the remaining environmental information, arbitrarily select two adjacent groups of environmental information in time, and judge the time interval between the environmental information uploaded by the two millimeter-wave radar sensors, two camera sensors, and two lidar sensors in the two groups of environmental information. If the time interval between the environmental information uploaded by two millimeter-wave radar sensors or two camera sensors or two lidar sensors is greater than T*β, then mark the corresponding sensor as an abnormal sensor. If within the preset time T2, the number of times the time interval between the adjacent two groups of environmental information uploaded by the abnormal sensor is greater than T*β is more than 3 times, then after the data packets transmitted by the abnormal sensor are confirmed by the administrator, they are either eliminated or retained. Mark the remaining environmental information groups as perspective data groups. Among them, β≥1, and β is a preset value.

5. The emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to claim 1, characterized in that, When the edge data processing module analyzes the environmental information to obtain the surrounding event types, it executes the following algorithm: F001: Obtain the environmental information in the first type of perspective data group as the analysis data. F002: Retrieve the environmental information collected by the camera sensor from the analysis data, and intercept the picture information collected by the camera every preset time T4. If the positions of the same object on the road in three consecutive pictures do not change, then mark the corresponding object as a suspected obstacle. F003: Along the driving direction of the road, if the vehicle distance between the frontmost suspected obstacle measured by the millimeter-wave radar sensor and the next adjacent obstacle behind it <L1, it is determined as a traffic accident event. If the vehicle distance between the frontmost suspected obstacle measured by the millimeter-wave radar sensor and the next adjacent obstacle behind it ≥L1, and the vehicle distance between any two other adjacent suspected obstacles in the same lane ≥L1, it is determined as a traffic jam event; L1 is a preset value. Along the driving direction of the road, if there are pedestrians in the environmental information collected by the camera sensor, it is determined as a pedestrian crossing event. If the vehicle presses on the lane dividing line for more than the preset time T5 in the environmental information collected by the camera sensor, it is determined as a lane departure event.

6. The emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to claim 5, characterized in that, When the edge data processing module analyzes the environmental information, it also includes: Z001: Obtain environmental information from the continuation class perspective data group as analysis data; Z002: Retrieve each set of environmental information collected by the roadside perception module from the analysis data. According to the time sequence, if there are abnormal events in all three adjacent sets of environmental information collected by the millimeter-wave radar sensor, camera sensor, and lidar sensor, then directly determine the abnormal event as a traffic accident event / traffic congestion event / pedestrian crossing event / lane departure event according to step F003. Z003: If the determination of abnormal events in the environmental information collected by the millimeter-wave radar sensor, camera sensor, and lidar sensor in three adjacent sets of environmental information is inconsistent, then: Choose one of the following: millimeter-wave radar sensor, camera sensor, or lidar sensor, and mark it as a standard parameter sensor; if the abnormal events collected by the standard parameter sensor in three adjacent sets of environmental information are consistent, then directly determine the abnormal event as a traffic accident event / traffic jam event / pedestrian crossing event / lane departure event according to step F003. Otherwise, the corresponding environmental information will be marked as ambiguous information. If ambiguous information is received three times consecutively within a preset time T6, the abnormal event will be judged as a traffic emergency abnormal event.

7. The emergency avoidance perception and decision-making system for unmanned vehicles based on vehicle-road cooperation according to claim 6, characterized in that, The millimeter-wave radar sensor is also used to detect the speed of vehicles on the road, and the lidar sensor acquires the distance and azimuth angle between the vehicle and the target.

Images