A method and system for assessing the dangerous behavior of autonomous vehicles

Abstract

The present invention is applicable to the technical field of dangerous behavior assessment, and in particular relates to a dangerous behavior assessment method and system for autonomous driving vehicles, the method comprising: delineating a dangerous warning area for the autonomous driving vehicle, wherein the dangerous warning area is composed of motor vehicle lanes and non-motor vehicle lanes, utilizing monitoring equipment pre-deployed in the dangerous warning area to collect video data, select traffic participants that affect each other, and define sample vehicles in the motor vehicle lanes and interference objects in the non-motor vehicle lanes; collecting behavioral characteristics of the interference objects, and reading out the linkage response of the sample vehicles at a preset step time. By determining the target response, the present invention can provide autonomous driving vehicles with real and referenceable decision samples, optimize the response strategy of autonomous driving vehicles in complex scenarios, reduce the occurrence of false triggering of emergency braking, reduce the safety risks of autonomous driving vehicles passing through dangerous warning areas, and improve traffic safety.

Description

Technical Field

[0001] This invention relates to the field of hazardous behavior assessment technology, and in particular to a method and system for assessing hazardous behavior of autonomous vehicles. Background Technology

[0002] Autonomous vehicles are typically used in limited scenarios or closed areas, such as express delivery, freight, and within industrial parks.

[0003] Taking unmanned logistics vehicles of express delivery companies as an example, when driving on straight road sections, the safety risks are relatively controllable due to the relatively simple road environment, fewer traffic participants, and stable vehicle trajectory. However, when passing through intersections, they are prone to dangerous behavior because they cannot understand the intentions of other traffic participants on the road.

[0004] For example, when an unmanned logistics vehicle is preparing to turn right in the motor vehicle lane, a rider of an electric bicycle in the adjacent non-motorized vehicle lane also turns on their right turn signal or slightly shifts to the right, indicating their intention to turn right. Because the unmanned logistics vehicle currently lacks accurate recognition and understanding of the intentions of traffic participants, it cannot predict the future behavior of the electric bicycle rider and thus trigger emergency braking to avoid a collision when approaching the intersection. However, the human-driven vehicle following closely behind may have judged from the riding behavior that the electric bicycle will turn right and assume that the unmanned logistics vehicle will pass through the intersection at a constant speed, so it is not prepared to brake in time, ultimately creating a safety hazard and possibly even causing a rear-end collision.

[0005] Therefore, "how to identify the travel intentions of traffic participants in the adjacent non-motorized vehicle lane when passing through an intersection" is the technical problem that this invention needs to solve. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for assessing the dangerous behavior of autonomous vehicles, in order to solve the problem mentioned in the background art of "how to identify the travel intentions of traffic participants in adjacent non-motorized vehicle lanes when passing through intersections".

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for assessing the dangerous behavior of an autonomous vehicle, the method comprising:

[0009] A danger warning zone for autonomous vehicles is defined, which consists of motor vehicle lanes and non-motor vehicle lanes. Video data is collected using monitoring equipment pre-deployed in the danger warning zone, and traffic participants that affect each other are selected. Sample vehicles are defined in the motor vehicle lanes, and interference objects are defined in the non-motor vehicle lanes.

[0010] Collect the behavioral characteristics of the interference objects, and read the linkage response of the sample vehicles at a preset step time. Integrate the behavioral characteristics and linkage response to generate a rule base and insert risk level items.

[0011] When the Bluetooth beacon signal of the autonomous vehicle is collected by the Bluetooth device deployed in the danger warning area, the target terminal is defined and the monitoring device is activated to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane, compare them with the rule base, and obtain the target response.

[0012] Establish a communication link between the target terminal and the Bluetooth device, and capture the target terminal in the video data. When the target terminal reaches the preset marker point, open the communication link and send the target response to the target terminal.

[0013] Furthermore, the steps of collecting video data, selecting mutually influencing traffic participants, defining sample vehicles in the motor vehicle lane, and defining interference objects in the non-motor vehicle lane include:

[0014] A behavior pattern recognition model is constructed. When a sample vehicle arrives at a marker point, a snapshot of the non-motorized vehicle lane area is extracted from the video data and input into the behavior pattern recognition model to output the potential intent.

[0015] Based on the potential intent, determine whether there are traffic participants in the motor vehicle lane and the non-motor vehicle lane who influence each other.

[0016] Furthermore, the step of defining the interference object in the non-motorized vehicle lane includes:

[0017] Acquire the driving data of the interfering object, wherein the driving data includes at least: vehicle speed and driving position;

[0018] The risk level of the deviation target response is determined based on the driving data.

[0019] Furthermore, the steps of collecting the behavioral characteristics of the interference object, reading the linkage response of the sample vehicle within a preset step time, integrating the behavioral characteristics and linkage response, generating a rule base, and inserting risk level items include:

[0020] Create a response plan that corresponds one-to-one with the risk level, and activate the corresponding response plan once the target response is determined;

[0021] Remove abnormal responses from the rule base, integrate the rule bases of all danger warning areas, build a management platform that coordinates all monitoring devices and Bluetooth devices, and upload the integrated rule base to the management platform.

[0022] Furthermore, the step of defining the target terminal and activating the monitoring device when the Bluetooth beacon signal of the autonomous vehicle is collected using a Bluetooth device deployed in the danger warning area includes:

[0023] Using the monitoring equipment, the attribute characteristics of the autonomous vehicle are collected, wherein the attribute characteristics include at least: license plate and appearance;

[0024] Determine whether the target terminal meets the attribute characteristics; if not, redefine the target terminal.

[0025] Furthermore, the method also includes:

[0026] Using the monitoring device as the center and a preset distance as the radius, a data processing range is constructed, and edge nodes are identified.

[0027] The video data is sent to edge nodes, where each monitoring device corresponds to at least one edge node.

[0028] Furthermore, the step of opening the communication link and sending the target response to the target terminal after the target terminal reaches the preset marker point includes:

[0029] From the video data, abnormal events are defined, and the total number of abnormal events in each danger warning area is counted.

[0030] Dangerous warning areas with a total number greater than a threshold are selected and defined as slow-moving road sections. When an autonomous vehicle reaches a slow-moving road section, a pre-configured deceleration control strategy is invoked and executed.

[0031] Furthermore, the system includes:

[0032] The delineation module is used to delineate the danger warning area for autonomous vehicles. The danger warning area consists of motor vehicle lanes and non-motor vehicle lanes. Using monitoring equipment pre-deployed in the danger warning area, video data is collected, and traffic participants that affect each other are selected. Sample vehicles are defined in the motor vehicle lanes, and interference objects are defined in the non-motor vehicle lanes.

[0033] The insertion module is used to collect the behavioral characteristics of the interference object, and at a preset step time, read the linkage response of the sample vehicle, integrate the behavioral characteristics and linkage response, generate a rule base, and insert risk level items.

[0034] The module is used to define the target terminal when the Bluetooth beacon signal of the autonomous vehicle is collected by the Bluetooth device deployed in the danger warning area, and to activate the monitoring device to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane, compare them with the rule base, and obtain the target response.

[0035] The delivery module is used to establish a communication link between the target terminal and the Bluetooth device, and to capture the target terminal in the video data. When the target terminal reaches the preset marker point, the communication link is opened and the target response is sent to the target terminal.

[0036] Furthermore, the delineation module includes:

[0037] The construction unit is used to construct a behavior pattern recognition model. When a sample vehicle arrives at the marker point, a snapshot of the non-motorized vehicle lane area is extracted from the video data, and the snapshot is input into the behavior pattern recognition model to output the potential intent.

[0038] The judgment unit is used to determine, based on the potential intent, whether there are traffic participants in the motor vehicle lane and the non-motor vehicle lane who affect each other;

[0039] An acquisition unit is used to acquire the driving data of the interfering object, wherein the driving data includes at least: vehicle speed and driving position;

[0040] An offset unit is used to offset the risk level of the target response based on the driving data.

[0041] Furthermore, the insertion module includes:

[0042] The activation unit is used to create a response plan that corresponds one-to-one with the risk level. Once the target response is determined, the corresponding response plan is activated.

[0043] The upload unit is used to delete abnormal responses from the rule base, integrate the rule bases of all danger warning areas, build a management platform that coordinates all monitoring devices and Bluetooth devices, and upload the integrated rule base to the management platform.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] By delineating danger warning zones, potential conflict behaviors can be detected in advance, assisting autonomous vehicles in understanding their travel intentions and improving traffic safety. By selecting interference targets, potential conflict targets can be identified in advance, providing a data foundation for travel intention recognition. By building a rule base, human experience can be used to supplement the decision-making ability of autonomous vehicles in uncertain environments. "Imitation learning" can be used to assist in decision optimization and provide a reference for emergency handling of autonomous vehicles. By determining target responses, real and referable decision samples can be provided for autonomous vehicles, optimizing their response strategies in complex scenarios, reducing false emergency braking, lowering the safety risks of autonomous vehicles passing through danger warning zones, and improving traffic safety. Attached Figure Description

[0046] Figure 1 This is a diagram showing the location of the danger warning area;

[0047] Figure 2 A flowchart illustrating the method for assessing the dangerous behavior of autonomous vehicles provided in an embodiment of the present invention;

[0048] Figure 3 This is a first sub-flowchart of the method for assessing the dangerous behavior of autonomous vehicles provided in an embodiment of the present invention;

[0049] Figure 4 This is a second sub-flowchart of the method for assessing the dangerous behavior of autonomous vehicles provided in an embodiment of the present invention;

[0050] Figure 5 A third sub-flowchart of the method for assessing the dangerous behavior of autonomous vehicles provided in an embodiment of the present invention;

[0051] Figure 6 The fourth sub-flowchart of the method for assessing the dangerous behavior of autonomous vehicles provided in this embodiment of the invention;

[0052] Figure 7 A block diagram illustrating the components of a method for assessing the dangerous behavior of autonomous vehicles provided in an embodiment of the present invention;

[0053] Figure 8 A block diagram showing the components of the delineation module in the hazardous behavior assessment system for autonomous vehicles provided in this embodiment of the invention;

[0054] Figure 9 A block diagram of the components of the insertion module in the hazardous behavior assessment system for autonomous vehicles provided in an embodiment of the present invention;

[0055] Figure 10 A block diagram of the module obtained in the hazardous behavior assessment system for autonomous vehicles provided in this embodiment of the invention.

[0056] Figure 11 A block diagram of the distribution module in the hazardous behavior assessment system for autonomous vehicles provided in this embodiment of the invention. Detailed Implementation

[0057] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0058] In Example 1, Figure 1 and Figure 2 The implementation flow of the hazardous behavior assessment method for autonomous vehicles provided in this embodiment of the invention is illustrated below, and is described in detail below:

[0059] S100: Delineate the danger warning zone for autonomous vehicles. The danger warning zone consists of motor vehicle lanes and non-motor vehicle lanes. Using monitoring equipment pre-deployed in the danger warning zone, video data is collected, and traffic participants that affect each other are selected. Sample vehicles are defined in the motor vehicle lanes, and interference objects are defined in the non-motor vehicle lanes.

[0060] The danger warning zone for autonomous vehicles is defined. This danger warning zone is located at traffic light intersections or other intersections and includes the motor vehicle lane and the adjacent non-motor vehicle lane. For example, the danger warning zone is the right-turn lane and the adjacent non-motor vehicle lane at a four-way intersection (as shown in the instruction manual). Figure 1 The two groups of areas A shown in this application (taking an autonomous vehicle needing to turn right as an example, with left turns being similar) utilize monitoring equipment deployed within the danger warning area. This monitoring equipment can be a high-definition camera or other video acquisition equipment to collect video data within the danger warning area, obtaining dynamic information of all traffic participants in the area, including various traffic entities such as motor vehicles, non-motor vehicles (such as electric bicycles), and pedestrians. Using existing image recognition and behavior analysis algorithms, target detection, tracking, and behavior recognition are performed on the collected video stream to filter out traffic participants with potential conflicts in space and time, i.e., traffic participants that influence each other. The analysis and processing of the video stream data can be carried out at the edge nodes.

[0061] Vehicles that are about to turn right in the motor vehicle lane are defined as sample vehicles, and pedestrians or non-motorized vehicles that may conflict with the sample vehicles and are located in the non-motorized vehicle lane are defined as interference objects.

[0062] S200: Collects the behavioral characteristics of the interference object, and reads the linkage response of the sample vehicle at a preset step time. It integrates the behavioral characteristics and linkage response to generate a rule base and inserts risk level items.

[0063] Using monitoring equipment, video data containing interfering objects is collected. Image recognition and pedestrian analysis algorithms are used to extract the behavioral characteristics of the interfering objects, such as turning on the turn signal (indicating an intention to change lanes or turn), the brake light illuminating (indicating that the vehicle is decelerating or about to stop), and the vehicle body slightly shifting to the right of the lane (which may be a sign of lane change, avoidance, or turning). After a preset step time interval (e.g., 0.5 seconds or 1 second) after detecting the behavioral characteristics, the linked responses of the sample vehicles are collected and recorded, such as deceleration, steering, emergency braking, path adjustment, or maintaining the current state. A mapping relationship of "behavioral characteristics - linked responses" is constructed, and all behavioral characteristics and linked responses are integrated to generate a rule base. Risk level items are inserted into the rule base to determine the risk level of each mapping relationship.

[0064] For example, when a sample vehicle is about to turn right, the monitoring equipment captures the following behavioral characteristics of an electric bicycle in the right-hand non-motorized vehicle lane: turning on the right turn signal, slightly veering to the right in the lane, but not slowing down; the time of the behavioral characteristics is recorded, assuming it is T seconds, with a preset step size of 1 second; after the sample vehicle detects the interference object's intention to turn right, it does not slow down but passes through the intersection at a relatively slow, constant speed; then the linkage response of the sample vehicle after T+1 seconds is recorded, i.e., it does not slow down and passes through the intersection at a constant speed; integrating the behavioral characteristics and linkage response, the mapping relationship "turning on the right turn signal, slightly veering to the right in the lane, but not slowing down - not slowing down, passing through the intersection at a constant speed" is obtained, the corresponding risk level is determined, and all obtained mapping relationships are integrated to obtain a rule base; where the higher the risk level, the higher the probability of a collision between the sample vehicle and the interference object.

[0065] In this application, by collecting the linkage response of sample vehicles when passing through intersections, a data foundation is provided for the driving operation and emergency handling of autonomous vehicles under the same traffic conditions, thereby avoiding sudden braking and reducing the possibility of safety accidents.

[0066] S300: When the Bluetooth beacon signal of the autonomous vehicle is collected by the Bluetooth device deployed in the danger warning area, the target terminal is defined and the monitoring device is activated to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane, compare them with the rule base, and obtain the target response.

[0067] Using Bluetooth devices pre-deployed within the danger warning area, Bluetooth beacon signals from all devices within the signal coverage area are collected. Based on the unique identifier in the Bluetooth beacon signal, the Bluetooth beacon signal corresponding to the autonomous vehicle is identified, and the corresponding autonomous vehicle is defined as the target terminal. The presence of a Bluetooth beacon signal from an autonomous vehicle can be determined by the fixed prefix or suffix of the unique identifier. Once the target terminal is identified, the monitoring equipment is activated to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane. These characteristics are compared with a rule base to find the corresponding linkage response and define it as the target response. In simple terms, the target response is the driving action performed by the sample vehicle under the same traffic conditions.

[0068] Continuing with the example in S200, when the real-time behavior characteristic is "turning on the right turn signal, slightly deviating to the right of the lane, but not decelerating", then the corresponding "but not decelerating - not decelerating, passing through the intersection at a constant speed" is defined as the target response. In other words, under the same traffic conditions, the autonomous vehicle does not need to decelerate and can pass through the intersection at a constant speed. Here, deceleration refers to an additional secondary deceleration. The autonomous vehicle should decelerate in advance when entering the intersection.

[0069] S400: Establish a communication link between the target terminal and the Bluetooth device, and capture the target terminal in the video data. When the target terminal reaches the preset marker point, open the communication link and send the target response to the target terminal.

[0070] By utilizing Bluetooth beacon signals, a communication link is established between the target terminal and the Bluetooth device. Through target recognition and tracking algorithms, the position of the target terminal in the video frame is determined. Once the target terminal is detected to have reached a preset marker point, which can be a zebra crossing or other location in front of an intersection, the communication link is opened, allowing data to interact between the Bluetooth device and the target terminal. The target response is then sent to the target terminal in real time via the Bluetooth device, guiding the autonomous vehicle to perform corresponding driving operations.

[0071] In Example 2, Figure 3 The implementation flow of the hazardous behavior assessment method for autonomous vehicles provided by an embodiment of the present invention is illustrated. The following details the steps of collecting video data, selecting mutually influencing traffic participants, defining sample vehicles in the motor vehicle lane, and defining interference objects in the non-motor vehicle lane:

[0072] S101: Construct a behavior pattern recognition model. When a vehicle arrives at a marker point, extract a snapshot of the non-motorized vehicle lane area from the video data and input the snapshot into the behavior pattern recognition model to output the potential intent.

[0073] A behavior pattern recognition model is constructed based on convolutional neural networks. This model is mainly used to identify the potential behavioral intentions of participants in non-motorized vehicle lanes. When a sample vehicle reaches a preset marker point, a snapshot of the current non-motorized vehicle lane area is extracted from the video stream. This snapshot includes the real-time position, posture, and direction of movement of traffic participants such as electric bicycles and pedestrians. This snapshot is used as input to the trained behavior pattern recognition model, and the output is the potential intention of the non-motorized vehicle or pedestrian. The potential intention includes whether the vehicle is preparing to cross the road, turn right, accelerate, decelerate, or suddenly change direction.

[0074] S102: Based on the potential intent, determine whether there are traffic participants in the motor vehicle lane and the non-motor vehicle lane who affect each other.

[0075] Based on the potential intentions of non-motorized vehicles and pedestrians, determine whether there are traffic participants in the non-motorized vehicle lane who could affect the passage of vehicles in the motorized vehicle lane.

[0076] In Example 3, Figure 3 The implementation flow of the hazardous behavior assessment method for autonomous vehicles provided by an embodiment of the present invention is illustrated below. The step of defining the interference object in the non-motorized vehicle lane is described in detail below:

[0077] S103: Obtain the driving data of the interfering object, wherein the driving data includes at least: vehicle speed and driving position.

[0078] The driving data of the interfering objects in the non-motorized vehicle lane were determined.

[0079] S104: Risk level of the deviation target response based on the driving data.

[0080] Based on driving data, the risk level of the target response is shifted. The shift includes increasing or decreasing the risk level. For example, if the speed of the interfering object exceeds a threshold, the risk level can be increased and the corresponding response plan can be activated to avoid a traffic accident.

[0081] In Example 4, Figure 4 The implementation flow of the hazardous behavior assessment method for autonomous vehicles provided by an embodiment of the present invention is illustrated. The following details the steps of collecting the behavioral characteristics of the interference object, reading the linkage response of the sample vehicle within a preset step time, integrating the behavioral characteristics and linkage response, generating a rule base, and inserting risk level items, as follows:

[0082] S201: Create a response plan that corresponds one-to-one with the risk level. Once the target response is determined, activate the corresponding response plan.

[0083] Create a corresponding response plan for each risk level, which can be divided into high, medium and low. Once the target response is determined, find the corresponding risk level and activate the corresponding response plan. The response plan can be, for example, continuous horn blaring or early deceleration.

[0084] S202: Remove abnormal responses from the rule base, integrate the rule bases of all danger warning areas, build a management platform that coordinates all monitoring devices and Bluetooth devices, and upload the integrated rule base to the management platform.

[0085] In the rule base, abnormal responses are identified. Abnormal responses refer to linkage responses that indicate a traffic accident has occurred or may occur. For example, if a sample vehicle does not slow down when turning right, and the interfering object also does not slow down, resulting in the two being too close to each other, the linkage response corresponding to the sample vehicle is defined as an abnormal response. A management platform is built, which is mainly used to analyze and process all rules in the rule base. Specific processing steps include deleting identical linkage responses and counting the occurrence frequency of linkage responses.

[0086] In Example 5, Figure 5The implementation flow of the hazardous behavior assessment method for autonomous vehicles provided by an embodiment of the present invention is illustrated. The following details the steps of defining the target terminal and activating the monitoring device when the Bluetooth beacon signal of the autonomous vehicle is collected using a Bluetooth device deployed in a hazardous warning area:

[0087] S301: Using the monitoring equipment, collect the attribute characteristics of the autonomous vehicle, wherein the attribute characteristics include at least: license plate and appearance.

[0088] Once the Bluetooth device receives the Bluetooth beacon signal from the autonomous vehicle, it uses monitoring equipment to collect the license plate information of all vehicles and identify the autonomous vehicle among them.

[0089] S302: Determine whether the target terminal meets the attribute characteristics. If not, redefine the target terminal.

[0090] The system determines whether the target terminal is an autonomous vehicle. If not, the target terminal is redefined. The advantage of this method is that by verifying the target terminal, the accuracy of target terminal identification can be improved, ensuring that the target response can be sent to the target terminal in a timely manner.

[0091] In Example 6, unlike Example 1, the method further includes:

[0092] Using the monitoring device as the center and a preset distance as the radius, a data processing range is constructed, and edge nodes are identified.

[0093] The video data is sent to edge nodes, where each monitoring device corresponds to at least one edge node.

[0094] Using the monitoring equipment deployed in the danger warning area as the center and a preset distance as the radius, a data processing range is constructed. The preset distance is determined by traffic management personnel. Edge nodes located within the data processing range are identified. These edge nodes are roadside shops or smart gateways, etc., and video data is sent to the edge nodes for processing.

[0095] In Example 7, Figure 6 The implementation flow of the dangerous behavior assessment method for autonomous vehicles provided by an embodiment of the present invention is illustrated. The following details the steps of opening the communication link and sending the target response to the target terminal after the target terminal reaches the preset marker point:

[0096] S401: Define abnormal events from the video data and count the total number of abnormal events in each danger warning area.

[0097] In the video data, abnormal events are defined, which refer to the state characteristics corresponding to when the sample vehicle or the interfering object brakes suddenly, has a traffic accident, or has obvious dangerous actions. The total number of abnormal events in each danger warning area is calculated.

[0098] S402: Select dangerous warning areas with a total number greater than the threshold and define them as slow-moving road sections. When the autonomous vehicle reaches the slow-moving road section, call and execute the pre-configured deceleration control strategy.

[0099] If the total number exceeds the threshold, the corresponding danger warning area is defined as a slow-moving section. When the autonomous vehicle is in a slow-moving section, the corresponding deceleration control strategy is executed, which is to control the vehicle speed within a preset range.

[0100] Figure 7 This diagram illustrates the structural block diagram of a hazardous behavior assessment system for autonomous vehicles provided in an embodiment of the present invention. The hazardous behavior assessment system 1 for autonomous vehicles includes:

[0101] The delineation module 11 is used to delineate the danger warning area for autonomous vehicles. The danger warning area consists of motor vehicle lanes and non-motor vehicle lanes. Using monitoring equipment pre-deployed in the danger warning area, video data is collected, and traffic participants that affect each other are selected. Sample vehicles are defined in the motor vehicle lanes, and interference objects are defined in the non-motor vehicle lanes.

[0102] Insertion module 12 is used to collect the behavioral characteristics of the interference object, and at a preset step time, read the linkage response of the sample vehicle, integrate the behavioral characteristics and linkage response, generate a rule base, and insert risk level items;

[0103] The module 13 is used to define the target terminal when the Bluetooth beacon signal of the autonomous vehicle is collected by the Bluetooth device deployed in the danger warning area, and to activate the monitoring device to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane, compare them with the rule base, and obtain the target response.

[0104] The sending module 14 is used to establish a communication link between the target terminal and the Bluetooth device, and to capture the target terminal in the video data. When the target terminal reaches the preset marker point, the communication link is opened and the target response is sent to the target terminal.

[0105] Figure 8 This diagram illustrates the structural block diagram of a hazardous behavior assessment system for autonomous vehicles provided in an embodiment of the present invention. The delineation module 11 includes:

[0106] The construction unit 111 is used to construct a behavior pattern recognition model. When a sample vehicle arrives at the marker point, a snapshot of the non-motorized vehicle lane area is extracted from the video data, and the snapshot is input into the behavior pattern recognition model to output the potential intent.

[0107] The judgment unit 112 is used to determine, based on the potential intention, whether there are traffic participants in the motor vehicle lane and the non-motor vehicle lane that affect each other;

[0108] Acquisition unit 113 is used to acquire driving data of the interfering object, wherein the driving data includes at least: vehicle speed and driving position;

[0109] Offset unit 114 is used to offset the risk level of the target response via the driving data.

[0110] Figure 9 This diagram illustrates the structural composition of a hazardous behavior assessment system for autonomous vehicles provided in an embodiment of the present invention. The insertion module 12 includes:

[0111] The activation unit 121 is used to create a response plan that corresponds one-to-one with the risk level. Once the target response is determined, the corresponding response plan is activated.

[0112] Upload unit 122 is used to delete abnormal responses from the rule base, integrate the rule bases of all danger warning areas, build a management platform that coordinates all monitoring devices and Bluetooth devices, and upload the integrated rule base to the management platform.

[0113] Figure 10 This diagram illustrates the structural block diagram of a hazardous behavior assessment system for autonomous vehicles provided in an embodiment of the present invention. The obtaining module 13 includes:

[0114] The acquisition unit 131 is used to acquire the attribute features of the autonomous vehicle using the monitoring equipment, wherein the attribute features include at least: license plate and appearance;

[0115] The definition unit 132 is used to determine whether the target terminal meets the attribute characteristics. If not, the target terminal is redefined.

[0116] Figure 11 This diagram illustrates the structural block diagram of a hazardous behavior assessment system for autonomous vehicles provided in an embodiment of the present invention. The distribution module 14 includes:

[0117] The statistics unit 141 is used to define abnormal events from the video data and count the total number of abnormal events in each danger warning area;

[0118] Calling unit 142 is used to select dangerous warning areas with a total number greater than the threshold and define them as slow-moving road sections. When the autonomous vehicle reaches the slow-moving road section, it calls and executes the pre-configured deceleration control strategy.

[0119] The delineation module 11 is mainly used to complete step S100, the insertion module 12 is mainly used to complete step S200, the obtaining module 13 is mainly used to complete step S300, and the sending module 14 is mainly used to complete step S400.

[0120] The construction unit 111 is mainly used to complete step S101, the judgment unit 112 is mainly used to complete step S102, the acquisition unit 113 is mainly used to complete step S103, and the offset unit 114 is mainly used to complete step S104.

[0121] The starting unit 121 is mainly used to complete step S201, and the uploading unit 122 is mainly used to complete step S202;

[0122] The acquisition unit 131 is mainly used to complete step S301, and the definition unit 132 is mainly used to complete step S302.

[0123] The statistics unit 141 is mainly used to complete step S401, and the calling unit 142 is mainly used to complete step S402.

[0124] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0125] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for assessing the dangerous behavior of autonomous vehicles, characterized in that, The method includes: A danger warning zone for autonomous vehicles is defined, which consists of motor vehicle lanes and non-motor vehicle lanes. Video data is collected using monitoring equipment pre-deployed in the danger warning zone, and traffic participants that affect each other are selected. Sample vehicles are defined in the motor vehicle lanes, and interference objects are defined in the non-motor vehicle lanes. Collect the behavioral characteristics of the interference objects, and read the linkage response of the sample vehicles at a preset step time. Integrate the behavioral characteristics and linkage response to generate a rule base and insert risk level items. When the Bluetooth beacon signal of the autonomous vehicle is collected by the Bluetooth device deployed in the danger warning area, the target terminal is defined and the monitoring device is activated to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane, compare them with the rule base, and obtain the target response. Establish a communication link between the target terminal and the Bluetooth device, and capture the target terminal in the video data. When the target terminal reaches the preset marker point, open the communication link and send the target response to the target terminal.

2. The method for assessing the dangerous behavior of autonomous vehicles according to claim 1, characterized in that, The steps of collecting video data, selecting mutually influencing traffic participants, defining sample vehicles in the motor vehicle lane, and defining interference objects in the non-motor vehicle lane include: A behavior pattern recognition model is constructed. When a sample vehicle arrives at a marker point, a snapshot of the non-motorized vehicle lane area is extracted from the video data and input into the behavior pattern recognition model to output the potential intent. Based on the potential intent, determine whether there are traffic participants in the motor vehicle lane and the non-motor vehicle lane who influence each other.

3. The method for assessing the dangerous behavior of autonomous vehicles according to claim 2, characterized in that, The step of defining the interference object in the non-motorized vehicle lane includes: Acquire the driving data of the interfering object, wherein the driving data includes at least: vehicle speed and driving position; The risk level of the deviation target response is determined based on the driving data.

4. The method for assessing the dangerous behavior of autonomous vehicles according to claim 3, characterized in that, The steps of collecting the behavioral characteristics of the interference object, reading the linkage response of the sample vehicle within a preset step time, integrating the behavioral characteristics and linkage response, generating a rule base, and inserting risk level items include: Create a response plan that corresponds one-to-one with the risk level, and activate the corresponding response plan once the target response is determined; Remove abnormal responses from the rule base, integrate the rule bases of all danger warning areas, build a management platform that coordinates all monitoring devices and Bluetooth devices, and upload the integrated rule base to the management platform.

5. The method for assessing the dangerous behavior of autonomous vehicles according to claim 1, characterized in that, The steps of defining the target terminal and activating the monitoring device when the Bluetooth beacon signal of the autonomous vehicle is collected using a Bluetooth device deployed in the danger warning area include: Using the monitoring equipment, the attribute characteristics of the autonomous vehicle are collected, wherein the attribute characteristics include at least: license plate and appearance; Determine whether the target terminal meets the attribute characteristics; if not, redefine the target terminal.

6. The method for assessing the dangerous behavior of autonomous vehicles according to claim 5, characterized in that, The method further includes: Using the monitoring device as the center and a preset distance as the radius, a data processing range is constructed, and edge nodes are identified. The video data is sent to edge nodes, where each monitoring device corresponds to at least one edge node.

7. The method for assessing the dangerous behavior of autonomous vehicles according to claim 6, characterized in that, The steps of opening the communication link and sending the target response to the target terminal after the target terminal reaches the preset marker point include: From the video data, abnormal events are defined, and the total number of abnormal events in each danger warning area is counted. Dangerous warning areas with a total number greater than a threshold are selected and defined as slow-moving road sections. When an autonomous vehicle reaches a slow-moving road section, a pre-configured deceleration control strategy is invoked and executed.

8. A hazardous behavior assessment system for autonomous vehicles, characterized in that, The system includes: The delineation module is used to delineate the danger warning area for autonomous vehicles. The danger warning area consists of motor vehicle lanes and non-motor vehicle lanes. Using monitoring equipment pre-deployed in the danger warning area, video data is collected, and traffic participants that affect each other are selected. Sample vehicles are defined in the motor vehicle lanes, and interference objects are defined in the non-motor vehicle lanes. The insertion module is used to collect the behavioral characteristics of the interference object, and at a preset step time, read the linkage response of the sample vehicle, integrate the behavioral characteristics and linkage response, generate a rule base, and insert risk level items. The module is used to define the target terminal when the Bluetooth beacon signal of the autonomous vehicle is collected by the Bluetooth device deployed in the danger warning area, and to activate the monitoring device to collect the real-time behavioral characteristics of traffic participants in the non-motorized vehicle lane, compare them with the rule base, and obtain the target response. The delivery module is used to establish a communication link between the target terminal and the Bluetooth device, and to capture the target terminal in the video data. When the target terminal reaches the preset marker point, the communication link is opened and the target response is sent to the target terminal.

9. The hazardous behavior assessment system for autonomous vehicles according to claim 8, characterized in that, The delineation module includes: The construction unit is used to construct a behavior pattern recognition model. When a sample vehicle arrives at the marker point, a snapshot of the non-motorized vehicle lane area is extracted from the video data, and the snapshot is input into the behavior pattern recognition model to output the potential intent. The judgment unit is used to determine, based on the potential intent, whether there are traffic participants in the motor vehicle lane and the non-motor vehicle lane who affect each other; An acquisition unit is used to acquire the driving data of the interfering object, wherein the driving data includes at least: vehicle speed and driving position; An offset unit is used to offset the risk level of the target response based on the driving data.

10. The hazardous behavior assessment system for autonomous vehicles according to claim 9, characterized in that, The insertion module includes: The activation unit is used to create a response plan that corresponds one-to-one with the risk level. Once the target response is determined, the corresponding response plan is activated. The upload unit is used to delete abnormal responses from the rule base, integrate the rule bases of all danger warning areas, build a management platform that coordinates all monitoring devices and Bluetooth devices, and upload the integrated rule base to the management platform.

Images