A method, apparatus, system, and storage medium for avoiding a malfunctioning vehicle

Abstract

This application discloses a method, apparatus, system, and storage medium for avoiding a disabled vehicle. The method includes: real-time acquisition of roadside perception data across the entire intersection area using at least one roadside perception unit deployed at the intersection; processing the roadside perception data using a multi-source perception fusion algorithm to output attribute information of all perceived targets within the intersection; associating and mapping the attribute information of the perceived targets with a pre-stored high-precision map to identify whether a disabled vehicle exists among the perceived targets; when a disabled vehicle exists, determining whether the autonomous driving vehicle needs to avoid it based on the roadside perception data and the state of the autonomous driving vehicle; and when the autonomous driving vehicle needs to avoid it, planning a driving trajectory for the autonomous driving vehicle to avoid the disabled vehicle. This solution can effectively improve vehicle driving safety.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a method, apparatus, system and storage medium for avoiding malfunctioning vehicles. Background Technology

[0002] As autonomous driving technology continues to develop, its safety and traffic efficiency have become key bottlenecks for its practical application. Current mainstream solutions rely heavily on onboard sensors (cameras, millimeter-wave radar, lidar) for single-vehicle perception and decision-making. However, these solutions suffer from the following drawbacks in complex scenarios such as intersections: First, onboard sensors have limited recognition distance and are easily obstructed by obstacles such as large vehicles and trees, leading to blind spots. Second, onboard sensors may not be able to accurately determine the status of vehicles that have been parked for a long time; for example, a malfunctioning vehicle may have been stopped for several minutes, but the onboard sensors may have only just identified it as a short-term stop. Furthermore, single-vehicle intelligence struggles to acquire comprehensive traffic information and cannot accurately assess the impact of a malfunctioning vehicle on the right-of-way of surrounding vehicles. Due to these shortcomings, when autonomous vehicles encounter malfunctioning vehicles, they often fail to perceive them in time and take effective avoidance measures, increasing the risk of traffic accidents.

[0003] Therefore, how to provide a method to avoid disabled vehicles in order to improve vehicle driving safety has become an urgent technical problem to be solved. Summary of the Invention

[0004] This application provides a method, apparatus, system, and storage medium for avoiding disabled vehicles, in order to improve vehicle driving safety.

[0005] This application provides a method for avoiding a disabled vehicle, including: Real-time roadside sensing data of the entire intersection area is collected by deploying at least one roadside sensing unit at the intersection. The roadside sensing data is processed using a multi-source sensing fusion algorithm to output the attribute information of all sensing targets within the intersection. The attribute information of the perceived target is associated and mapped with a pre-stored high-precision map to identify whether there is a faulty vehicle among the perceived targets. When a malfunctioning vehicle is present, the autonomous vehicle is determined to avoid the obstacle based on the roadside perception data and the status of the autonomous vehicle. When the autonomous vehicle needs to avoid a collision, a driving trajectory is planned for the autonomous vehicle to avoid the malfunctioning vehicle.

[0006] The beneficial effects of this application are as follows: Roadside perception data of the entire intersection area is collected in real time by deploying at least one roadside perception unit at the intersection; a multi-source perception fusion algorithm is used to process the roadside perception data to output attribute information of all perceived targets within the intersection; the attribute information of the perceived targets is associated and mapped with a pre-stored high-precision map to identify whether a malfunctioning vehicle exists among the perceived targets; when a malfunctioning vehicle exists, the autonomous vehicle is judged whether it needs to avoid the malfunctioning vehicle based on the roadside perception data and the state of the autonomous vehicle; when the autonomous vehicle needs to avoid the malfunctioning vehicle, a driving trajectory to avoid the malfunctioning vehicle is planned for the autonomous vehicle. This solution utilizes sensors deployed at the intersection to collect roadside perception data of the entire intersection area, avoiding blind spots caused by relying solely on data collected from vehicle-mounted sensors. Furthermore, by utilizing the global perception data provided by the roadside, it is possible to effectively identify malfunctioning vehicles that have been parked for a long time and make more comprehensive avoidance decisions, planning a globally better and safer avoidance trajectory, thereby effectively improving vehicle driving safety.

[0007] In one embodiment, roadside sensing data of the entire intersection area is collected in real time by at least one roadside sensing unit deployed at the intersection, including: Raw sensing data is collected by roadside sensing units set up in multiple directions at intersections, and the raw sensing data is transmitted to roadside or edge computing units in real time. The original sensing data is processed by a preset sensing model in the roadside or edge computing unit to generate roadside sensing data for the entire intersection area. The roadside sensing data includes at least the outline, orientation, and speed and position information of the sensing target relative to the sensing unit.

[0008] In one embodiment, the attribute information of the perceived target is associated and mapped with a pre-stored high-precision map to identify whether a faulty vehicle exists among the perceived targets, including: The attribute information of the perceived target is projected onto the high-precision map to determine whether the perceived target is located within the intersection area; When the perceived target is located within the intersection area, it is determined whether the perceived target is a vehicle in an abnormal parking state; When the sensing target is a vehicle in an abnormal parking state, the duration of the vehicle being in an abnormal parking state is monitored. When the duration exceeds a preset time threshold, the vehicle is determined to be a faulty vehicle.

[0009] In one embodiment, the method further includes: The impact risk of vehicles in abnormal parking states on traffic flow at intersections is assessed in real time or periodically based on the roadside sensing data. The preset time threshold is dynamically adjusted based on the degree of risk of impact on traffic flow at the intersection.

[0010] In one embodiment, determining whether the autonomous vehicle needs to avoid a collision based on the roadside perception data and the state of the autonomous vehicle includes: Determine whether the real-time location and affected area of ​​the faulty vehicle occupy the preset driving path of the autonomous vehicle; When the preset driving path of an autonomous vehicle is occupied, determine whether the road section in front of the malfunctioning vehicle has normal traffic conditions. When the road section in front of the disabled vehicle is passable, the avoidance decision is determined to be that the disabled vehicle needs to be avoided.

[0011] In one embodiment, planning a driving trajectory for the autonomous vehicle to avoid the malfunctioning vehicle includes: Based on the real-time location and impact range of the faulty vehicle, and combined with the safety redundancy distance of the autonomous vehicle, the spatial range that needs to be avoided is calculated. Outside the defined avoidance space, multiple candidate translation paths are generated along a direction perpendicular to the original driving path of the autonomous vehicle. Each generated candidate translation path is smoothly connected to the original driving path of the autonomous vehicle to form multiple complete alternative avoidance paths. The final avoidance trajectory is selected from the multiple complete alternative avoidance paths.

[0012] In one embodiment, selecting the final avoidance trajectory from the plurality of complete alternative avoidance paths includes: A comprehensive evaluation is conducted on each of the generated alternative avoidance paths. The evaluation factors for the comprehensive evaluation include at least the driving safety of the path and the switching cost between the path and the original path of the autonomous vehicle. From the comprehensive evaluation results, the alternative path with the highest safety and the lowest switching cost is selected as the final avoidance driving trajectory.

[0013] This application also provides a device for avoiding a disabled vehicle, comprising: The data acquisition module is used to collect roadside sensing data of the entire intersection area in real time through at least one roadside sensing unit deployed at the intersection. The processing module is used to process the roadside sensing data using a multi-source sensing fusion algorithm to output the attribute information of all sensing targets within the intersection. The identification module is used to associate and map the attribute information of the perceived target with a pre-stored high-precision map in order to identify whether there is a faulty vehicle among the perceived targets; The judgment module is used to determine whether the autonomous vehicle needs to avoid a faulty vehicle based on the roadside perception data and the state of the autonomous vehicle. The planning module is used to plan a driving trajectory for the autonomous vehicle to avoid the malfunctioning vehicle when the autonomous vehicle needs to avoid it.

[0014] In one embodiment, the acquisition module includes: The data acquisition submodule is used to collect raw sensing data by collecting roadside sensing units set up in multiple directions at the intersection, and to transmit the raw sensing data to the roadside or edge computing unit in real time. The processing submodule is used to process the original sensing data through a preset sensing model in the roadside or edge computing unit to generate roadside sensing data for the entire intersection area. The roadside sensing data includes at least the outline, orientation, and speed and position information of the sensing target relative to the sensing unit.

[0015] In one embodiment, the identification module includes: The first judgment submodule is used to project the attribute information of the perceived target onto the high-precision map to determine whether the perceived target is located within the intersection area; The determination submodule is used to determine whether the perceived target is a vehicle in an abnormal parking state when the perceived target is located within the intersection area; The monitoring submodule is used to monitor the duration of the abnormal parking state of the vehicle when the sensing target is a vehicle in an abnormal parking state. The first determining submodule is used to determine that the vehicle is a faulty vehicle when the duration exceeds a preset time threshold.

[0016] In one embodiment, the apparatus further includes: The assessment module is used to assess the risk level of the impact of vehicles in abnormal parking states on traffic flow at intersections in real time or periodically based on the roadside perception data. The adjustment module is used to dynamically adjust the preset time threshold based on the degree of risk of impact on traffic flow at the intersection.

[0017] In one embodiment, the determining module includes: The second judgment submodule is used to determine whether the real-time location and impact range of the faulty vehicle occupy the preset driving path of the autonomous driving vehicle. The third judgment submodule is used to determine whether the road section in front of the faulty vehicle has normal passage conditions when the preset driving path of the autonomous vehicle is occupied. The second determining submodule is used to determine that the avoidance decision is to avoid the disabled vehicle when the road section in front of the disabled vehicle has normal traffic conditions.

[0018] In one embodiment, the planning module includes: The calculation submodule is used to calculate the space range that needs to be avoided based on the real-time location and impact range of the faulty vehicle, combined with the safety redundancy distance of the autonomous vehicle. The generation submodule is used to generate multiple candidate translation paths outside the defined avoidance space, along a direction perpendicular to the original driving path of the autonomous vehicle. The smoothing submodule is used to smoothly connect each generated candidate translation path with the original driving path of the autonomous vehicle to form multiple complete alternative avoidance paths. The selection submodule is used to select the final avoidance driving trajectory from the multiple complete alternative avoidance paths.

[0019] In one embodiment, the selection submodule is further configured to: A comprehensive evaluation is conducted on each of the generated alternative avoidance paths. The evaluation factors for the comprehensive evaluation include at least the driving safety of the path and the switching cost between the path and the original path of the autonomous vehicle. From the comprehensive evaluation results, the alternative path with the highest safety and the lowest switching cost is selected as the final avoidance driving trajectory.

[0020] This application also provides a system for avoiding a disabled vehicle, including: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to implement the method for avoiding a disabled vehicle as described in any of the above embodiments.

[0021] This application also provides a computer-readable storage medium that, when the instructions in the storage medium are executed by a processor corresponding to a system for avoiding a disabled vehicle, enables the system for avoiding a disabled vehicle to implement the method for avoiding a disabled vehicle described in any of the above embodiments.

[0022] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0023] The technical solution of this application will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0024] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a method for avoiding a disabled vehicle in one embodiment of this application; Figure 2 This is a flowchart illustrating the identification of a faulty vehicle in one embodiment of this application; Figure 3 This is a flowchart illustrating the decision-making process for whether to yield to a disabled vehicle in one embodiment of this application. Figure 4 This is a flowchart illustrating the generation of an avoidance path in one embodiment of this application; Figure 5 This is a schematic diagram of the structure of a device for avoiding a disabled vehicle according to an embodiment of this application; Figure 6 This is a schematic diagram of the hardware structure of a system for avoiding a disabled vehicle according to an embodiment of this application. Detailed Implementation

[0025] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application.

[0026] Figure 1 This is a flowchart of a method for avoiding a disabled vehicle according to an embodiment of this application, as shown below. Figure 1 As shown, the method can be implemented as follows: S101-S105: In step S101, roadside sensing data of the entire intersection area is collected in real time by at least one roadside sensing unit deployed at the intersection. In step S102, a multi-source sensing fusion algorithm is used to process the roadside sensing data to output the attribute information of all sensing targets within the intersection. In step S103, the attribute information of the perceived target is associated and mapped with a pre-stored high-precision map to identify whether there is a faulty vehicle among the perceived targets; In step S104, when there is a faulty vehicle, it is determined whether the autonomous vehicle needs to avoid it based on the roadside perception data and the status of the autonomous vehicle. In step S105, when the autonomous vehicle needs to avoid a collision, a driving trajectory is planned for the autonomous vehicle to avoid the malfunctioning vehicle.

[0027] In this application, roadside sensing data of the entire intersection area is collected in real time by deploying at least one roadside sensing unit at the intersection. Specifically, raw sensing data is collected by roadside sensing units set up in multiple directions at the intersection, and the raw sensing data is transmitted to the roadside or edge computing unit in real time. For example, sensing units (such as cameras, millimeter-wave radars) set up in multiple directions at the intersection, such as the four corners of the intersection, the central median, traffic light poles, etc., ensure that the coverage of the sensing units has no overlapping blind spots, and achieve full sensing coverage of the intersection area to collect raw sensing data of the entire intersection area, including video data, point cloud data, etc. Then, the raw sensing data is processed by a preset sensing model in the roadside or edge computing unit to generate roadside sensing data of the entire intersection area. For sensing data from different sources, the data includes the outline, orientation, and speed and position information of the sensing target relative to the sensing unit, and may also include the size of the sensing target and its attitude relative to the sensing unit.

[0028] A multi-source perception fusion algorithm is used to process the roadside perception data to output the attribute information of all perceived targets within the intersection. Addressing the limitations of data collected by a single perception unit (such as camera interference from lighting conditions and radar's inability to identify target types), the multi-source perception fusion algorithm complements, verifies, and fuses data collected by different types of perception units, improving the accuracy and completeness of the perception data and ultimately outputting precise attribute information for all perceived targets within the intersection. First, data preprocessing: The raw data collected by each perception unit is filtered and denoised to remove invalid and abnormal data, such as noise in camera images and interference echoes in radar data. Second, data association and matching: Data on the same target collected by different perception units is associated and matched, for example, matching vehicle appearance features identified by a camera with vehicle position and speed data sensed by radar to determine different dimensions of the same target. Then, data fusion is performed: Commonly used fusion algorithms include probability-based Bayesian fusion algorithms, optimization-based Kalman filter fusion algorithms, and deep learning-based end-to-end fusion algorithms. The matched data is then weighted and fused using the algorithm to output comprehensive attribute information of the target. Finally, the attribute information is output: The final output of the perceived target attribute information includes the target's precise location, real-time speed, driving direction, target type, vehicle license plate number (if required), pedestrian behavior intention (such as whether they are preparing to cross the road), etc., providing accurate basis for subsequent traffic decisions and control.

[0029] The attribute information of the perceived target is associated and mapped with a pre-stored high-precision map to identify whether a malfunctioning vehicle exists among the perceived targets. By projecting the attribute information of the perceived target onto the high-precision map, it is determined whether the perceived target is located within an intersection range; when the perceived target is located within an intersection range, it is determined whether the perceived target is a vehicle in an abnormal parking state; when the perceived target is a vehicle in an abnormal parking state, the duration of the vehicle being in an abnormal parking state is monitored; when the duration exceeds a preset time threshold, the vehicle is determined to be a malfunctioning vehicle. Figure 2 This is a flowchart illustrating the identification of faulty vehicles in one embodiment of this application, such as... Figure 2 As shown, firstly, the perceived target is projected onto a high-precision map to filter out the perceived targets located within the intersection area. Then, based on the motion state of the perceived target (e.g., the perceived target speed is less than a preset speed), it is determined whether it is an abnormally parked vehicle. Finally, the duration of the abnormally parked vehicle's state is monitored. If its stationary state duration exceeds a preset threshold (e.g., 20 seconds), the system classifies it as a "malfunctioning vehicle" and records its detailed information. In a preferred embodiment, the system can also analyze the congestion impact of the malfunctioning vehicle on traffic flow in all directions at the intersection in real time and dynamically adjust the judgment threshold. For example, during peak traffic hours, the time threshold can be shortened by 20%-30%, or the threshold can be shortened to a preset value, to initiate the avoidance process more quickly. Furthermore, when a sudden traffic event (such as an accident or construction) occurs at the intersection, an emergency mode is automatically triggered, adjusting the time threshold to the preset minimum threshold to prioritize the safety of traffic at the intersection.

[0030] When a malfunctioning vehicle is present, the system determines whether the autonomous vehicle needs to yield based on the roadside perception data and the status of the autonomous vehicle. Once the system identifies the malfunctioning vehicle, it will make a decision based on the real-time status (position, speed, planned path) of the autonomous vehicle about to pass through the intersection. Figure 3 This is a flowchart illustrating the decision-making process for whether to yield to a disabled vehicle in one embodiment of this application. Figure 3As shown, firstly, it is determined whether the real-time location and affected area of ​​the malfunctioning vehicle occupy the preset driving path of the autonomous vehicle (i.e., the original route). If it occupies the preset driving path of the autonomous vehicle, it is determined whether the road section in front of the malfunctioning vehicle has normal traffic conditions, i.e., whether the avoidance operation is feasible. If the road section in front of the malfunctioning vehicle has normal traffic conditions, the avoidance decision is determined to require avoiding the malfunctioning vehicle. If the malfunctioning vehicle does not occupy the preset driving path of the autonomous vehicle, it may be that the malfunctioning vehicle has deviated from the original route and completed the avoidance, turning, or stopping in a safe area. In this case, there is no need to perform the avoidance operation, and traffic can be restored after it completes the subsequent processing, avoiding unnecessary operation to interfere with normal traffic flow. If the road section in front is not passable, passing means that there may be risk factors such as obstacles, construction, accident scenes, and congested traffic. At this time, rashly performing the avoidance operation will not only fail to pass smoothly, but may also cause secondary accidents such as rear-end collisions and scrapes. In this case, it directly enters the "waiting for passage or prompting takeover" stage, and processing is carried out after the road traffic conditions are restored or the malfunctioning vehicle completes takeover.

[0031] The determination of whether the road section in front of the disabled vehicle has normal traffic conditions also includes comprehensively assessing the traffic conditions of the road section by combining lane occupancy information, traffic light status, and congestion at upstream and downstream intersections in the high-precision map; when there is temporary road construction in front of the disabled vehicle, priority is given to choosing an avoidance route that does not conflict with the construction area, and if there is no feasible route, a waiting or takeover prompt is triggered.

[0032] When the autonomous vehicle needs to avoid a malfunctioning vehicle, a driving trajectory to avoid the malfunctioning vehicle is planned. Once avoidance is determined, the system initiates path planning. Specifically, based on the real-time location and impact range of the malfunctioning vehicle, and combined with the autonomous vehicle's safety redundancy distance, the required avoidance space is calculated. For example, based on the malfunctioning vehicle's outline and location, combined with the vehicle's dimensions and safety distance, the required avoidance space is calculated. Outside the determined avoidance space, multiple candidate translation paths are generated along a direction perpendicular to the autonomous vehicle's original driving path. Each generated candidate translation path is smoothly connected to the autonomous vehicle's original driving path to form multiple complete alternative avoidance paths. The final avoidance trajectory is selected from these multiple complete alternative avoidance paths. A comprehensive evaluation is performed on each of the formed alternative avoidance paths, where the evaluation factors include at least: the path's driving safety and the switching cost between the path and the autonomous vehicle's original planned path. From the evaluation results of the comprehensive evaluation, the alternative path with the highest safety and the lowest switching cost is selected as the final avoidance trajectory.

[0033] Figure 4 This is a flowchart of generating an avoidance path in one embodiment of this application, such as... Figure 4 As shown, the original driving path of the autonomous vehicle can be first translated at equal intervals outside the space where it needs to avoid obstacles, generating several parallel "candidate translation paths". Then, using a curve fitting algorithm, the start and end points of each candidate translation path are smoothly and continuously connected to the original path, forming multiple complete "alternative avoidance paths". Each alternative path is then quantitatively evaluated, with evaluation indicators mainly including: the safe distance between the path and other surrounding dynamic and static obstacles (safety), and the comfort and efficiency losses caused by changes in path curvature (switching cost). The system selects the path with the highest comprehensive score, i.e., the safest and least costly path, as the final avoidance trajectory and issues it to the autonomous vehicle for execution.

[0034] In addition, during the operation of autonomous vehicles, the status changes of malfunctioning vehicles can be monitored in real time (such as whether they are started or towed away). When the status of a malfunctioning vehicle changes, the avoidance trajectory is dynamically corrected. When an autonomous vehicle encounters a new obstacle (such as a pedestrian or non-motorized vehicle) during its operation, the avoidance path is replanned based on real-time perception data to ensure driving safety.

[0035] When multiple autonomous vehicles are present at an intersection, the location, status, and avoidance intentions of the malfunctioning vehicle are shared through vehicle-to-infrastructure (V2I) communication technology. Based on the V2I communication data, conflict-free collaborative avoidance paths are planned for multiple autonomous vehicles to prevent new traffic conflicts between vehicles.

[0036] In the process of target recognition, this application uses roadside sensors to continuously and efficiently monitor the conditions within the intersection, enabling faster and more effective identification of malfunctioning vehicles within the intersection; in the decision-making process, roadside perception data is used to more comprehensively assess the surrounding environment and generate the globally optimal avoidance route.

[0037] The beneficial effects of this application are as follows: Roadside perception data of the entire intersection area is collected in real time by deploying at least one roadside perception unit at the intersection; a multi-source perception fusion algorithm is used to process the roadside perception data to output attribute information of all perceived targets within the intersection; the attribute information of the perceived targets is associated and mapped with a pre-stored high-precision map to identify whether a malfunctioning vehicle exists among the perceived targets; when a malfunctioning vehicle exists, the autonomous vehicle is judged whether it needs to avoid the malfunctioning vehicle based on the roadside perception data and the state of the autonomous vehicle; when the autonomous vehicle needs to avoid the malfunctioning vehicle, a driving trajectory to avoid the malfunctioning vehicle is planned for the autonomous vehicle. This solution utilizes sensors deployed at the intersection to collect roadside perception data of the entire intersection area, avoiding blind spots caused by relying solely on data collected from vehicle-mounted sensors. Furthermore, by utilizing the global perception data provided by the roadside, it is possible to effectively identify malfunctioning vehicles that have been parked for a long time and make more comprehensive avoidance decisions, planning a globally better and safer avoidance trajectory, thereby effectively improving vehicle driving safety.

[0038] In one embodiment, step S101 above can be implemented as steps A1-A2 as follows: In step A1, raw sensing data is collected by roadside sensing units set up in multiple directions at the intersection, and the raw sensing data is transmitted to the roadside or edge computing unit in real time. In step A2, the original sensing data is processed by a preset sensing model in the roadside or edge computing unit to generate roadside sensing data for the entire intersection area. The roadside sensing data includes at least the outline, orientation, and speed and position information of the sensing target relative to the sensing unit.

[0039] In one embodiment, step S103 above can be implemented as steps B1-B4 as follows: In step B1, the attribute information of the perceived target is projected onto the high-precision map to determine whether the perceived target is located within the intersection area; In step B2, when the sensing target is located within the intersection area, it is determined whether the sensing target is a vehicle in an abnormal parking state; In step B3, when the sensing target is a vehicle in an abnormal parking state, the duration of the vehicle being in an abnormal parking state is monitored. In step B4, when the duration exceeds a preset time threshold, the vehicle is determined to be a faulty vehicle.

[0040] In one embodiment, the method may also be implemented as follows: C1-C2: In step C1, the risk level of the impact of the vehicle in an abnormal parking state on the traffic flow at the intersection is assessed in real time or periodically based on the roadside sensing data. In step C2, the preset time threshold is dynamically adjusted based on the degree of risk of impact on traffic flow at the intersection.

[0041] In one embodiment, step S104 above can be implemented as steps D1-D3 as follows: In step D1, it is determined whether the real-time location and impact range of the faulty vehicle occupy the preset driving path of the autonomous driving vehicle; In step D2, when the preset driving path of the autonomous vehicle is occupied, it is determined whether the road section in front of the malfunctioning vehicle has normal traffic conditions. In step D3, when the road section in front of the disabled vehicle has normal traffic conditions, the avoidance decision is determined to be that the disabled vehicle needs to be avoided.

[0042] In one embodiment, step S105 above can be implemented as steps E1-E4 as follows: In step E1, based on the real-time location and impact range of the faulty vehicle, and combined with the safety redundancy distance of the autonomous vehicle, the spatial range that needs to be avoided is calculated. In step E2, outside the determined avoidance space, multiple candidate translation paths are generated along a direction perpendicular to the original driving path of the autonomous vehicle. In step E3, each generated candidate translation path is smoothly connected to the original driving path of the autonomous vehicle to form multiple complete alternative avoidance paths. In step E4, the final avoidance trajectory is selected from the multiple complete alternative avoidance paths.

[0043] In one embodiment, step E4 above can be implemented as steps E41-E42: In step E41, a comprehensive evaluation is performed on each of the generated alternative avoidance paths. The evaluation factors for the comprehensive evaluation include at least: the driving safety of the path and the switching cost between the path and the original path of the autonomous vehicle. In step E42, the alternative path with the highest safety and the lowest switching cost is selected from the comprehensive evaluation results as the final avoidance driving trajectory.

[0044] Figure 5 This is a schematic diagram of a device for avoiding a disabled vehicle according to an embodiment of this application, as shown below. Figure 5 As shown, it includes: The data acquisition module 501 is used to collect roadside sensing data of the entire intersection area in real time through at least one roadside sensing unit deployed at the intersection. The processing module 502 is used to process the roadside sensing data using a multi-source sensing fusion algorithm to output the attribute information of all sensing targets within the intersection. The identification module 503 is used to associate and map the attribute information of the perceived target with a pre-stored high-precision map in order to identify whether there is a faulty vehicle in the perceived target. The judgment module 504 is used to determine whether the autonomous vehicle needs to avoid a faulty vehicle based on the roadside perception data and the state of the autonomous vehicle when a faulty vehicle is present. The planning module 505 is used to plan a driving trajectory for the autonomous vehicle to avoid the malfunctioning vehicle when the autonomous vehicle needs to avoid it.

[0045] In one embodiment, the acquisition module includes: The data acquisition submodule is used to collect raw sensing data by collecting roadside sensing units set up in multiple directions at the intersection, and to transmit the raw sensing data to the roadside or edge computing unit in real time. The processing submodule is used to process the original sensing data through a preset sensing model in the roadside or edge computing unit to generate roadside sensing data for the entire intersection area. The roadside sensing data includes at least the outline, orientation, and speed and position information of the sensing target relative to the sensing unit.

[0046] In one embodiment, the identification module includes: The first judgment submodule is used to project the attribute information of the perceived target onto the high-precision map to determine whether the perceived target is located within the intersection area; The determination submodule is used to determine whether the perceived target is a vehicle in an abnormal parking state when the perceived target is located within the intersection area; The monitoring submodule is used to monitor the duration of the abnormal parking state of the vehicle when the sensing target is a vehicle in an abnormal parking state. The first determining submodule is used to determine that the vehicle is a faulty vehicle when the duration exceeds a preset time threshold.

[0047] In one embodiment, the apparatus further includes: The assessment module is used to assess the risk level of the impact of vehicles in abnormal parking states on traffic flow at intersections in real time or periodically based on the roadside perception data. The adjustment module is used to dynamically adjust the preset time threshold based on the degree of risk of impact on traffic flow at the intersection.

[0048] In one embodiment, the determining module includes: The second judgment submodule is used to determine whether the real-time location and impact range of the faulty vehicle occupy the preset driving path of the autonomous driving vehicle. The third judgment submodule is used to determine whether the road section in front of the faulty vehicle has normal passage conditions when the preset driving path of the autonomous vehicle is occupied. The second determining submodule is used to determine that the avoidance decision is to avoid the disabled vehicle when the road section in front of the disabled vehicle has normal traffic conditions.

[0049] In one embodiment, the planning module includes: The calculation submodule is used to calculate the space range that needs to be avoided based on the real-time location and impact range of the faulty vehicle, combined with the safety redundancy distance of the autonomous vehicle. The generation submodule is used to generate multiple candidate translation paths outside the defined avoidance space, along a direction perpendicular to the original driving path of the autonomous vehicle. The smoothing submodule is used to smoothly connect each generated candidate translation path with the original driving path of the autonomous vehicle to form multiple complete alternative avoidance paths. The selection submodule is used to select the final avoidance driving trajectory from the multiple complete alternative avoidance paths.

[0050] In one embodiment, the selection submodule is further configured to: A comprehensive evaluation is conducted on each of the generated alternative avoidance paths. The evaluation factors for the comprehensive evaluation include at least the driving safety of the path and the switching cost between the path and the original path of the autonomous vehicle. From the comprehensive evaluation results, the alternative path with the highest safety and the lowest switching cost is selected as the final avoidance driving trajectory.

[0051] Figure 6 This is a schematic diagram of the hardware structure of a system for avoiding a disabled vehicle according to an embodiment of this application, as shown below. Figure 6 As shown, the system for avoiding a disabled vehicle includes: At least one processor 620; and, Memory 604 communicatively connected to the at least one processor 620; wherein, The memory 604 stores instructions that can be executed by the at least one processor 620 to implement the method for avoiding a disabled vehicle as described in any of the above embodiments.

[0052] Reference Figure 6The system 600 for avoiding a disabled vehicle may include one or more of the following components: a processing component 602, a memory 604, a power supply component 606, an input / output (I / O) interface 608, a sensor component 610, and a communication component 612.

[0053] Processing component 602 typically controls the overall operation of system 600 for avoiding a disabled vehicle. Processing component 602 may include one or more processors 620 to execute instructions to complete all or part of the steps of the method described above. Furthermore, processing component 602 may include one or more modules to facilitate interaction between processing component 602 and other components. The processor 620 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0054] Memory 604 is configured to store various types of data to support the operation of system 600 for avoiding disabled vehicles. Examples of this data include instructions for any application or method operating on system 600 for avoiding disabled vehicles. Memory 604 may be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. Memory 604 may also be an external storage device of the terminal device, such as a plug-in hard disk equipped on the terminal device. Memory 604 may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. Memory 604 is used to store programs and data required by this application. Memory 604 may also be used to temporarily store data that has been output or will be output.

[0055] Power supply component 606 provides power to various components of system 600 for avoiding a disabled vehicle. Power supply component 606 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to system 600 for avoiding a disabled vehicle.

[0056] I / O interface 608 provides an interface between processing component 602 and peripheral interface modules, such as keyboards, click wheels, buttons, etc.

[0057] Sensor assembly 610 includes one or more sensors for providing various aspects of the status assessment of the system 600 for avoiding a disabled vehicle. Additionally, sensor assembly 610 can detect the on / off state of the system 600 for avoiding a disabled vehicle, the relative positioning of components, and the operational status of the system 600 or a component of the system 600 for avoiding a disabled vehicle. In some embodiments, sensor assembly 610 may include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor, etc.

[0058] Communication component 612 is configured to enable system 600 for avoiding disabled vehicles to provide wired or wireless communication capabilities with other devices and cloud platforms. System 600 for avoiding disabled vehicles can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 616 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 616 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0059] In an exemplary embodiment, the system 600 for avoiding a disabled vehicle may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the method for avoiding a disabled vehicle as described in any of the above embodiments.

[0060] This application also provides a computer-readable storage medium that, when the instructions in the storage medium are executed by a processor corresponding to a system for avoiding a disabled vehicle, enables the system for avoiding a disabled vehicle to implement the method for avoiding a disabled vehicle described in any of the above embodiments.

[0061] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0062] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0063] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0064] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0065] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method of avoiding a disabled vehicle, characterized by, include: Real-time roadside sensing data of the entire intersection area is collected by deploying at least one roadside sensing unit at the intersection. The roadside sensing data is processed using a multi-source sensing fusion algorithm to output the attribute information of all sensing targets within the intersection. The attribute information of the perceived target is associated and mapped with a pre-stored high-precision map to identify whether there is a faulty vehicle among the perceived targets. When a malfunctioning vehicle is present, the autonomous vehicle is determined to avoid the obstacle based on the roadside perception data and the status of the autonomous vehicle. When the autonomous vehicle needs to avoid a collision, a driving trajectory is planned for the autonomous vehicle to avoid the malfunctioning vehicle.

2. The method of claim 1, wherein, The method of collecting roadside sensing data of the entire intersection area in real time through at least one roadside sensing unit deployed at the intersection includes: Raw sensing data is collected by roadside sensing units set up in multiple directions at intersections, and the raw sensing data is transmitted to roadside or edge computing units in real time. The original sensing data is processed by a preset sensing model in the roadside or edge computing unit to generate roadside sensing data for the entire intersection area. The roadside sensing data includes at least the outline, orientation, and speed and position information of the sensing target relative to the sensing unit.

3. The method of claim 1, wherein, The step of associating and mapping the attribute information of the perceived target with a pre-stored high-precision map to identify whether there is a faulty vehicle among the perceived targets includes: The attribute information of the perceived target is projected onto the high-precision map to determine whether the perceived target is located within the intersection area; When the perceived target is located within the intersection area, it is determined whether the perceived target is a vehicle in an abnormal parking state; When the sensing target is a vehicle in an abnormal parking state, the duration of the vehicle being in an abnormal parking state is monitored. When the duration exceeds a preset time threshold, the vehicle is determined to be a faulty vehicle.

4. The method of claim 3, wherein, The method further includes: The impact risk of vehicles in abnormal parking states on traffic flow at intersections is assessed in real time or periodically based on the roadside sensing data. The preset time threshold is dynamically adjusted based on the degree of risk of impact on traffic flow at the intersection.

5. The method of claim 1, wherein, The step of determining whether the autonomous vehicle needs to avoid a collision based on the roadside perception data and the state of the autonomous vehicle includes: Determine whether the real-time location and affected area of ​​the faulty vehicle occupy the preset driving path of the autonomous vehicle; When the preset driving path of an autonomous vehicle is occupied, determine whether the road section in front of the malfunctioning vehicle has normal traffic conditions. When the road section in front of the disabled vehicle is passable, the avoidance decision is determined to be that the disabled vehicle needs to be avoided.

6. The method of claim 1, wherein, Planning a driving trajectory for the autonomous vehicle to avoid the malfunctioning vehicle includes: Based on the real-time location and impact range of the faulty vehicle, and combined with the safety redundancy distance of the autonomous vehicle, the spatial range that needs to be avoided is calculated. Outside the defined avoidance space, multiple candidate translation paths are generated along a direction perpendicular to the original driving path of the autonomous vehicle. Each generated candidate translation path is smoothly connected to the original driving path of the autonomous vehicle to form multiple complete alternative avoidance paths. The final avoidance trajectory is selected from the multiple complete alternative avoidance paths.

7. The method of claim 6, wherein, The step of selecting the final avoidance trajectory from the multiple complete alternative avoidance paths includes: A comprehensive evaluation is conducted on each of the generated alternative avoidance paths. The evaluation factors for the comprehensive evaluation include at least the driving safety of the path and the switching cost between the path and the original path of the autonomous vehicle. From the comprehensive evaluation results, the alternative path with the highest safety and the lowest switching cost is selected as the final avoidance driving trajectory.

8. An apparatus for avoiding a disabled vehicle, characterized by include: The data acquisition module is used to collect roadside sensing data of the entire intersection area in real time through at least one roadside sensing unit deployed at the intersection. The processing module is used to process the roadside sensing data using a multi-source sensing fusion algorithm to output the attribute information of all sensing targets within the intersection. The identification module is used to associate and map the attribute information of the perceived target with a pre-stored high-precision map in order to identify whether there is a faulty vehicle among the perceived targets; The judgment module is used to determine whether the autonomous vehicle needs to avoid a faulty vehicle based on the roadside perception data and the state of the autonomous vehicle. The planning module is used to plan a driving trajectory for the autonomous vehicle to avoid the malfunctioning vehicle when the autonomous vehicle needs to avoid it.

9. A system for avoiding a disabled vehicle, the system comprising: include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor to implement the method of avoiding a disabled vehicle as described in any one of claims 1-7.

10. A computer readable storage medium characterized by, When the instructions in the storage medium are executed by the processor corresponding to the system for avoiding a disabled vehicle, the system for avoiding a disabled vehicle is able to implement the method for avoiding a disabled vehicle as described in any one of claims 1-7.

Images