Collision risk detection method, apparatus and device

Abstract

The application discloses a collision risk detection method, device and equipment. The method comprises the following steps: identifying an obstacle on a driving road of a vehicle; predicting a first future motion track of the obstacle; acquiring a second future motion track of the vehicle; and acquiring a risk value of collision between the vehicle and the obstacle according to the first future motion track and the second future motion track. By using the processing mode, the collision risk of the vehicle driving on the road is quantified by a probability method, so that the method is not dependent on map modeling, and a safety officer is not required to take over forcibly. As long as the safety officer gives feedback that the warning has been received in a specific case, the automatic driving behavior is not changed in the case of receiving the feedback, and excessive human resources are avoided.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, specifically to collision risk detection methods and devices, and vehicles. Background Technology

[0002] Autonomous vehicles, also known as driverless cars, computer-driven cars, or wheeled mobile robots, are intelligent vehicles that achieve driverless operation through computer systems. They have been widely adopted and applied in industries such as instant delivery and express delivery services.

[0003] Collision risk detection is a fundamental requirement for autonomous driving. Collision risk detection technology can assess the risk of collisions between autonomous vehicles and other road users in real time. Adjusting the control of autonomous driving based on the collision risk detection results contributes to the safety of autonomous driving systems. Currently, a common vehicle collision risk detection method uses Principal Component Analysis (PCA) and Support Vector Machines for risk modeling. Another common method is to generate probability distributions based on prior knowledge of the map environment.

[0004] However, in the process of realizing this invention, the inventors found that the above solution has at least the following problems: 1) The method of generating probability distribution based on prior knowledge of the map environment depends on map modeling when calculating risk value; 2) The method of risk modeling using principal component analysis (PCA) and support vector machine requires mandatory takeover by safety personnel, which cannot meet the requirements of L4 autonomous driving standards.

[0005] In summary, existing technologies cannot simultaneously achieve the advantages of not relying on map modeling and not requiring safety personnel to forcibly take over. Summary of the Invention

[0006] This application provides a collision risk detection method to address the shortcomings of existing technologies that fail to simultaneously achieve the advantages of not relying on map modeling and not requiring mandatory intervention by a safety operator. This application also provides a collision risk detection device and electronic equipment.

[0007] This application provides a collision risk detection method, including:

[0008] Identify obstacles on the vehicle's path;

[0009] Predict the first trajectory of an obstacle in the future time period;

[0010] Obtain the vehicle's second trajectory in the future time period;

[0011] Based on the first and second motion trajectories, the risk value of a collision between the vehicle and an obstacle in the future time period is obtained.

[0012] Optionally, the obstacle includes a dynamic obstacle;

[0013] The prediction of the obstacle's first trajectory in a future time period is made using at least one of the following methods:

[0014] Based on the pedestrian's historical movement trajectory, predict the pedestrian's first movement trajectory in the future time period;

[0015] Based on the vehicle's historical trajectory, predict the vehicle's first trajectory in the future time period;

[0016] Based on the historical movement trajectory of non-motorized vehicles, predict the first movement trajectory of non-motorized vehicles in the future time period.

[0017] Optional, also includes:

[0018] Environmental perception information collected from the space surrounding the vehicle;

[0019] Based on environmental perception information of the space surrounding the vehicle, the state type of the obstacle is determined, including dynamic obstacles or static obstacles.

[0020] Optionally, obtaining the second motion trajectory of the vehicle in a future time period includes:

[0021] Based on the vehicle's driving route planning information, obtain the vehicle's second trajectory in the future time period.

[0022] Optionally, obtaining the risk value of a collision between the vehicle and an obstacle in a future time period based on the first and second motion trajectories includes:

[0023] Based on the first and second motion trajectories, obtain the risk value of collision between the vehicle and the obstacle at various points in time within a future time period;

[0024] Based on the risk value of collisions at various points in time, obtain the risk value of collisions between the vehicle and obstacles in future time periods.

[0025] Optionally, the step of obtaining the risk value of a collision between the vehicle and the obstacle at various points in time within a future time period based on the first and second motion trajectories includes:

[0026] Based on the first trajectory, determine the location of the obstacle at various points in time within the future time period;

[0027] Based on the second trajectory, the vehicle's position information at various points in time within a future period can be determined;

[0028] For each time point, based on the range information of the obstacle at that time point and the position information of the vehicle at that time point, the risk value of the collision between the vehicle and the obstacle at that time point is obtained.

[0029] Optionally, obtaining the risk value of a collision between the vehicle and the obstacle at that time point based on the obstacle's range information and the vehicle's position information at that time point includes:

[0030] Based on the range information at the stated time point, sample the location information of multiple obstacles;

[0031] For each obstacle location, based on the obstacle location information and the vehicle's location information, the number of collision points between the vehicle and the obstacle is obtained;

[0032] Based on the number of obstacle location samples and the number of collision points, the risk value of a collision between the vehicle and an obstacle at the specified time point is obtained.

[0033] Optionally, the first motion trajectory includes: the location information of the obstacle at a given time point, and the location distribution covariance information;

[0034] The step of obtaining the risk value of collision between the vehicle and the obstacle at various points in time within a future period based on the first and second motion trajectories includes:

[0035] For any given time point, the Monte Carlo algorithm is used to sample the location of the obstacle at that time point based on the covariance and location information.

[0036] Based on the second trajectory, the vehicle's position information at various points in time within a future period can be determined;

[0037] For each sampling location of the obstacle, based on the obstacle location information and the vehicle's location information, the number of collision points between the vehicle and the obstacle is obtained;

[0038] Based on the number of obstacle location samples and the number of collision points, the risk value of a collision between the vehicle and an obstacle at the specified time point is obtained.

[0039] Optional, also includes:

[0040] Based on the risk value and at least one risk treatment threshold, a risk treatment strategy is executed.

[0041] Optionally, the risk handling strategy is executed based on the risk value and at least one risk alarm threshold, employing at least one of the following methods:

[0042] If the risk value is less than the first risk handling threshold, the vehicle will continue to drive.

[0043] If the risk value is greater than the first risk handling threshold and less than the second risk handling threshold, the vehicle will continue driving and the risk value will be displayed.

[0044] If the risk value is greater than the second risk handling threshold and less than the third risk handling threshold, the risk value is displayed; if feedback information is received, the vehicle is controlled to continue driving; if no feedback information is received, the vehicle is controlled to stop driving.

[0045] This application also provides a collision detection device, including:

[0046] An obstacle recognition unit is used to identify obstacles on the road where the vehicle is traveling.

[0047] An obstacle trajectory prediction unit is used to predict the first motion trajectory of an obstacle in a future time period.

[0048] The vehicle trajectory acquisition unit is used to acquire the second motion trajectory of the vehicle in a future time period;

[0049] The risk determination unit is used to obtain the risk value of a collision between the vehicle and an obstacle in the future time period based on the first motion trajectory and the second motion trajectory.

[0050] This application also provides electronic devices, including:

[0051] This includes: the processor, and the memory;

[0052] The memory is used to store a program that implements the collision risk detection method described above. The program of the method is executed by the processor when the device is powered on.

[0053] This application also provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the various methods described above.

[0054] This application also provides a computer program product including instructions that, when run on a computer, cause the computer to perform the various methods described above.

[0055] Compared with the prior art, this application has the following advantages:

[0056] The collision risk detection method provided in this application identifies obstacles on the vehicle's driving path; predicts the first future trajectory of the obstacle; obtains the second future trajectory of the vehicle; and obtains a collision risk value between the vehicle and the obstacle based on the first and second future trajectories. This approach quantifies the collision risk of the vehicle on the road using a probabilistic method, thus avoiding reliance on map modeling and the need for mandatory intervention by a safety driver. Only under specific circumstances is the safety driver provided with confirmation that a warning has been received; upon receiving this feedback, the autonomous driving behavior remains unchanged, avoiding excessive human intervention. Attached Figure Description

[0057] Figure 1 This is a flowchart of an embodiment of the collision risk detection method provided in this application;

[0058] Figure 2 This is a schematic diagram of a scenario representing an embodiment of the collision risk detection method provided in this application;

[0059] Figure 3 This is a schematic diagram of the collision probability of an embodiment of the collision risk detection method provided in this application. Detailed Implementation

[0060] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0061] This application provides a collision detection method and apparatus, a collision detection model construction method and apparatus, a vehicle, and an electronic device. The various solutions are described in detail in the following embodiments.

[0062] First Embodiment

[0063] Please refer to Figure 1 This is a flowchart illustrating an embodiment of a collision detection method provided in this application. The subject executing this method includes, but is not limited to, autonomous vehicles, such as unmanned delivery vehicles. The collision detection method provided in this application includes:

[0064] Step S101: Identify obstacles on the vehicle's driving path.

[0065] During autonomous driving, the vehicle (the one performing collision risk detection) interacts with various road users, who are considered obstacles. These obstacles include, but are not limited to, bicycles, cars, pedestrians, etc. To identify obstacles in the vehicle's path, readily available and mature technologies can be employed, such as machine learning-based object recognition algorithms, which will not be elaborated upon here.

[0066] like Figure 2 As shown, a vehicle is traveling from south to north, while a road participant 1 is traveling from west to east. As both the vehicle and the road participant move, they are in different positions at different times. At time tn, the vehicle collides with road participant 1.

[0067] The method provided in this application can detect the risk of a collision between a vehicle and dynamic or static obstacles traveling on the road. Dynamic obstacles can be vehicles in motion, pedestrians, etc., while static obstacles can be vehicles involved in accidents on the road, etc.

[0068] In one example, the method may further include the following steps: collecting environmental perception information of the space surrounding the vehicle; and determining the state type of the obstacle, such as a dynamic obstacle or a static obstacle, based on the environmental perception information of the space surrounding the vehicle.

[0069] In practice, the system can obtain the type (vehicle, person, others, etc.), location, and speed of obstacles based on environmental perception information of the space surrounding the vehicle. The state type of the obstacle can be determined through feature analysis (vehicle lights, vehicle speed, other surrounding obstacles, etc., or by using deep learning-based pattern recognition). For example, if the obstacle is a temporary stop due to an accident, it is a static obstacle; if the obstacle has a certain speed, it is a dynamic obstacle.

[0070] Step S103: Predict the first trajectory of the obstacle in the future time period.

[0071] Using the method provided in this application, an autonomous vehicle can determine the collision risk with road users in real time during operation and quantify the collision risk as a probability value. To obtain the risk value of a collision between the vehicle and an obstacle, it is necessary to first obtain the motion trajectories of each obstacle and the vehicle in the future time period. For ease of description, this embodiment refers to the motion trajectory of the obstacle in the future time period as the first motion trajectory and the motion trajectory of the vehicle in the future time period as the second motion trajectory.

[0072] In one example, the obstacle is a dynamic obstacle; step S103 can be implemented as follows: the prediction module predicts the motion trajectory of the dynamic obstacle over a future period of time based on the historical motion trajectories of various objects. The motion trajectory is represented by a sequence of position points and their corresponding covariance. For example, the pedestrian trajectory prediction module predicts the pedestrian's motion trajectory over a future period of time based on the pedestrian's historical motion trajectory; the vehicle trajectory prediction module predicts the vehicle's motion trajectory over a future period of time based on the vehicle's historical motion trajectory; and the non-motorized vehicle's first motion trajectory over a future period of time is predicted based on the non-motorized vehicle's historical motion trajectory.

[0073] In practical implementation, the trajectory of any moving obstacle over a future period of time (e.g., 5 seconds) can be probabilistically modeled, represented by a sequence of n points: (x0, y0, θ0), (x1, y1, θ1), ..., (xt, yt, θt), ..., (xn, yn, θn). Here, the subscripts represent time points, (x, y) represent the position of the obstacle at each time point, θ represents the covariance at that time, and (xt, yt, θt) represents the range of the obstacle at time t. In practice, one trajectory point can be generated every second or every 0.1 seconds.

[0074] In another example, the obstacle is a static obstacle, and the first movement trajectory of the static obstacle in a future time period is the same location.

[0075] Step S105: Obtain the vehicle's second motion trajectory in the future time period.

[0076] In this embodiment, the vehicle's driving route planning module can be used to obtain the vehicle's driving route planning information; based on the vehicle's driving route planning information, the vehicle's second motion trajectory in the future time period can be obtained.

[0077] In practice, since the future pose of the vehicle depends on the route planning information, and the vehicle's route is usually highly certain, the vehicle's position at each moment is a specific location rather than a range, meaning that the covariance of its future pose does not need to be considered.

[0078] Step S107: Based on the first and second motion trajectories, obtain the risk value of a collision between the vehicle and an obstacle in the future time period.

[0079] For each obstacle, the risk of a collision between the vehicle and the obstacle in the future time period can be determined based on the trajectory of the obstacle and the vehicle in the future time period. For example, the risk of the vehicle colliding with vehicle 1 in the next 5 seconds is 30%, and the risk of colliding with vehicle 2 in the next 5 seconds is 55%.

[0080] In one example, step S107 may include the following sub-steps:

[0081] Step S1071: Based on the first and second motion trajectories, obtain the risk value of collision between the vehicle and the obstacle at various points in time within the future time period.

[0082] In one example, step S1071 may include the following sub-steps:

[0083] Step S1071-1A: Based on the first motion trajectory, the range information of the obstacle at various points in time within the future time period can be determined.

[0084] Step S1071-2A: Based on the second motion trajectory, the location information of the vehicle at various points in time within the future time period can be determined.

[0085] Step S1071-3A: For each time point, based on the range information of the obstacle at that time point and the position information of the vehicle at that time point, obtain the risk value of the collision between the vehicle and the obstacle at that time point.

[0086] In one example, step S1071-3A may include the following sub-steps: 1) sampling multiple obstacle location information based on the range information at the time point, such as sampling 100 locations; 2) for each obstacle location, obtaining the number of collision points between the vehicle and the obstacle based on the obstacle location information and the vehicle's location information, such as 60 collision points; 3) obtaining the risk value of the collision between the vehicle and the obstacle at the time point based on the obstacle location sampling number information and the number of collision points, such as 60 / 100 = 60%, i.e., the risk probability is 60%.

[0087] In another example, a probabilistic model is performed on the trajectory of any moving obstacle in the next n seconds. The first trajectory includes: the obstacle's position information at a given time point, and the position distribution covariance information, i.e., (xt, yt, θt) mentioned above. Step S1071 may include the following sub-steps:

[0088] Step S1071-1B: For any given time point, using the Monte Carlo algorithm, sample the obstacle positions corresponding to that time point based on the covariance and position information. For example, sample 100 positions that follow a Gaussian distribution.

[0089] Step S1071-2B: Based on the second motion trajectory, the position information of the vehicle at various points in time within the future time period can be determined.

[0090] Step S1071-3B: For each sampling location of the obstacle, based on the obstacle location information and the vehicle's location information, obtain the number of collision points between the vehicle and the obstacle.

[0091] Step S1071-4B: Based on the obstacle location sampling quantity information and the collision location point information, obtain the risk value of the vehicle colliding with the obstacle at the time point.

[0092] This approach allows the Monte Carlo method to sample obstacles at predicted positions xt and yt based on the covariance θt at any time t, and to calculate the number of sampling points that collide with the vehicle's position, effectively improving the accuracy of collision risk detection.

[0093] Step S1073: Based on the risk value of collision at each time point, obtain the risk value of collision between the vehicle and the obstacle in the future time period.

[0094] By combining the risk values ​​of a collision between the vehicle and an obstacle at various points in time, the risk value of a collision between the vehicle and an obstacle in the future can be obtained. For example... Figure 3 As shown, the maximum risk value of a collision at various points in time within a future time period can be used as the risk value of a collision between the vehicle and an obstacle in that future time period. For example, if the probability of the maximum collision risk within 5 seconds is determined to be between 2 and 2.5 seconds, the risk of the vehicle colliding with the obstacle is approximately 70%. Using this approach can effectively improve the accuracy of collision risk detection.

[0095] In one example, the method may further include the following steps:

[0096] Step S201: Execute a risk management strategy based on the risk value and at least one risk management threshold.

[0097] Different levels of warnings are issued to the autonomous driving system and the autonomous driving safety driver based on collision risk values. At different risk thresholds, strategies such as deceleration, stopping, and requesting human intervention can be implemented to improve the safety of the autonomous human-machine co-driving system.

[0098] In one example, step S201 may take at least one of the following approaches: if the risk value is less than a first risk handling threshold, control the vehicle to continue driving; if the risk value is greater than the first risk handling threshold and less than a second risk handling threshold, control the vehicle to continue driving and display the risk value; if the risk value is greater than the second risk handling threshold and less than a third risk handling threshold, display the risk value; if feedback information is received, control the vehicle to continue driving; if no feedback information is received, control the vehicle to stop driving.

[0099] For example, three different thresholds (q1, q2, q3) are set to implement the strategy. Strategy 1: Collision risk probability [0, 0.3], normal driving, no need to alert safety personnel; Strategy 2: Collision risk probability [0.3, 0.6], normal driving, but alerting safety personnel, no feedback required; Strategy 3: Collision risk probability [0.6, 1.0], alerting safety personnel, requiring feedback, normal driving after feedback, forced stopping if no feedback is given.

[0100] As can be seen from the above embodiments, the collision risk detection method provided in this application identifies obstacles on the vehicle's driving road; predicts the first future trajectory of the obstacle; obtains the second future trajectory of the vehicle; and obtains the risk value of a collision between the vehicle and the obstacle based on the first and second future trajectories. This approach quantifies the collision risk of a vehicle driving on the road using a probabilistic method, thus avoiding reliance on map modeling and the need for a safety driver to forcibly take over. Only under specific circumstances is the safety driver given feedback confirming receipt of the warning; upon receiving feedback, the autonomous driving behavior is not altered, avoiding excessive human intervention.

[0101] Second Embodiment

[0102] In the above embodiments, a collision risk detection method was provided. Correspondingly, this application also provides a collision risk detection device. This device corresponds to the embodiments of the above method. Since the device embodiments are basically similar to the method embodiments, the description is relatively simple, and relevant parts can be referred to in the description of the method embodiments. The device embodiments described below are merely illustrative.

[0103] This application also provides a collision risk detection device, comprising: an obstacle recognition unit, an obstacle trajectory prediction unit, a vehicle trajectory acquisition unit, and a risk determination unit. The obstacle recognition unit is used to identify obstacles on the vehicle's driving path; the obstacle trajectory prediction unit is used to predict a first motion trajectory of the obstacle in a future time period; the vehicle trajectory acquisition unit is used to acquire a second motion trajectory of the vehicle in a future time period; and the risk determination unit is used to obtain a risk value for a collision between the vehicle and the obstacle in the future time period based on the first and second motion trajectories.

[0104] In one example, the obstacle includes a dynamic obstacle; the obstacle trajectory prediction unit is specifically used to predict the first movement trajectory of a pedestrian in a future time period based on the pedestrian's historical movement trajectory; predict the first movement trajectory of a vehicle in a future time period based on the vehicle's historical movement trajectory; and predict the first movement trajectory of a non-motorized vehicle in a future time period based on the non-motorized vehicle's historical movement trajectory.

[0105] In one example, the device may further include: an environmental information acquisition unit and an obstacle state type determination unit. The environmental information acquisition unit is used to acquire environmental perception information of the space surrounding the vehicle; the obstacle state type determination unit is used to determine the state type of the obstacle based on the environmental perception information of the space surrounding the vehicle, wherein the state type includes dynamic obstacles or static obstacles.

[0106] In one example, the vehicle trajectory acquisition unit is specifically used to acquire the second motion trajectory of the vehicle in a future time period based on the vehicle's driving route planning information.

[0107] In one example, the risk determination unit includes: a single-point risk determination unit and a multi-point risk processing unit. The single-point risk determination unit is used to obtain the risk value of a collision between the vehicle and an obstacle at various time points within a future time period based on the first motion trajectory and the second motion trajectory; the multi-point risk processing unit is used to obtain the risk value of a collision between the vehicle and an obstacle within a future time period based on the risk value of the collision at each time point.

[0108] In one example, the single-point risk determination unit includes: an obstacle range determination unit, a vehicle position determination unit, and a single-point collision risk acquisition unit.

[0109] The obstacle range determination unit is used to determine the range information of the obstacle at various points in time within a future time period based on the first motion trajectory.

[0110] The vehicle position determination unit is used to determine the position information of the vehicle at various points in time within a future time period based on the second motion trajectory.

[0111] The single-point collision risk acquisition unit is used to acquire the risk value of a collision between the vehicle and the obstacle at each time point, based on the range information of the obstacle at that time point and the position information of the vehicle at that time point.

[0112] In one example, the single-point collision risk acquisition unit is specifically used to sample multiple obstacle location information based on the range information at the time point; for each obstacle location, based on the obstacle location information and the vehicle's location information, obtain the number of collision points between the vehicle and the obstacle; and based on the obstacle location sampling number information and the number of collision points, obtain the risk value of the collision between the vehicle and the obstacle at the time point.

[0113] In one example, the first motion trajectory includes: the location information of the obstacle at a given time point, and the location distribution covariance information; the single-point risk determination unit is specifically used to sample the obstacle position at any given time point using a Monte Carlo algorithm based on the covariance information and location information corresponding to that time point; based on the second motion trajectory, the position information of the vehicle at various time points in the future time period can be determined; for each sampled location of the obstacle, based on the obstacle position information and the vehicle position information, the number of collision points between the vehicle and the obstacle is obtained; based on the obstacle position sampling number information and the number of collision points, the risk value of the collision between the vehicle and the obstacle at that time point is obtained.

[0114] In one example, the apparatus may further include a risk processing unit for executing a risk processing strategy based on the risk value and at least one risk processing threshold.

[0115] In one example, the risk processing unit is specifically configured to: control the vehicle to continue driving if the risk value is less than a first risk processing threshold; control the vehicle to continue driving and display the risk value if the risk value is greater than the first risk processing threshold and less than a second risk processing threshold; display the risk value if the risk value is greater than the second risk processing threshold and less than a third risk processing threshold; control the vehicle to continue driving if feedback information is received; and control the vehicle to stop driving if no feedback information is received.

[0116] Third Embodiment

[0117] In the above embodiments, a collision risk detection method is provided. Correspondingly, this application also provides an electronic device, which includes, but is not limited to, an autonomous vehicle. The embodiments of this device correspond to the embodiments of the above method. Since the device embodiments are basically similar to the method embodiments, the description is relatively simple; relevant details can be found in the descriptions of the method embodiments. The device embodiments described below are merely illustrative.

[0118] This application also provides an electronic device, including a processor and a memory. The memory is used to store a program for implementing the collision risk detection method provided according to the above embodiments, wherein the device is powered on and the program of the method is executed by the processor.

[0119] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

[0120] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0121] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0122] 1. Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include non-transitory computer-readable media, such as modulated data and carrier waves.

[0123] 2. 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 embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

Claims

1. A collision risk detection method, characterized in that, include: Identify obstacles on the vehicle's path; Predict the first trajectory of an obstacle in the future time period; The first motion trajectory includes: the location information of the obstacle at a given time point, and the location distribution covariance information; Obtain the vehicle's second trajectory in the future time period; Based on the first and second motion trajectories, the risk value of a collision between the vehicle and an obstacle in a future time period is obtained, including: determining the position information of the vehicle at various time points within the future time period based on the second motion trajectory; for any given time point, the following processing is performed: using a Monte Carlo algorithm, sampling the obstacle position corresponding to the time point based on the covariance and position information; for each sampled obstacle position corresponding to the time point, obtaining the number of collision points between the vehicle and the obstacle corresponding to the time point based on the obstacle position information and the vehicle position information; and obtaining the risk value of a collision between the vehicle and the obstacle at that time point based on the obstacle position sampling number information and the number of collision points.

2. The method according to claim 1, characterized in that, The obstacles include dynamic obstacles; The prediction of the obstacle's first trajectory in a future time period is made using at least one of the following methods: Based on the historical movement trajectories of pedestrians, predict the first movement trajectory of pedestrians in the future time period; based on the historical movement trajectories of vehicles, predict the first movement trajectory of vehicles in the future time period; based on the historical movement trajectories of non-motorized vehicles, predict the first movement trajectory of non-motorized vehicles in the future time period.

3. The method according to claim 2, characterized in that, Also includes: Environmental perception information collected from the space surrounding the vehicle; Based on environmental perception information of the space surrounding the vehicle, the state type of the obstacle is determined, including dynamic obstacles or static obstacles.

4. The method according to claim 1, characterized in that, The acquisition of the vehicle's second trajectory over a future time period includes: Based on the vehicle's driving route planning information, obtain the vehicle's second trajectory in the future time period.

5. The method according to claim 1, characterized in that, The step of obtaining a risk value for a collision between the vehicle and an obstacle in the future time period based on the first and second motion trajectories includes: Based on the first and second motion trajectories, obtain the risk value of collision between the vehicle and the obstacle at various points in time within a future time period; Based on the risk value of collisions at various points in time, obtain the risk value of collisions between the vehicle and obstacles in future time periods.

6. The method according to claim 5, characterized in that, The step of obtaining the risk value of collision between the vehicle and the obstacle at various points in time within a future period based on the first and second motion trajectories includes: Based on the first trajectory, determine the location of the obstacle at various points in time within the future time period; Based on the second trajectory, the vehicle's position information at various points in time within a future period can be determined; For each time point, based on the range information of the obstacle at that time point and the position information of the vehicle at that time point, the risk value of the collision between the vehicle and the obstacle at that time point is obtained.

7. The method according to claim 6, characterized in that, The step of obtaining a risk value for a collision between the vehicle and the obstacle at a given time point, based on the obstacle's location information and the vehicle's position information at that time point, includes: Based on the range information at the stated time point, sample the location information of multiple obstacles; For each obstacle location, based on the obstacle location information and the vehicle's location information, the number of collision points between the vehicle and the obstacle is obtained; Based on the number of obstacle location samples and the number of collision points, the risk value of a collision between the vehicle and an obstacle at the specified time point is obtained.

8. The method according to claim 1, characterized in that, Also includes: Based on the risk value and at least one risk treatment threshold, a risk treatment strategy is executed.

9. The method according to claim 8, characterized in that, The risk handling strategy is executed based on the risk value and at least one risk alarm threshold, employing at least one of the following methods: If the risk value is less than the first risk handling threshold, the vehicle will continue to drive. If the risk value is greater than the first risk handling threshold and less than the second risk handling threshold, the vehicle will continue driving and the risk value will be displayed. If the risk value is greater than the second risk handling threshold and less than the third risk handling threshold, then the risk value is displayed. If feedback is received, the vehicle will continue driving; if no feedback is received, the vehicle will stop driving.

10. A collision detection device, characterized in that, include: An obstacle recognition unit is used to identify obstacles on the road where the vehicle is traveling. An obstacle trajectory prediction unit is used to predict the first motion trajectory of an obstacle in a future time period. The first motion trajectory includes: the location information of the obstacle at a given time point, and the location distribution covariance information; The vehicle trajectory acquisition unit is used to acquire the second motion trajectory of the vehicle in a future time period; The risk determination unit is used to obtain a risk value for a collision between the vehicle and an obstacle in a future time period based on a first motion trajectory and a second motion trajectory. This includes: determining the vehicle's position information at various points in time within the future time period based on the second motion trajectory; for any given point in time, performing the following processing: sampling the obstacle position corresponding to the time point using a Monte Carlo algorithm based on the covariance and position information; for each sampled obstacle position corresponding to the time point, obtaining the number of collision points between the vehicle and the obstacle corresponding to the time point based on the obstacle position information and the vehicle's position information; and obtaining the risk value for a collision between the vehicle and the obstacle at that time point based on the obstacle position sampling number information and the number of collision points.

11. An electronic device, characterized in that, include: This includes: the processor, and the memory; The memory is used to store a program for implementing the collision risk detection method according to any one of claims 1-9, wherein the device is powered on and the program of the method is run by the processor.

Images