Obstacle mis-detection identification method and apparatus, electronic device, and medium

Abstract

The present disclosure provides an obstacle mis-detection identification method and device, electronic equipment and medium, and relates to the field of artificial intelligence. The specific implementation scheme is: obtaining an obstacle collection image collected by a target vehicle; determining an obstacle identification image from the obstacle collection image, and determining the number of images of the obstacle identification image; wherein the obstacle identification image is an obstacle collection image in which a risk obstacle is identified, and the risk obstacle is a candidate obstacle detected as having a collision risk with the target vehicle; determining attribute information of the risk obstacle according to the obstacle identification image; wherein the attribute information includes at least one of a position, a surface area, and a volume; and identifying mis-detection of the risk obstacle according to the attribute information and / or the number of images. The present disclosure achieves the effect of identifying mis-detected risk obstacles, improves the accuracy and reliability of risk obstacle determination, and thus improves the driving safety of autonomous vehicles.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, specifically to the fields of autonomous driving, intelligent transportation, electronic maps and cloud computing, and particularly to a method, apparatus, electronic device and medium for identifying false obstacle detection. Background Technology

[0002] With the growth of car ownership and the advancement of technology, more and more vehicles are equipped with autonomous driving capabilities. Autonomous vehicles are typically equipped with various sensor systems, such as vision sensors, millimeter-wave radar sensors, and lidar sensors, to ensure that they can fully perceive obstacles under different conditions.

[0003] In existing technologies, the main focus is on how to identify risky obstacles that pose a collision risk to autonomous vehicles. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and medium for identifying false obstacle detection to improve the accuracy and reliability of risk obstacle determination.

[0005] According to one aspect of this disclosure, a method for identifying false obstacle detection is provided, comprising:

[0006] Acquire obstacle images captured by the target vehicle;

[0007] Obstacle recognition images are determined from the obstacle acquisition images, and the number of obstacle recognition images is determined; wherein, the obstacle recognition images are obstacle acquisition images that identify risk obstacles, and the risk obstacles are candidate obstacles detected as having a collision risk with the target vehicle;

[0008] The attribute information of the risk obstacle is determined based on the obstacle recognition image; wherein the attribute information includes at least one of location, surface area, and volume;

[0009] Based on the attribute information and / or the number of images, the risk obstacle is falsely detected and identified.

[0010] According to another aspect of this disclosure, an obstacle false detection identification device is provided, comprising:

[0011] The image acquisition module is used to acquire obstacle images captured by the target vehicle;

[0012] An image quantity determination module is used to determine obstacle recognition images from the obstacle acquisition images and to determine the number of obstacle recognition images; wherein, the obstacle recognition image is an obstacle acquisition image that identifies a risk obstacle, and the risk obstacle is a candidate obstacle detected as having a collision risk with the target vehicle;

[0013] An attribute information determination module is used to determine the attribute information of the risk obstacle based on the obstacle recognition image; wherein the attribute information includes at least one of location, surface area, and volume;

[0014] The false detection identification module is used to identify the risk obstacle based on the attribute information and / or the number of images.

[0015] According to another aspect of this disclosure, an electronic device is provided, comprising:

[0016] At least one processor; and

[0017] A memory that is communicatively connected to at least one processor; wherein,

[0018] The memory stores instructions that can be executed by at least one processor to enable the at least one processor to perform any of the methods of this disclosure.

[0019] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause a computer to perform any of the methods of this disclosure.

[0020] According to another aspect of this disclosure, a computer program product is provided, including a computer program and a method for the computer program to be executed by a processor according to any of the methods disclosed herein.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0022] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0023] Figure 1 This is a flowchart of a method for identifying false obstacle detections according to embodiments of this disclosure;

[0024] Figure 2 This is a flowchart of another obstacle false detection identification method disclosed in the embodiments of this disclosure;

[0025] Figure 3 This is a flowchart of a method for determining some risk obstacles according to embodiments of this disclosure;

[0026] Figure 4 This is a schematic diagram of the structure of some obstacle false detection identification devices disclosed in the embodiments of this disclosure;

[0027] Figure 5 This is a block diagram of an electronic device used to implement the obstacle false detection identification method disclosed in the embodiments of this disclosure. Detailed Implementation

[0028] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0029] Autonomous vehicles are typically equipped with various sensor systems, such as vision sensors, millimeter-wave radar sensors, and lidar sensors, to ensure adequate obstacle perception under different conditions. The capabilities of different sensors vary across different scenarios; by leveraging the complementary strengths of data collected from multiple sensors in different spaces, downstream confidence assessments can be made more effective.

[0030] Existing technologies primarily focus on identifying hazardous obstacles that pose a collision risk to autonomous vehicles. For example, they utilize sensor-based information about the vehicle's current movement, filtering this information and applying either a fixed set of avoidance action parameters or a differentiated set of parameters tailored to the autonomous vehicle's driving parameters to identify such obstacles and ensure vehicle safety.

[0031] However, any sensor may experience data acquisition anomalies, leading to false detections of risky obstacles. Current technologies lack further identification and filtering mechanisms for these false detections, resulting in low accuracy and reliability in obstacle identification. Furthermore, since false detections of risky obstacles can trigger a series of abnormal driving behaviors in autonomous vehicles, this undoubtedly leads to lower driving safety for autonomous vehicles.

[0032] Figure 1This is a flowchart of some obstacle false detection identification methods disclosed in embodiments of this disclosure. This embodiment can be applied to situations where further false detection identification of risky obstacles posing a collision risk is required. The method of this embodiment can be executed by the obstacle false detection identification device disclosed in embodiments of this disclosure. The device can be implemented in software and / or hardware and can be integrated into any electronic device with computing capabilities.

[0033] like Figure 1 As shown, the obstacle false detection identification method disclosed in this embodiment may include:

[0034] S101. Acquire obstacle images captured by the target vehicle.

[0035] The target vehicle refers to any vehicle equipped with autonomous driving capabilities, including but not limited to sedans, SUVs (Sport Utility Vehicles), MPVs (Multi-Purpose Vehicles), buses, and trucks. Obstacle acquisition images are images captured by the target vehicle to perceive surrounding obstacles. These images are acquired using visual sensors mounted on the target vehicle, such as high-precision cameras or high-resolution cameras, to capture images of the environment surrounding the target vehicle.

[0036] By acquiring obstacle images captured by the target vehicle,

[0037] S102. Determine the obstacle recognition image from the obstacle acquisition image, and determine the number of obstacle recognition images.

[0038] Among them, the obstacle recognition image is the obstacle acquisition image that identifies the risky obstacle. In other words, the risky obstacle recognition based on the obstacle acquisition image includes two cases: "successful identification of risky obstacles" and "failure to identify risky obstacles". The obstacle acquisition image that "successfully identifies risky obstacles" is then used as the obstacle recognition image.

[0039] Risk obstacles are candidate obstacles detected that pose a collision risk to the target vehicle. A candidate obstacle represents the set of all obstacles surrounding the target vehicle. Collision risk indicates the probability of a collision with the target vehicle even if it brakes.

[0040] In one implementation, the target vehicle detects the collision risk of each candidate obstacle and, based on the detection results, identifies the risky obstacles that pose a collision risk to the target vehicle from among the candidate obstacles. The number of risky obstacles can be one or more.

[0041] The target vehicle acquires obstacle images from its visual sensors and uses a pre-defined target detection algorithm to identify obstacles in each image. Based on the identification results, it determines whether a risky obstacle has been successfully identified in each image. If a risky obstacle is successfully identified in any image, that image is designated as the obstacle identification image. This process is repeated across all obstacle images to count the number of identified obstacle identification images.

[0042] By identifying obstacle recognition images from obstacle acquisition images and determining the number of obstacle recognition images, a data foundation is laid for subsequent false detection and identification of risky obstacles based on the number of images.

[0043] S103. Determine the attribute information of the risky obstacle based on the obstacle recognition image.

[0044] The attribute information of a risk obstacle represents its inherent physical properties, including but not limited to at least one of location, surface area, and volume. This attribute information is obtained by identifying obstacles in an obstacle recognition image; that is, by identifying any given obstacle recognition image, the attribute information corresponding to the risk obstacle in that image is obtained.

[0045] In one implementation, for obstacle recognition images in which risk obstacles are successfully identified, an edge detection algorithm is used to determine the position of the risk obstacle in each obstacle acquisition image; and, a three-dimensional acquisition image is obtained by three-dimensional reconstruction of the obstacle acquisition image, and the surface area and volume of the risk obstacle in each obstacle acquisition image are determined based on the three-dimensional acquisition image.

[0046] By determining the attribute information of risky obstacles based on obstacle recognition images, a data foundation is laid for subsequent false detection and identification of risky obstacles based on the attribute information.

[0047] S104. Based on attribute information and / or the number of images, identify false detections of risk obstacles.

[0048] Among them, the false detection identification of risk obstacles refers to the verification of whether there is actually a collision risk between the risk obstacle and the target vehicle.

[0049] In one implementation, the proportion of obstacle recognition images in the obstacle acquisition images is determined based on the number of obstacle recognition images and the total number of obstacle acquisition images. Then, based on the proportion, it is determined whether the risk obstacle is stably identified in the obstacle acquisition images. If it is determined that the risk obstacle is not stably identified, it is determined that the risk obstacle is in a false detection state. If it is determined that the risk obstacle is stably identified, it is determined that the risk obstacle is not in a false detection state.

[0050] In another implementation, based on the attribute information corresponding to the risky obstacle in each obstacle recognition image, the attribute information difference between temporally adjacent obstacle recognition images is determined, such as position difference, surface area difference, or volume difference, etc. Then, based on the attribute information difference, it is determined whether there is a sudden change in the attribute information of the risky obstacle between any adjacent obstacle recognition images, such as a sudden change in position, surface area, or volume, etc. If so, it is determined that the risky obstacle is in a false detection state; otherwise, it is determined that the risky obstacle is not in a false detection state.

[0051] In another implementation, the proportion of obstacle recognition images in the total number of obstacle acquisition images is determined based on the number of obstacle recognition images and the total number of obstacle acquisition images. Then, based on the proportion, it is determined whether risk obstacles are stably identified in the obstacle acquisition images. If risk obstacles are stably identified, the attribute information difference between temporally adjacent obstacle recognition images is determined based on the attribute information corresponding to the risk obstacles in each obstacle recognition image. Then, based on the attribute information difference, it is determined whether there is a sudden change in the attribute information of the risk obstacle between any adjacent obstacle recognition images. If so, it is determined that the risk obstacle is in a false detection state; otherwise, it is determined that the risk obstacle is not in a false detection state.

[0052] This disclosure acquires obstacle acquisition images from a target vehicle, determines obstacle recognition images from these images, and determines the number of obstacle recognition images. The obstacle recognition images are those that identify potential obstacles, and the potential obstacles are candidate obstacles detected as posing a collision risk to the target vehicle. The attribute information of the potential obstacles is determined based on the obstacle recognition images. The attribute information includes at least one of location, surface area, and volume. Based on the attribute information and / or the number of images, potential obstacles are falsely detected, improving the accuracy and reliability of obstacle determination. This avoids abnormal avoidance or braking of falsely detected potential obstacles by the target vehicle during autonomous driving, thus improving the driving safety of the target vehicle during autonomous driving.

[0053] Figure 2This is a flowchart of another obstacle false detection identification method disclosed in the embodiments of this disclosure, which is further optimized and extended based on the above technical solution, and can be combined with the above optional implementation methods.

[0054] like Figure 2 As shown, the obstacle false detection identification method disclosed in this embodiment may include:

[0055] S201. Acquire obstacle images captured by the target vehicle.

[0056] S202. The moment when the risk obstacle is detected is taken as the first moment, and the image of the obstacle corresponding to the first moment is taken as the first acquired image. The second moment before the first moment and the third moment after the first moment are determined.

[0057] The second and third moments can be adjusted and set according to actual business needs. Preferably, the moment that is before the first moment and is three image acquisition moments apart from the first moment is taken as the second moment, and the moment that is after the first moment and is three image acquisition moments apart from the first moment is taken as the third moment.

[0058] S203. The obstacle acquisition image corresponding to the first moment and the second moment is used as the second acquisition image, and the obstacle acquisition image corresponding to the first moment and the third moment is used as the third acquisition image.

[0059] For example, suppose the moment before the first moment and three image acquisition moments apart from the first moment is designated as the second moment, and the moment after the first moment and three image acquisition moments apart from the first moment is designated as the third moment. Then, the three obstacle acquisition images that are sequentially before the first acquisition image will be designated as the second acquisition image; and the three obstacle acquisition images that are sequentially after the first acquisition image will be designated as the third acquisition image.

[0060] S204. Use the first acquired image, the second acquired image, and the third acquired image as initial acquired images, and determine the obstacle recognition image from the initial acquired images.

[0061] In one implementation, the set of acquired images, including the first acquired image, the second acquired image, and the third acquired image, is used as the initial acquired image, and then obstacle recognition images that identify risk obstacles are selected from the initial acquired image.

[0062] By taking the moment when a risky obstacle is detected as the first moment and the image of the obstacle corresponding to the first moment as the first acquired image; determining the second moment before the first moment and the third moment after the first moment; taking the image of the obstacle corresponding to the first moment and the second moment as the second acquired image, and taking the image of the obstacle corresponding to the first moment and the third moment as the third acquired image; and using the first, second, and third acquired images as the initial acquired images, and determining the obstacle recognition image from the initial acquired images, this method achieves the selection of a specific range of images from the obstacle acquired images as the initial acquired images, and then determining the obstacle recognition image from the initial acquired images. This greatly reduces the amount of data processing required for subsequent false detection and recognition of risky obstacles, improves the recognition efficiency of false detection and recognition of risky obstacles, and shortens the recognition time.

[0063] Optionally, an obstacle recognition image is determined from the initially acquired image, including the following steps A, B, and C:

[0064] A. Determine the current driving scenario of the target vehicle based on its current driving trajectory.

[0065] The vehicle's driving trajectory is the automatically predicted trajectory when the target vehicle is in autonomous driving mode, which the target vehicle uses as a basis for autonomous driving. Driving scenarios include, but are not limited to, turning scenarios and straight-line scenarios.

[0066] In one implementation, the trajectory curvature of the vehicle's driving trajectory is determined, and the trajectory curvature is matched with the curvature range corresponding to a 5-turn scenario and the curvature range corresponding to a straight-line scenario.

[0067] If the trajectory curvature falls within the turning curvature range, then the target vehicle is determined to be currently in a turning scenario.

[0068] If the trajectory curvature falls within the straight-line curvature range, then the target vehicle is determined to be currently in a straight-line scenario.

[0069] B. Identify abnormal images from the initial acquired images based on the driving scenario, and use the initial acquired images after removing the abnormal images as optimized acquired images.

[0070] 0. Among them, abnormal acquisition images refer to initial acquisition images that are abnormal and may pose risks and obstacles to subsequent acquisition.

[0071] Misidentification of objects can cause interference.

[0072] In one implementation, when the driving scenario is a turning scenario, abnormal acquisition images are identified from the initial acquisition images based on the curvature difference of the vehicle's trajectory between adjacent initial acquisition images. The initial acquisition images after removing the abnormal acquisition images are used as optimized acquisition images.

[0073] 5. In another implementation, when the driving scenario is a straight-ahead scenario, then according to each initial...

[0074] The trajectory curvature of the vehicle's driving path in the initial acquired image, as well as the road curvature of the road where the target vehicle is located, are used to identify abnormal acquired images from the initial acquired image. The initial acquired image after removing abnormal acquired images is used as the optimized acquired image.

[0075] C. Determine obstacle recognition images from optimized acquired images.

[0076] 0. Determine the current location of the target vehicle based on its current driving trajectory.

[0077] In driving scenarios, abnormal images are identified from the initial acquired images based on the driving scenario. The initial acquired images after removing abnormal images are used as optimized acquired images. Obstacle recognition images are then determined from the optimized acquired images. This process filters abnormal acquired images and ensures the accuracy of obstacle recognition images determined based on the filtered optimized acquired images.

[0078] 5. Optionally, step B includes steps B11 and B12:

[0079] B11. In the case of a turning driving scenario, determine the trajectory curvature of the vehicle's driving trajectory in the initial acquired image.

[0080] B12. Determine the first curvature difference between adjacent initial acquisition images and identify adjacent initial acquisition images with a first curvature difference greater than the first curvature threshold as abnormal acquisition images.

[0081] In one implementation, a first curvature difference of the trajectory curvature between all adjacent initial acquisition images is determined, and the first curvature difference is compared with a first curvature threshold. If the first curvature difference of the trajectory curvature between any two adjacent initial acquisition images is greater than the first curvature threshold, then the two adjacent initial acquisition images are regarded as abnormal acquisition images.

[0082] By determining the trajectory curvature of the vehicle's trajectory in the initial acquired image when the driving scenario is a turning scenario, determining the first curvature difference between adjacent initial acquired images, and identifying adjacent initial acquired images with a first curvature difference greater than a first curvature threshold as abnormal acquired images, the system can identify images with abnormal deviations in the vehicle's trajectory, thus avoiding interference from these abnormal images in the subsequent false detection and identification of risk obstacles.

[0083] Optionally, step B may also include steps B21 and B22:

[0084] B21. In the case of a straight-line driving scenario, determine the trajectory curvature of the vehicle's driving trajectory in the initial acquired image, as well as the road curvature of the road where the target vehicle is located.

[0085] B22. Determine the second curvature difference between the trajectory curvature and the road curvature, and use the initial acquired images where the second curvature difference is greater than the second curvature threshold as abnormal acquired images.

[0086] In one implementation, a second curvature difference between the trajectory curvature and the road curvature in each initial acquisition image is determined, and the second curvature difference corresponding to each initial acquisition image is compared with a second curvature threshold. If the second curvature difference corresponding to any initial acquisition image is greater than the second curvature threshold, then the initial acquisition image is regarded as an abnormal acquisition image.

[0087] By determining the trajectory curvature of the vehicle and the road curvature of the road where the target vehicle is located in the initial acquisition image when the driving scenario is straight, and determining the second curvature difference between the trajectory curvature and the road curvature, the initial acquisition image with the second curvature difference greater than the second curvature threshold is identified as an abnormal acquisition image. This enables the identification of images with abnormal curvature of the vehicle trajectory, such as the target vehicle suddenly making a U-turn in a straight scenario, and avoids these abnormal images from interfering with the subsequent false detection and identification of risk obstacles.

[0088] S205. Determine the attribute information of the risky obstacle based on the obstacle recognition image.

[0089] S206. Based on attribute information and / or the number of images, identify false detections of risk obstacles.

[0090] Optionally, S206 includes:

[0091] Based on the number of images and the total number of images in the optimized acquisition, determine the proportion of obstacle recognition images in the optimized acquisition images; compare the proportion with a proportion threshold, and if the proportion is less than the proportion threshold, determine that the risky obstacle is in a false detection state.

[0092] In one implementation, the ratio between the number of optimized images that identified risky obstacles and the total number of optimized images is calculated as the proportion of obstacle-identified images in the optimized images. This proportion is compared with a proportion threshold. If the proportion is less than the threshold, it indicates that risky obstacles are sometimes identified and sometimes not in the optimized images, meaning that risky obstacles are not consistently identified in the optimized images, thus determining that the risky obstacle is in a false detection state.

[0093] By determining the proportion of obstacle recognition images in the optimized acquisition images based on the number of images and the total number of images in the optimized acquisition images, the proportion is compared with a proportion threshold. If the proportion is less than the proportion threshold, the risk obstacle is determined to be in a false detection state. This achieves the effect of identifying risk obstacles that have not been stably identified, thereby determining that these risk obstacles are in a false detection state and improving the accuracy and reliability of risk obstacle identification.

[0094] Optionally, S206 also includes:

[0095] 1) Determine the first attribute information of the risk obstacle in the first recognition image and the second attribute information of the obstacle in the second recognition image; wherein the first recognition image and the second recognition image are adjacent images in the obstacle recognition image.

[0096] In one implementation, first attribute information corresponding to a risk obstacle in a first identification image and second attribute information corresponding to a second identification image are determined, wherein the first identification image and the second identification image are any two temporally adjacent images in the obstacle identification image.

[0097] 2) Determine the attribute information difference between the first attribute information and the second attribute information, and compare the attribute information difference with the attribute difference threshold.

[0098] In one implementation, the difference between the first attribute information and the second attribute information of the risk obstacle is calculated, including but not limited to at least one of the following: the difference in position of the target vehicle, the difference in surface area, and the difference in volume.

[0099] The attribute information difference is compared with the attribute difference threshold of the corresponding category. For example, the location difference is compared with the location difference threshold; the surface area difference is compared with the surface area difference threshold; and the volume difference is compared with the volume difference threshold.

[0100] 3) If the attribute information difference is greater than the attribute difference threshold, the risk obstacle is determined to be in a false detection state.

[0101] In one implementation, if the attribute information difference is greater than the attribute difference threshold, it indicates that the attribute information of the risk obstacle has changed abruptly between the first recognition image and the second recognition image, thereby determining that the risk obstacle is in a false detection state.

[0102] For example, if the position difference is greater than the position difference threshold, it means that the position of the risk obstacle has changed between the first recognition image and the second recognition image, that is, the position of the risk obstacle is not accurately determined, and thus the risk obstacle is determined to be in a false detection state.

[0103] When the surface area difference is greater than the surface area difference threshold and / or the volume difference is greater than the volume difference threshold, it indicates that the shape of the risk obstacle has changed abruptly between the first recognition image and the second recognition image. That is, the shape detection of the risk obstacle is inaccurate, and thus it is determined that the risk obstacle is in a false detection state.

[0104] By determining the first attribute information corresponding to the risk obstacle in the first recognition image and the second attribute information corresponding to the risk obstacle in the second recognition image, wherein the first recognition image and the second recognition image are adjacent images in the obstacle recognition image, the attribute information difference between the first attribute information and the second attribute information is determined, and the attribute information difference is compared with the attribute information difference threshold. If the attribute information difference is greater than the attribute information difference threshold, the risk obstacle is determined to be in a false detection state. This realizes the identification of risk obstacles with sudden changes in attribute information, and thus determines that these risk obstacles are in a false detection state, thereby improving the accuracy and reliability of risk obstacle determination.

[0105] Optionally, after identifying false detections of risk obstacles, the process may also include:

[0106] Data annotations were performed on risky obstacles that were falsely detected and those that were not, and a training dataset was constructed based on the annotation results. The training dataset was used to train a false detection identification model, which was then deployed in the target vehicle so that the vehicle could subsequently use the model to identify falsely detected risky obstacles.

[0107] Figure 3 This is a flowchart of a method for determining some risk obstacles disclosed in the embodiments of this disclosure. Based on the above technical solutions, it is further optimized and extended, and can be executed before "acquiring obstacle acquisition images collected by the target vehicle", and can be combined with the above optional implementation methods.

[0108] like Figure 3 As shown, the method for determining risk obstacles disclosed in this embodiment may include:

[0109] S301. Based on the target vehicle's driving trajectory, the location of the candidate obstacles, and the type of the candidate obstacles, determine the target obstacle from the candidate obstacles.

[0110] The obstacle position of a candidate obstacle represents its location information in the world coordinate system. The obstacle types for candidate obstacles include static obstacles and dynamic obstacles.

[0111] In one implementation, auxiliary obstacles are selected from the candidate obstacles based on their locations and the target vehicle's trajectory. Then, the target obstacle is determined from the auxiliary obstacles based on their types and the vehicle's trajectory.

[0112] Optionally, S301 includes:

[0113] Based on the vehicle's trajectory and the location of obstacles, auxiliary obstacles are determined from the candidate obstacles; based on the vehicle's trajectory and the type of obstacle in the auxiliary obstacles, the target obstacle is determined from the auxiliary obstacles.

[0114] In one implementation, the distance between the location of each candidate obstacle and the vehicle's trajectory is determined, and candidate obstacles with distance values ​​less than or equal to a distance threshold are designated as auxiliary obstacles. The obstacle type of the auxiliary obstacles is determined. If the obstacle type is a static obstacle, a target obstacle is determined from the auxiliary obstacles based on the obstacle's location and the vehicle's trajectory. If the obstacle type is a dynamic obstacle, the trajectory of the auxiliary obstacles is further determined, and the target obstacle is determined from the auxiliary obstacles based on the obstacle's trajectory and the vehicle's trajectory.

[0115] By identifying auxiliary obstacles from candidate obstacles based on vehicle trajectory and obstacle location, and then identifying target obstacles from auxiliary obstacles based on vehicle trajectory and obstacle type, the system achieves the effect of screening candidate obstacles based on obstacle location and type, eliminating non-risk obstacles and improving the accuracy and reliability of risk obstacle identification.

[0116] Optionally, based on the vehicle's trajectory and the type of obstacle among the auxiliary obstacles, the target obstacle is determined from the auxiliary obstacles, including:

[0117] When the obstacle type is a static obstacle, the target obstacle is determined from the auxiliary obstacles based on the positional relationship between the position of the auxiliary obstacles and the vehicle's trajectory.

[0118] In one implementation, if any auxiliary obstacle is a static obstacle, it is determined whether the location of the auxiliary obstacle is on the vehicle's driving trajectory. If so, it indicates that the static auxiliary obstacle has a potential risk of colliding with the target vehicle, and thus the auxiliary obstacle is determined to be the target obstacle.

[0119] By identifying the target obstacle from the auxiliary obstacles based on the positional relationship between the auxiliary obstacle's position and the vehicle's trajectory when the obstacle type is static, the identification of static obstacles with potential collision risk with the target vehicle is achieved.

[0120] Optionally, based on the vehicle's trajectory and the type of obstacle among the auxiliary obstacles, the target obstacle is determined from the auxiliary obstacles, including:

[0121] When the obstacle type is dynamic obstacle, determine the obstacle trajectory of the auxiliary obstacle; based on the positional relationship between the obstacle trajectory and the vehicle trajectory, determine the target obstacle from the auxiliary obstacles.

[0122] Among them, the obstacle trajectory is the predicted trajectory of the dynamic obstacle based on the motion information of the dynamic obstacle.

[0123] In one implementation, when any auxiliary obstacle is a dynamic obstacle, the obstacle's trajectory is predicted, and it is determined whether the obstacle's trajectory intersects with the vehicle's trajectory. If so, it indicates that the dynamic auxiliary obstacle has a potential risk of colliding with the target vehicle, and thus the auxiliary obstacle is determined to be the target obstacle.

[0124] By determining the obstacle trajectory of auxiliary obstacles when the obstacle type is dynamic, and identifying the target obstacle from the auxiliary obstacles based on the positional relationship between the obstacle trajectory and the vehicle trajectory, the identification of dynamic obstacles with potential collision risk with the target vehicle is achieved.

[0125] S302. Based on the target vehicle's current speed and braking deceleration, predict the target vehicle's stopping position, and determine the risk obstacle from the target obstacles based on the stopping position.

[0126] Among them, braking deceleration refers to the deceleration of the target vehicle when it performs emergency braking in autonomous driving mode.

[0127] In one implementation, the braking position of the target vehicle is predicted based on its current vehicle speed and braking deceleration. Based on the braking position and the obstacle position or trajectory of the target obstacle, a risk obstacle is identified from the target obstacles.

[0128] By identifying target obstacles from candidate obstacles based on vehicle trajectory, obstacle location, and obstacle type, and predicting the target vehicle's braking position based on its current speed and braking deceleration, and then identifying risk obstacles from the target obstacles based on the braking position, this method provides a way to determine risk obstacles and lays a data foundation for subsequent false detection and identification of risk obstacles.

[0129] Optionally, based on the target vehicle's current speed and braking deceleration, predict the target vehicle's stopping position, including:

[0130] 1) Determine the braking reaction distance of the target vehicle based on the vehicle speed and braking reaction time.

[0131] Braking reaction time is used to simulate the reaction time required for a person to brake a vehicle after sensing an obstacle. Braking reaction time can be set and adjusted based on empirical values.

[0132] In one implementation, the braking reaction distance corresponding to the target vehicle is determined based on the product of vehicle speed and braking reaction time.

[0133] 2) Determine the actual braking distance of the target vehicle based on the vehicle speed and braking deceleration.

[0134] In one implementation, the actual braking distance of the target vehicle when its speed is zero, i.e., when the target vehicle comes to a complete stop, is calculated based on the vehicle speed and braking deceleration.

[0135] 3) Determine the total braking distance based on the braking reaction distance and the actual braking distance, and predict the stopping position of the target vehicle based on the total braking distance and the vehicle's trajectory.

[0136] In one implementation, the sum of the braking reaction distance and the actual braking distance is taken as the total braking distance. Based on the total braking distance and the vehicle's trajectory, the braking position of the target vehicle on the vehicle's trajectory is determined.

[0137] By determining the braking reaction distance of the target vehicle based on vehicle speed and braking reaction time, and the actual braking distance of the target vehicle based on vehicle speed and braking deceleration, the total braking distance is determined based on the braking reaction distance and the actual braking distance. Based on the total braking distance and the vehicle's trajectory, the stopping position of the target vehicle is predicted. This makes the calculation of the total braking distance more consistent with real-world scenarios and ensures the accuracy of the final calculated stopping position.

[0138] Optionally, risk obstacles can be identified from the target obstacles based on the braking position, including:

[0139] 1) Determine the braking trajectory of the target vehicle from the vehicle's driving trajectory based on the braking position.

[0140] In one implementation, the trajectory of the vehicle's driving path located between the target vehicle's current position and its braking position is taken as the braking trajectory of the target vehicle.

[0141] 2) When the obstacle type of the target obstacle is a static obstacle, the risk obstacle is determined from the target obstacle based on the positional relationship between the obstacle position and the braking trajectory.

[0142] In one implementation, for any target obstacle that is a static obstacle, it is determined whether the obstacle's position is on the braking trajectory. If so, it indicates that even if the target vehicle takes braking measures, the target obstacle still poses a risk of colliding with the target vehicle, and thus the target obstacle is determined to be a risk obstacle.

[0143] 3) When the obstacle type of the target obstacle is a dynamic obstacle, the risk obstacle is determined from the target obstacle based on the positional relationship between the obstacle's travel trajectory and braking trajectory.

[0144] In one implementation, for any target obstacle that is a dynamic obstacle, it is determined whether there is an intersection between the obstacle's travel trajectory and its braking trajectory. If so, it indicates that even if the target vehicle takes braking measures, the target obstacle still poses a risk of colliding with the target vehicle, and thus the target obstacle is determined to be a risk obstacle.

[0145] By determining the braking trajectory of the target vehicle from its braking position, and considering that the target obstacle is a static obstacle, risk obstacles are identified from the target obstacles based on the positional relationship between the obstacle's position and the braking trajectory. Similarly, if the target obstacle is a dynamic obstacle, risk obstacles are identified from the target obstacles based on the positional relationship between the obstacle's driving trajectory and the braking trajectory. This approach achieves comprehensive identification of both static and dynamic risk obstacles, ensuring the reliability, comprehensiveness, and accuracy of risk obstacle identification.

[0146] This disclosure also provides a preferred embodiment of a method for identifying false obstacle detection, including:

[0147] 1) Acquire the initial image of the target vehicle.

[0148] 2) Determine whether the target vehicle is in a straight-ahead scenario based on the vehicle's trajectory in the initial acquired image. If not, proceed to step 3); otherwise, proceed to step 4.

[0149] 3) If the target vehicle is in a turning scene, determine whether there is any deviation in the vehicle's trajectory between any adjacent initial acquisition images. If so, remove the adjacent initial acquisition images to obtain the optimized acquisition images.

[0150] 4) If the target vehicle is in a straight-moving scene, determine whether there is any initial image in which the curvature between the vehicle's trajectory and the road is too large. If so, remove the initial image and obtain the optimized image.

[0151] 5) For optimized image acquisition, determine whether risky obstacles are consistently detected. If not, determine that the risky obstacle is in a false detection state. If so, proceed to step 6).

[0152] 6) For obstacle recognition images in the optimized acquisition images, determine whether the risky obstacle has a positional jump between adjacent obstacle recognition images. If so, determine that the risky obstacle is in a false detection state; otherwise, proceed to step 7).

[0153] 7) For obstacle recognition images in optimized acquisition images, determine whether the risk obstacle has a sudden change in shape between adjacent obstacle recognition images, including but not limited to sudden changes in surface area and / or volume. If so, determine that the risk obstacle is in a false detection state; if not, determine that the risk obstacle is not in a false detection state.

[0154] This disclosure also provides a preferred embodiment of a method for determining risk barriers, including:

[0155] 1) Obtain the status information of the target vehicle and determine whether the target vehicle is in autonomous driving mode based on the status information. If not, stop executing the subsequent steps; if so, proceed to step 2).

[0156] 2) Determine the location and type of candidate obstacles based on road information. Then, select the ROI area of ​​the obstacle based on the obstacle location, obstacle type, and the vehicle trajectory of the target vehicle, and determine the auxiliary obstacle based on the ROI area.

[0157] 3) Determine whether the auxiliary obstacle intersects with the target vehicle's trajectory, filter out non-intersecting auxiliary obstacles, and use the remaining auxiliary obstacles as the target obstacle.

[0158] 4) Based on the vehicle's driving trajectory and vehicle driving information (such as vehicle speed, braking reaction time, and braking deceleration), determine whether each target obstacle poses a collision risk with the target vehicle, and designate the target obstacles that are determined to pose a collision risk with the target vehicle as risk obstacles.

[0159] 5) Record obstacle information of the risk obstacles and vehicle driving information of the target vehicle.

[0160] Figure 4 This is a schematic diagram of the structure of an obstacle false detection identification device disclosed in some embodiments of this disclosure. It can be applied to situations where further false detection identification is needed for risky obstacles posing a collision risk. The device in this embodiment can be implemented in software and / or hardware and can be integrated into any electronic device with computing capabilities.

[0161] like Figure 4 As shown, the obstacle false detection identification device 40 disclosed in this embodiment may include an image acquisition module 41, an image quantity determination module 42, an attribute information determination module 43, and a false detection identification module 42, wherein:

[0162] Image acquisition module 41 is used to acquire obstacle acquisition images captured by the target vehicle;

[0163] The image quantity determination module 42 is used to determine obstacle recognition images from the obstacle acquisition images and determine the number of obstacle recognition images; wherein, the obstacle recognition image is an obstacle acquisition image that identifies a risk obstacle, and the risk obstacle is a candidate obstacle detected as having a collision risk with the target vehicle;

[0164] The attribute information determination module 43 is used to determine the attribute information of the risk obstacle based on the obstacle recognition image; wherein the attribute information includes at least one of location, surface area and volume;

[0165] The false detection identification module 44 is used to identify the risk obstacle based on the attribute information and / or the number of images.

[0166] Optionally, the image quantity determination module 42 is specifically used for:

[0167] The moment when a risky obstacle is detected is taken as the first moment, and the image of the obstacle corresponding to the first moment is taken as the first acquired image;

[0168] Determine the second moment before the first moment, and the third moment after the first moment;

[0169] The obstacle image acquired between the first and second moments is used as the second acquired image, and the obstacle image acquired between the first and third moments is used as the third acquired image.

[0170] The first, second, and third acquired images are used as initial acquired images, and obstacle recognition images are determined from the initial acquired images.

[0171] Optionally, the image quantity determination module 42 is further used for:

[0172] Based on the target vehicle's current driving trajectory, determine the current driving scenario of the target vehicle;

[0173] Based on the driving scenario, abnormal images are identified from the initial acquired images, and the initial acquired images after removing the abnormal images are used as optimized acquired images.

[0174] Obstacle recognition images are determined from optimized acquired images.

[0175] Optionally, the image quantity determination module 42 is further used for:

[0176] In the case of a turning driving scenario, determine the trajectory curvature of the vehicle's driving trajectory in the initial acquired image;

[0177] Determine the first curvature difference between adjacent initial acquisition images, and identify adjacent initial acquisition images whose first curvature difference is greater than the first curvature threshold as abnormal acquisition images.

[0178] Optionally, the image quantity determination module 42 is further used for:

[0179] In the case of a straight-line driving scenario, determine the trajectory curvature of the vehicle's trajectory in the initial acquired image, as well as the road curvature of the road where the target vehicle is located.

[0180] Determine the second curvature difference between the trajectory curvature and the road curvature, and use the initial acquired images where the second curvature difference is greater than the second curvature threshold as abnormal acquired images.

[0181] Optional, the false detection identification module 44 is specifically used for:

[0182] Based on the number of images and the total number of optimized images, determine the proportion of obstacle recognition images in the optimized images;

[0183] The quantity percentage is compared with the percentage threshold. If the quantity percentage is less than the percentage threshold, the risk obstacle is determined to be in a false detection state.

[0184] Optional, the false detection identification module 44 is specifically used for:

[0185] Determine the first attribute information corresponding to the risk obstacle in the first recognition image, and the second attribute information corresponding to the obstacle in the second recognition image; wherein, the first recognition image and the second recognition image are adjacent images in the obstacle recognition image;

[0186] Determine the attribute information difference between the first attribute information and the second attribute information, and compare the attribute information difference with the attribute difference threshold;

[0187] If the difference in attribute information is greater than the attribute difference threshold, the risk obstacle is determined to be in a false detection state.

[0188] Optionally, the device also includes a risk obstacle determination module, specifically used for:

[0189] The target obstacle is determined from the candidate obstacles based on the target vehicle's driving trajectory, the location of the candidate obstacles, and the type of the candidate obstacles.

[0190] Based on the target vehicle's current speed and braking deceleration, predict the target vehicle's stopping position, and identify the risky obstacle from the target obstacles based on the stopping position.

[0191] Optionally, the risk obstacle determination module is also used for:

[0192] Based on the vehicle's trajectory and the location of obstacles, auxiliary obstacles are determined from the candidate obstacles;

[0193] The target obstacle is determined from the auxiliary obstacles based on the vehicle's trajectory and the types of obstacles in the auxiliary obstacles.

[0194] Optionally, the risk obstacle determination module is also used for:

[0195] When the obstacle type is a static obstacle, the target obstacle is determined from the auxiliary obstacles based on the positional relationship between the position of the auxiliary obstacles and the vehicle's trajectory.

[0196] Optionally, the risk obstacle determination module is also used for:

[0197] When the obstacle type is dynamic obstacle, determine the obstacle travel trajectory of the auxiliary obstacle;

[0198] The target obstacle is determined from the auxiliary obstacles based on the positional relationship between the obstacle's trajectory and the vehicle's trajectory.

[0199] Optionally, the risk obstacle determination module is also used for:

[0200] Determine the corresponding braking reaction distance of the target vehicle based on the vehicle speed and braking reaction time;

[0201] Determine the actual braking distance of the target vehicle based on the vehicle speed and braking deceleration;

[0202] Based on the braking reaction distance and the actual braking distance, the total braking distance is determined, and based on the total braking distance and the driving trajectory, the stopping position of the target vehicle is predicted.

[0203] Optionally, the risk obstacle determination module is also used for:

[0204] Determine the braking trajectory of the target vehicle from its driving trajectory based on its stopping position;

[0205] When the obstacle type of the target obstacle is a static obstacle, the risk obstacle is determined from the target obstacles based on the positional relationship between the obstacle position and the braking trajectory;

[0206] When the target obstacle is a dynamic obstacle, risk obstacles are identified from the target obstacles based on the positional relationship between the obstacle's travel trajectory and braking trajectory.

[0207] The obstacle false detection identification device 40 disclosed in this embodiment can execute the obstacle false detection identification method disclosed in this embodiment, and has the corresponding functional modules and beneficial effects of executing the method. Content not described in detail in this embodiment can be referred to the description in the method embodiments of this disclosure.

[0208] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0209] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0210] Figure 5 A schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0211] like Figure 5 As shown, device 500 includes a computing unit 501, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 502 or a computer program loaded from storage unit 508 into random access memory (RAM) 503. RAM 503 may also store various programs and data required for the operation of device 500. The computing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. Input / output (I / O) interface 505 is also connected to bus 504.

[0212] Multiple components in device 500 are connected to I / O interface 505, including: input unit 506, such as keyboard, mouse, etc.; output unit 507, such as various types of monitors, speakers, etc.; storage unit 508, such as disk, optical disk, etc.; and communication unit 509, such as network card, modem, wireless transceiver, etc. Communication unit 509 allows device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0213] The computing unit 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above, such as the obstacle false detection identification method. For example, in some embodiments, the obstacle false detection identification method can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by the computing unit 501, one or more steps of the obstacle false detection identification method described above can be performed. Alternatively, in other embodiments, the computing unit 501 may be configured by any other suitable means (e.g., by means of firmware) to perform a method for identifying false obstacle detections.

[0214] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0215] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0216] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0217] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0218] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0219] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0220] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0221] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for identifying false obstacle detections, comprising: Acquire obstacle images captured by the target vehicle; The trajectory curvature of the target vehicle's driving trajectory is determined, and the trajectory curvature is matched with the turning curvature range. If it falls within the turning curvature range, the driving scenario of the target vehicle is determined to be a turning scenario. The trajectory curvature is matched with the straight-line curvature range. If it falls within the straight-line curvature range, the driving scenario is determined to be a straight-line scenario. Based on the driving scenario, abnormal acquisition images are identified from the initial acquisition images, and the initial acquisition images after removing the abnormal acquisition images are used as optimized acquisition images; wherein, the initial acquisition images include the obstacle acquisition image at the moment when the risk obstacle is detected, as well as at least one frame of obstacle acquisition image before and after that moment; Obstacle recognition images are determined from the optimized acquired images; wherein, the obstacle recognition image is an obstacle acquisition image that identifies risky obstacles, and the risky obstacles are candidate obstacles detected that pose a collision risk to the target vehicle; Determine the number of images in the obstacle recognition image; The attribute information of the risk obstacle is determined based on the obstacle recognition image; wherein the attribute information includes at least one of location, surface area, and volume; Based on the number of images and the total number of images in the optimized acquisition images, determine the proportion of the obstacle recognition images in the optimized acquisition images; If the percentage of the number is less than the percentage threshold, then the risk obstacle is determined to be in a false detection state; If the proportion of the number is not less than the proportion threshold, then the first attribute information of the risk obstacle in the first recognition image and the second attribute information of the obstacle in the second recognition image are determined; wherein, the first recognition image and the second recognition image are adjacent images in the obstacle recognition image; Determine the attribute information difference between the first attribute information and the second attribute information, and compare the attribute information difference with an attribute difference threshold; wherein, the attribute information difference includes at least one of position difference, surface area difference, and volume difference, and the attribute difference threshold includes at least one of position difference threshold, surface area difference threshold, and volume difference threshold corresponding to the category of the attribute information difference; If the attribute information difference is greater than the attribute difference threshold, then the risk obstacle is determined to be in a false detection state; otherwise, the risk obstacle is determined not to be in a false detection state. The process of determining abnormal acquisition images from the initially acquired images based on the driving scenario includes: In the case where the driving scenario is a turning scenario, the trajectory curvature of the vehicle's driving trajectory in the initial acquired image is determined; a first curvature difference value of the trajectory curvature between adjacent initial acquired images is determined, and adjacent initial acquired images with the first curvature difference value greater than a first curvature threshold are regarded as abnormal acquired images; In the case where the driving scenario is a straight-line scenario, the trajectory curvature of the vehicle's driving trajectory and the road curvature of the road where the target vehicle is located are determined in the initial acquired image; a second curvature difference between the trajectory curvature and the road curvature is determined, and the initial acquired image with the second curvature difference greater than the second curvature threshold is taken as the abnormal acquired image.

2. The method according to claim 1, wherein, Determining the obstacle recognition image from the obstacle acquisition image includes: The moment when the risk obstacle is detected is taken as the first moment, and the image of the obstacle corresponding to the first moment is taken as the first image. Determine the second moment before the first moment and the third moment after the first moment; The obstacle image acquired between the first time point and the second time point is used as the second acquired image, and the obstacle image acquired between the first time point and the third time point is used as the third acquired image; The first acquired image, the second acquired image, and the third acquired image are used as initial acquired images, and the obstacle recognition image is determined from the initial acquired images.

3. The method according to claim 1, further comprising, before acquiring the obstacle acquisition image collected by the target vehicle: The target obstacle is determined from the candidate obstacles based on the vehicle trajectory of the target vehicle, the location of the candidate obstacles, and the type of the candidate obstacles. Based on the target vehicle's current speed and braking deceleration, predict the target vehicle's stopping position, and determine the risk obstacle from the target obstacles based on the stopping position.

4. The method according to claim 3, wherein, The step of determining the target obstacle from the candidate obstacles based on the vehicle's driving trajectory, the obstacle position of the candidate obstacles, and the obstacle type of the candidate obstacles includes: Based on the vehicle's trajectory and the location of the obstacle, an auxiliary obstacle is determined from the candidate obstacles; The target obstacle is determined from the auxiliary obstacles based on the vehicle's driving trajectory and the obstacle type of the auxiliary obstacles.

5. The method according to claim 4, wherein, The step of determining the target obstacle from the auxiliary obstacles based on the vehicle's driving trajectory and the obstacle type of the auxiliary obstacles includes: When the obstacle type is a static obstacle, the target obstacle is determined from the auxiliary obstacles based on the positional relationship between the obstacle position of the auxiliary obstacle and the vehicle's driving trajectory.

6. The method according to claim 4, wherein, The step of determining the target obstacle from the auxiliary obstacles based on the vehicle's driving trajectory and the obstacle type of the auxiliary obstacles includes: When the obstacle type is a dynamic obstacle, determine the obstacle travel trajectory of the auxiliary obstacle; The target obstacle is determined from the auxiliary obstacles based on the positional relationship between the obstacle's trajectory and the vehicle's trajectory.

7. The method according to claim 3, wherein, The step of predicting the stopping position of the target vehicle based on its current speed and braking deceleration includes: Based on the vehicle speed and braking reaction time, determine the braking reaction distance corresponding to the target vehicle; The actual braking distance corresponding to the target vehicle is determined based on the vehicle speed and the braking deceleration. Based on the braking reaction distance and the actual braking distance, the total braking distance is determined, and based on the total braking distance and the vehicle's trajectory, the stopping position of the target vehicle is predicted.

8. The method according to claim 3, wherein, The step of determining the risk obstacle from the target obstacle based on the braking position includes: The braking trajectory of the target vehicle is determined from the vehicle's driving trajectory based on the braking position; If the obstacle type of the target obstacle is a static obstacle, a risk obstacle is determined from the target obstacles based on the positional relationship between the obstacle position of the target obstacle and the braking trajectory; When the obstacle type of the target obstacle is a dynamic obstacle, the risk obstacle is determined from the target obstacle based on the positional relationship between the obstacle's travel trajectory and the braking trajectory.

9. A device for identifying false obstacle detection, comprising: The image acquisition module is used to acquire obstacle images captured by the target vehicle; The image quantity determination module is used to determine the trajectory curvature of the target vehicle's driving trajectory, match the trajectory curvature with the turning curvature range, and if it belongs to the turning curvature range, determine that the current driving scenario of the target vehicle is a turning scenario; match the trajectory curvature with the straight-line curvature range, and if it belongs to the straight-line curvature range, determine that the driving scenario is a straight-line scenario; Based on the driving scenario, abnormal acquisition images are identified from the initial acquisition images, and the initial acquisition images after removing the abnormal acquisition images are used as optimized acquisition images; wherein, the initial acquisition images include the obstacle acquisition image at the moment when the risk obstacle is detected, as well as at least one frame of obstacle acquisition image before and after that moment; Obstacle recognition images are determined from the optimized acquired images; wherein, the obstacle recognition image is an obstacle acquisition image that identifies risky obstacles, and the risky obstacles are candidate obstacles detected that pose a collision risk to the target vehicle; Determine the number of images in the obstacle recognition image; An attribute information determination module is used to determine the attribute information of the risk obstacle based on the obstacle recognition image; wherein the attribute information includes at least one of location, surface area, and volume; The false detection identification module is used to determine the proportion of the obstacle identification images in the optimized acquisition images based on the number of images and the total number of images in the optimized acquisition images; If the percentage of the number is less than the percentage threshold, then the risk obstacle is determined to be in a false detection state; If the proportion of the number is not less than the proportion threshold, then the first attribute information of the risk obstacle in the first recognition image and the second attribute information of the obstacle in the second recognition image are determined; wherein, the first recognition image and the second recognition image are adjacent images in the obstacle recognition image; Determine the attribute information difference between the first attribute information and the second attribute information, and compare the attribute information difference with an attribute difference threshold; wherein, the attribute information difference includes at least one of position difference, surface area difference, and volume difference, and the attribute difference threshold includes at least one of position difference threshold, surface area difference threshold, and volume difference threshold corresponding to the category of the attribute information difference; If the attribute information difference is greater than the attribute difference threshold, then the risk obstacle is determined to be in a false detection state; otherwise, the risk obstacle is determined not to be in a false detection state. The image acquisition module is specifically used for: In the case where the driving scenario is a turning scenario, the trajectory curvature of the vehicle's driving trajectory in the initial acquired image is determined; a first curvature difference value of the trajectory curvature between adjacent initial acquired images is determined, and adjacent initial acquired images with the first curvature difference value greater than a first curvature threshold are regarded as abnormal acquired images; In the case where the driving scenario is a straight-line scenario, the trajectory curvature of the vehicle's driving trajectory and the road curvature of the road where the target vehicle is located are determined in the initial acquired image; a second curvature difference between the trajectory curvature and the road curvature is determined, and the initial acquired image with the second curvature difference greater than the second curvature threshold is taken as the abnormal acquired image.

10. The apparatus according to claim 9, wherein, The image quantity determination module is specifically used for: The moment when the risk obstacle is detected is taken as the first moment, and the image of the obstacle corresponding to the first moment is taken as the first image. Determine the second moment before the first moment and the third moment after the first moment; The obstacle image acquired between the first time point and the second time point is used as the second acquired image, and the obstacle image acquired between the first time point and the third time point is used as the third acquired image; The first acquired image, the second acquired image, and the third acquired image are used as initial acquired images, and the obstacle recognition image is determined from the initial acquired images.

11. The apparatus according to claim 9, further comprising a risk obstacle determination module, specifically used for: The target obstacle is determined from the candidate obstacles based on the vehicle trajectory of the target vehicle, the location of the candidate obstacles, and the type of the candidate obstacles. Based on the target vehicle's current speed and braking deceleration, predict the target vehicle's stopping position, and determine the risk obstacle from the target obstacles based on the stopping position.

12. The apparatus according to claim 11, wherein, The risk obstacle determination module is further used for: Based on the vehicle's trajectory and the location of the obstacle, an auxiliary obstacle is determined from the candidate obstacles; The target obstacle is determined from the auxiliary obstacles based on the vehicle's driving trajectory and the obstacle type of the auxiliary obstacles.

13. The apparatus according to claim 12, wherein, The risk obstacle determination module is further used for: When the obstacle type is a static obstacle, the target obstacle is determined from the auxiliary obstacles based on the positional relationship between the obstacle position of the auxiliary obstacle and the vehicle's driving trajectory.

14. The apparatus according to claim 12, wherein, The risk obstacle determination module is further used for: When the obstacle type is a dynamic obstacle, determine the obstacle travel trajectory of the auxiliary obstacle; The target obstacle is determined from the auxiliary obstacles based on the positional relationship between the obstacle's trajectory and the vehicle's trajectory.

15. The apparatus according to claim 11, wherein, The risk obstacle determination module is further used for: Based on the vehicle speed and braking reaction time, determine the braking reaction distance corresponding to the target vehicle; The actual braking distance corresponding to the target vehicle is determined based on the vehicle speed and the braking deceleration. Based on the braking reaction distance and the actual braking distance, the total braking distance is determined, and based on the total braking distance and the driving trajectory, the stopping position of the target vehicle is predicted.

16. The apparatus according to claim 11, wherein, The risk obstacle determination module is further used for: The braking trajectory corresponding to the target vehicle is determined from the driving trajectory based on the braking position; If the obstacle type of the target obstacle is a static obstacle, a risk obstacle is determined from the target obstacles based on the positional relationship between the obstacle position of the target obstacle and the braking trajectory; When the obstacle type of the target obstacle is a dynamic obstacle, the risk obstacle is determined from the target obstacle based on the positional relationship between the obstacle's travel trajectory and the braking trajectory.

17. An electronic device comprising: At least one processor; as well as 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 enable the at least one processor to perform the method of any one of claims 1-8.

18. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-8.

19. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-8.

Images