Obstacle type identification method and device and unmanned vehicle
Through dust mis-detection detection strategies for different obstacle types, combined with unmanned vehicle sensor data, identify and judge whether the target obstacle is dust, the problems of low robustness and safety in the existing technology are solved and the false detection is reduced The rate of recognition accuracy and the safety and operation efficiency of autonomous vehicles are improved.
Patent Information
- Application Number
- CN202411561679.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-04
- Publication Date
- 2025-02-18
AI Technical Summary
The existing autonomous driving obstacle detection technology is low in robustness and safety in harsh environments, with a high false detection rate, and cannot effectively deal with complex scenarios.
Provide an identification method for obstacle types. By obtaining the types of obstacles that need to be detected by dust mis-detection, combining the sensor data of unmanned vehicles, dust mis-detection detection strategies for different obstacle types are adopted to determine whether the target obstacle meets dust. Falsely check and determine that it is dust.
It reduces the error detection rate, improves the accuracy of obstacle type identification, and enhances the safety and operating efficiency of autonomous driving vehicles.
Smart Images
Figure CN119478892A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of perception and unmanned driving technology, and in particular to an obstacle type recognition method, device and unmanned vehicle. Background Art
[0002] The obstacle detection technology for autonomous driving mainly includes detecting obstacles and identifying different types of obstacles, such as vehicles, pedestrians, stationary objects, etc. Traditional obstacle detection technology relies on a variety of sensors to detect these obstacles, including lidar, radar, camera, ultrasonic radar, etc. The function of identifying different types of obstacles is achieved by analyzing and processing the data from these sensors.
[0003] However, in the actual working process, since the vehicle will encounter various harsh environments during driving, some rain, fog and dust will affect the vehicle's sensors and cameras, and the autonomous driving obstacle detection technology needs to handle complex scenes. For example, dust attached to the camera lens will affect the clarity of the image, causing the system to mistakenly recognize the existence of other objects, such as night driving, rain and snow. Traditional obstacle detection technology can only be used for detection in relatively simple scenes, resulting in low robustness and safety of traditional obstacle detection technology, and unable to cope with various uncertainties. Therefore, obstacle detection technology still has a lot of room for improvement in ensuring vehicle safety and operating efficiency. Summary of the invention
[0004] The embodiments of the present disclosure provide a method, device and unmanned vehicle for identifying obstacle types, so as to solve the problems of low robustness and safety and high false detection rate of existing obstacle detection technologies.
[0005] Based on the above problems, in a first aspect, a method for identifying obstacle types is provided, comprising:
[0006] Acquire at least one obstacle type for which dust false detection detection is required; wherein the dust false detection detection is used to detect false detection behavior of falsely detecting dust as a physical obstacle;
[0007] Based on the dust false detection detection strategies corresponding to the at least one obstacle type, and the obstacle data detected by the unmanned vehicle sensor, respectively, performing dust false detection detection on the target obstacles having the at least one obstacle type;
[0008] In the case where the target obstacle meets the dust false detection condition, it is determined that the target obstacle is dust.
[0009] In combination with the first aspect, in a possible implementation manner, the obstacle type includes an isolated static obstacle type;
[0010] The dust false detection detection strategy corresponding to the at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor respectively performs dust false detection detection on the target obstacle having the at least one obstacle type, including:
[0011] Determining a target obstacle belonging to the isolated static obstacle type from obstacle data detected by the unmanned vehicle sensor;
[0012] For each target obstacle, determining whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle;
[0013] If there is an obstacle belonging to the dynamic obstacle type, determining whether there is a trajectory point in the historical driving trajectory points that passes through the target obstacle based on the historical driving trajectory information of the obstacle belonging to the dynamic obstacle type;
[0014] If there is a trajectory point that passes through the target obstacle, it is determined that the target obstacle meets the dust false detection.
[0015] In combination with the first aspect, in a possible implementation manner, before determining whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle, the method further includes:
[0016] Determining, from the obstacle data, obstacles that are of the dust type;
[0017] For each target obstacle, determining whether there is an obstacle of dust type within a second distance threshold range around the target obstacle;
[0018] The step of determining, for each target obstacle, whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle includes:
[0019] If there is an obstacle of the dust type, then for each target obstacle, it is determined whether there is an obstacle of the dynamic obstacle type within the first distance threshold range around the target obstacle.
[0020] In combination with the first aspect, in a possible implementation manner, the first distance threshold range is greater than the second distance threshold range.
[0021] In combination with the first aspect, in a possible implementation manner, the method further includes:
[0022] If there is no track point passing through the target obstacle, it is determined that it is not a dust false detection; and / or
[0023] If there is no dust-type obstacle, it is determined that it is not a false dust detection; and / or
[0024] If the target obstacle, the obstacle of the dust type, or the obstacle of the dynamic obstacle type does not exist in the obstacle data, it is determined that it is not a dust false detection.
[0025] In combination with the first aspect, in a possible implementation manner, the obstacle type includes a dynamic obstacle type;
[0026] The dust false detection detection strategy corresponding to the at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor respectively performs dust false detection detection on the target obstacle having the at least one obstacle type, including:
[0027] Determining a target obstacle belonging to the dynamic obstacle type from obstacle data detected by the unmanned vehicle sensor;
[0028] For each target obstacle, determine the position information of the target obstacle obtained from the obstacle data, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle self-vehicle coordinate system;
[0029] Determining, according to the position coordinates, whether the target obstacle is located in a complete coverage area of the corresponding sensor of the unmanned vehicle;
[0030] If it is located in the fully covered area, the area of the target obstacle is determined based on the obstacle data corresponding to the target obstacle, and when the area is less than a preset area threshold, it is determined that the target obstacle meets the dust false detection.
[0031] In combination with the first aspect, in a possible implementation manner, if located in the fully covered area, the method further includes:
[0032] Determine the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset time;
[0033] The determining, based on the sensor data corresponding to the target obstacle, the area of the target obstacle, and determining that the target obstacle meets the dust false detection condition when the area is less than a preset area threshold, includes:
[0034] If the number is less than a preset number threshold, the area of the target obstacle is determined based on the sensor data corresponding to the target obstacle, and when the area is less than a preset area threshold, it is determined that the target obstacle meets the dust false detection.
[0035] In combination with the first aspect, in a possible implementation manner, the method further includes:
[0036] Based on a detection source of obstacle data corresponding to the target obstacle, determining whether the detection source is a single sensor;
[0037] The step of determining the position information of each target obstacle and converting the position information into the position coordinates of the target obstacle in the self-driving vehicle coordinate system includes:
[0038] If it comes from a single sensor, the position information of each target obstacle is determined, and the position information is converted into the position coordinates of the target obstacle in the coordinate system of the unmanned vehicle.
[0039] In combination with the first aspect, in a possible implementation manner, the method further includes:
[0040] If it is not from a single sensor, it is determined not to be a false dust detection; and / or
[0041] If it is not located in the fully covered area, it is determined that it is not a false dust detection; and / or
[0042] If the number is greater than or equal to a preset number threshold, it is determined that it is not a false dust detection; and / or
[0043] When the area is greater than or equal to the preset area threshold, it is determined that it is not a false dust detection.
[0044] In combination with the first aspect, in a possible implementation manner, the obstacle data is stored in the form of a queue; and different obstacle types are respectively stored in different obstacle queues;
[0045] The method further comprises:
[0046] After determining that the target obstacle is dust, the target obstacle type is changed to a dust type, deleted from the current obstacle queue, and added to a dust type obstacle queue.
[0047] In a second aspect, a device for identifying obstacle types is provided, comprising:
[0048] An obstacle acquisition module is used to acquire at least one obstacle type that needs to be tested for dust false detection; wherein the dust false detection is used to detect false detection behavior of falsely detecting dust as a physical obstacle;
[0049] A type identification module, configured to perform dust false detection detection on target obstacles having the at least one obstacle type based on the dust false detection detection strategies corresponding to the at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor;
[0050] The type determination module is used to determine that the target obstacle is dust when the target obstacle meets the dust false detection condition.
[0051] In a third aspect, an unmanned vehicle is provided, comprising the obstacle type identification device described in the second aspect, or in combination with any possible implementation manner of the second aspect.
[0052] The beneficial effects of the embodiments of the present disclosure include:
[0053] The embodiments of the present disclosure provide an obstacle type recognition method, device and unmanned vehicle, which are applied to autonomous driving scenarios. In view of the situation that dust may be misdetected as a physical obstacle during obstacle detection, in the embodiments of the present disclosure, the obstacle types that may be misdetected are first obtained, and the obstacles that may be misdetected are identified in a targeted manner; then, different dust misdetection detection strategies set for different types of obstacles are used, and the target obstacle is subjected to dust misdetection detection in combination with the obstacle data detected by the unmanned vehicle sensor, and then it is determined whether it meets the dust misdetection condition; finally, if it is found through the dust misdetection detection that the target obstacle meets the dust misdetection condition, the target obstacle is determined to be dust, and the obstacle detection result is further analyzed and judged according to the sensor detection characteristics and the misdetected target characteristics, so as to reduce the misdetection rate and improve the accuracy of obstacle type recognition, which not only provides a guarantee for the safety of the autonomous driving vehicle, but also improves the vehicle's operating efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Figure 1 A flowchart of a method for identifying obstacle types provided in an embodiment of the present disclosure;
[0055] Figure 2 A flow chart of an obstacle type identification method for isolated static obstacle type obstacles provided in an embodiment of the present disclosure;
[0056] Figure 3 A schematic diagram of a historical trajectory point passing through an isolated static obstacle type obstacle provided by an embodiment of the present disclosure;
[0057] Figure 4 A flow chart of an obstacle type identification method for dynamic obstacle type obstacles provided in an embodiment of the present disclosure;
[0058] Figure 5 A schematic diagram of a sensor full coverage area provided by an embodiment of the present disclosure;
[0059] Figure 6 A schematic diagram of an obstacle type identification device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0060] The embodiments of the present disclosure provide a method, device and unmanned vehicle for identifying obstacle types. The preferred embodiments of the present disclosure are described below in conjunction with the drawings in the specification. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present disclosure, and are not used to limit the present disclosure. In addition, the embodiments and features in the embodiments of the present application can be combined with each other if there is no conflict.
[0061] The present disclosure provides a method for identifying obstacle types. Figure 1 As shown, including:
[0062] S101, obtaining at least one obstacle type for which dust false detection detection is required; wherein the dust false detection detection is used to detect false detection behavior of falsely detecting dust as a physical obstacle;
[0063] S102, performing dust false detection detection on target obstacles having at least one obstacle type based on the dust false detection detection strategy corresponding to at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor;
[0064] S103: When the target obstacle meets the condition of dust false detection, determine that the target obstacle is dust.
[0065] In the disclosed embodiment, a more accurate recognition method is provided for obstacle recognition of an autonomous driving unmanned vehicle, which effectively reduces the occurrence of dust false detection. When implemented, the target obstacle of the obstacle type that needs to be detected by dust false detection is first obtained. Then, the corresponding dust false detection detection strategy is matched according to the obstacle type, and a series of dust false detection detection steps corresponding to the strategy are performed on each target obstacle based on the corresponding dust false detection detection strategy and the obstacle data detected by the unmanned vehicle sensor, so as to achieve targeted use of appropriate strategies for different types of target obstacles to determine whether they are dust. Finally, if the target obstacle is determined to meet the dust false detection after the detection of the corresponding strategy is completed, it can be determined that the target obstacle is dust. It can be implemented in mining scenarios, which can reduce the occurrence of false detection behaviors of falsely detecting dust as physical obstacles, thereby reducing the emergency brakes caused by false detections, improving the efficiency of autonomous driving operations, and improving economic benefits.
[0066] In yet another embodiment provided by the present disclosure, the obstacle type includes an isolated static obstacle type.
[0067] The above S102, based on the dust false detection detection strategy corresponding to at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor, respectively performs dust false detection detection on target obstacles having at least one obstacle type, can be implemented as follows:
[0068] Step 1: Determine the target obstacle that is an isolated static obstacle type from the obstacle data detected by the unmanned vehicle sensor;
[0069] Step 2: for each target obstacle, determine whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle;
[0070] Step 3: If there is an obstacle belonging to the dynamic obstacle type, then based on the historical driving trajectory information of the obstacle belonging to the dynamic obstacle type, determine whether there is a trajectory point in the historical driving trajectory points that passes through the target obstacle;
[0071] Step 4: If there is a trajectory point that passes through the target obstacle, it is determined that the target obstacle meets the dust false detection.
[0072] In the disclosed embodiment, for a target obstacle whose obstacle type is an isolated static obstacle type, a dust false detection detection strategy for isolated static obstacles is used to perform dust false detection detection on it. First, from the obstacle data, a target obstacle (e.g., a retaining wall) belonging to the isolated static obstacle type is determined. For each target obstacle belonging to the isolated static obstacle type, it is determined whether there is an obstacle belonging to a dynamic obstacle type (e.g., an unmanned vehicle, an excavator, a pedestrian, etc.) within a first distance threshold range around the target obstacle (e.g., within the first distance threshold range (e.g., 30 meters) radiating from the target obstacle as the center). If there is, it is determined whether the obstacle of the dynamic obstacle type passes through the target obstacle through the historical driving trajectory information of the obstacle of the dynamic obstacle type. It should be noted that the time corresponding to the historical driving trajectory here cannot be too long. For example, the nearest preset number of driving trajectory points of the obstacle driving trajectory points of the dynamic obstacle type can be taken as the historical driving trajectory. If there is a historical driving trajectory point that passes through the target obstacle, given the characteristics that dust can be passed through but isolated static obstacles cannot be passed through, it can be determined that the target obstacle does not meet the isolated static obstacle type, and further it can be determined that the target obstacle is a dust false detection.
[0073] In another embodiment provided by the present disclosure, before executing the above step 2: determining whether there is an obstacle of a dynamic obstacle type within the first distance threshold range around the target obstacle, the following steps are also included:
[0074] Step 1: Determine the dust type obstacles from the obstacle data;
[0075] Step 2: for each target obstacle, determine whether there is an obstacle of dust type within a second distance threshold range around the target obstacle;
[0076] For each target obstacle, it is determined whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle.
[0077] Then the above step 2, for each target obstacle, determining whether there is an obstacle of a dynamic obstacle type within the first distance threshold range around the target obstacle, may include the following steps:
[0078] Step 3: If there is an obstacle of the dust type, for each target obstacle, determine whether there is an obstacle of the dynamic obstacle type within the first distance threshold range around the target obstacle.
[0079] In the embodiment of the present disclosure, in order to further ensure the accuracy of detection, before determining whether there is an obstacle of the dynamic obstacle type within the first distance threshold range around the target obstacle, it is also possible to determine an obstacle of the dust type from the obstacle data. For the target obstacle of the isolated static obstacle type, it is determined whether there is an obstacle of the dust type within the second distance threshold range around the target obstacle (for example, within the second distance threshold (such as 10 meters) radiating from the target obstacle as the center). If there is no obstacle of the dust type in the area within the second distance threshold range around the unmanned vehicle, that is, there is no dust, then the probability that the target obstacle is a false detection of dust is extremely small, and it can be determined that the target obstacle is not a false detection of dust; if there is a dust type obstacle, subsequent detections such as whether there is an obstacle of the dynamic obstacle type within the first distance threshold range around the target obstacle can be continued.
[0080] In another embodiment provided by the present disclosure, the first distance threshold range is greater than the second distance threshold range.
[0081] In another embodiment provided by the present disclosure, the above method further includes the following steps:
[0082] Step 1: If there is no track point that crosses the target obstacle, it is determined that it is not a false dust detection; and / or
[0083] Step 2: If there is no dust-type obstacle, it is determined that it is not a false dust detection; and / or
[0084] Step 3: If there is no target obstacle, dust obstacle or dynamic obstacle in the obstacle data, it is determined that it is not a dust false detection.
[0085] In the embodiment of the present disclosure, in the above method, when the target obstacle type is an isolated static obstacle, the following implementation methods are also included to determine that the target obstacle is not a dust false detection, and the dust false detection detection of the target obstacle can be terminated:
[0086] For a target obstacle of an isolated static obstacle type, determining whether there is an obstacle of a dynamic obstacle type within a first distance threshold around the target obstacle; if there is no obstacle of the dynamic obstacle type, determining that the target obstacle is not a dust false detection;
[0087] For a target obstacle of an isolated static obstacle type, determining whether there is an obstacle of a dynamic obstacle type within a first distance threshold around the target obstacle; if there is an obstacle of a dynamic obstacle type, determining whether there is a trajectory point in the historical driving trajectory points that passes through the target obstacle based on the historical driving trajectory information of the obstacle of the dynamic obstacle type; if there is no trajectory point that passes through the target obstacle, determining that it is not a dust false detection;
[0088] For a target obstacle whose target obstacle type is an isolated static obstacle type, determining whether there is an obstacle of the dust type within a second distance threshold range around the target obstacle; if there is no obstacle of the dust type, determining that it is not a dust false detection;
[0089] For a target obstacle of an isolated static obstacle type, the obstacle data detected by the unmanned vehicle sensor is obtained. If the target obstacle, the obstacle of the dust type, or the obstacle of the dynamic obstacle type does not exist in the obstacle data, it is determined that it is not a dust false detection.
[0090] Figure 2 A flow chart of an obstacle type identification method for isolated static obstacle type obstacles provided by an embodiment of the present disclosure, such as Figure 2 As shown, the following steps are included:
[0091] S201, obtaining at least one obstacle type for which dust false detection is required;
[0092] S202: When the obstacle type is an isolated static obstacle type obstacle, determine whether the obstacle data contains a dust type obstacle, a dynamic obstacle type obstacle, and an isolated static obstacle type obstacle; if all of them are contained, proceed to step S203; if at least one type of obstacle is missing, proceed to step S209;
[0093] S203, determining a target obstacle that is an isolated static obstacle type from obstacle data detected by the unmanned vehicle sensor;
[0094] S204, determining obstacles of dust type from the obstacle data;
[0095] S205, for each target obstacle, determine whether there is an obstacle of dust type within the second distance threshold range around the target obstacle; if yes, proceed to step S206; if no, proceed to step S209;
[0096] S206: for each target obstacle, determine whether there is an obstacle of the dynamic obstacle type within the first distance threshold range around the target obstacle; if yes, proceed to step S207; if no, proceed to step S209;
[0097] S207, based on the historical driving trajectory information of the obstacle belonging to the dynamic obstacle type, determine whether there is a trajectory point in the historical driving trajectory points that passes through the target obstacle; if there is a trajectory point that passes through the target obstacle, such as Figure 3 As shown (taking the isolated static obstacle type obstacle as a retaining wall and the dynamic obstacle type obstacle as a mining truck as an example), the process proceeds to step S208. If there is no trajectory point passing through the target obstacle, the process proceeds to step S209;
[0098] S208: Determine that the target obstacle meets the dust false detection condition, and determine that the target obstacle is dust. This process ends.
[0099] S209: Determine that it is not a false dust detection. This process ends.
[0100] In yet another embodiment provided by the present disclosure, the obstacle type includes a dynamic obstacle type;
[0101] The above S102, based on the dust false detection detection strategy corresponding to at least one obstacle type, and the obstacle data detected by the unmanned vehicle sensor, respectively performs dust false detection detection on target obstacles having at least one obstacle type, can be implemented as the following steps:
[0102] Step 1: Determine the target obstacle that belongs to the dynamic obstacle type from the obstacle data detected by the unmanned vehicle sensor;
[0103] Step 2: for each target obstacle, determine the position information of the target obstacle obtained from the obstacle data, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle coordinate system;
[0104] Step 3: Determine whether the target obstacle is located in the complete coverage area of the corresponding sensor of the unmanned vehicle according to the position coordinates;
[0105] Step 4: If the target obstacle is located in a fully covered area, the area of the target obstacle is determined based on the obstacle data corresponding to the target obstacle, and when the area is less than a preset area threshold, the target obstacle is determined to be a dust false detection.
[0106] In the embodiment of the present disclosure, for a target obstacle whose obstacle type is a dynamic obstacle type, a dust false detection detection strategy of the corresponding dynamic obstacle type is used to detect whether the target obstacle is a dust false detection. First, the target obstacle belonging to the dynamic obstacle type is determined from the obstacle data detected by the unmanned vehicle sensor. Then, for each target obstacle, the position information of the target obstacle obtained from the obstacle data is determined, and the position information is converted into the position coordinates of the target obstacle in the unmanned vehicle self-vehicle coordinate system. According to the position coordinates, it is determined whether the target obstacle is located in the complete coverage area of the corresponding sensor of the unmanned vehicle. If it is not located in the complete coverage area of the corresponding sensor of the unmanned vehicle, the complete vehicle cannot be detected, which affects the subsequent judgment of the area of the target obstacle. If it is located in the complete coverage area of the corresponding sensor of the unmanned vehicle, the target obstacle can be further detected. Since the area of the dust detected when the dust is falsely detected as a dynamic obstacle type is generally much smaller than the area of the dynamic obstacle, the area of the target obstacle is determined. When the area is less than a preset area threshold (such as 5m 2 ), it is determined that the target obstacle meets the dust false detection.
[0107] In another embodiment provided by the present disclosure, if the area is located in a fully covered area, the method further includes the following steps:
[0108] Determine the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset time.
[0109] Then the above step 4, determining the area of the target obstacle based on the sensor data corresponding to the target obstacle, and determining that the target obstacle meets the dust false detection when the area is less than the preset area threshold, can be implemented as follows:
[0110] If the number is less than a preset number threshold (such as 10 times), the area of the target obstacle is determined based on the sensor data corresponding to the target obstacle, and if the area is less than the preset area threshold, it is determined that the target obstacle meets the dust false detection.
[0111] In the disclosed embodiment, since the movement of dust is irregular, whether the target obstacle is a dynamic obstacle type obstacle is determined by the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset period of time. If the number of detections is less than the preset number threshold, the target obstacle is likely to be irregularly moving dust, and it can be preliminarily determined that the target obstacle meets the dust false detection. Conversely, if the number of detections is greater than or equal to the preset number threshold, the target obstacle is likely to be a dynamic obstacle type obstacle, and it can be determined that the target obstacle is not a dust false detection. After preliminarily determining the dynamic obstacle type obstacle, further detection is performed to determine whether the area of the target obstacle is less than the preset area threshold.
[0112] In another embodiment provided by the present disclosure, the above method further includes the following steps:
[0113] Based on the detection source of the obstacle data corresponding to the target obstacle, determining whether the detection source is a single sensor;
[0114] Then the above step 2, for each target obstacle, determines the position information of the target obstacle, and converts the position information into the position coordinates of the target obstacle in the unmanned vehicle's own vehicle coordinate system, can be implemented as follows:
[0115] If it comes from a single sensor, the position information of each target obstacle is determined, and the position information is converted into the position coordinates of the target obstacle in the unmanned vehicle's own vehicle coordinate system.
[0116] In the disclosed embodiment, for the target obstacle, it is determined whether the detection source of the obstacle data corresponding to it is a single sensor. If not, the probability that the target obstacle is a dust false detection is very small, so it can be directly determined that it is not a dust false detection; if it is, the target obstacle may be a dust false detection, and further detection can be performed; therefore, for the target obstacle whose detection source is a single sensor, the position information of the target obstacle is determined, and the position information is converted into the position coordinates of the target obstacle in the unmanned vehicle self-vehicle coordinate system to facilitate subsequent detection and judgment. Taking the laser radar as an example, the position information of the obstacle point cloud data collected by the laser radar can be relative to the laser radar coordinate system, and needs to be converted into the position coordinates in the unmanned vehicle self-vehicle coordinate system to make the judgment more accurate. The unmanned vehicle self-vehicle coordinate system can be a coordinate system fixed on the body of the vehicle. In a possible implementation, the center point of the rear axle of the vehicle can be used as the coordinate origin, the direction of the vehicle is the x-axis direction, the left side of the vehicle is the y-axis direction, and the vertical upward is the z-axis direction to form a right-handed coordinate system. In a possible implementation, the laser radar coordinate system can be formed with the sensor as the origin, the direction of the unmanned vehicle as the y-axis, the right side of the unmanned vehicle as the x-axis, and the vertical upward as the z-axis to form a right-handed coordinate system. There is a certain conversion relationship between the coordinates of the vehicle coordinate system and the laser radar coordinate system, which will not be repeated here.
[0117] In another embodiment provided by the present disclosure, the above method further includes the following steps:
[0118] Step 1: If it is not from a single sensor, it is determined that it is not a false dust detection; and / or
[0119] Step 2: If it is not in the fully covered area, it is determined that it is not a false dust detection; and / or
[0120] Step 3: If the number is greater than or equal to the preset number threshold, it is determined that it is not a false dust detection; and / or
[0121] Step 4: When the area is greater than or equal to the preset area threshold, it is determined that it is not a false dust detection.
[0122] In the embodiment of the present disclosure, in the above method, when the target obstacle type is a dynamic obstacle, the following implementation methods are also included to determine that the target obstacle is not a dust false detection, and the dust false detection detection of the target obstacle can be terminated:
[0123] Based on the detection source of the obstacle data corresponding to the target obstacle, determine whether the detection source is a single sensor. If it is not from a single sensor, determine that it is not a false dust detection.
[0124] For each target obstacle, determine the position information of the target obstacle obtained from the obstacle data, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle coordinate system. According to the position coordinates, determine whether the target obstacle is located in the full coverage area of the corresponding sensor of the unmanned vehicle. If it is not located in the full coverage area, it is determined that it is not a dust false detection;
[0125] If the target obstacle is located in the fully covered area, determine the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset time. If the number is greater than or equal to the preset number threshold, it is determined that it is not a false detection of dust;
[0126] If the target obstacle is located in the fully covered area, the area of the target obstacle is determined based on the sensor data corresponding to the target obstacle, and if the area is greater than or equal to a preset area threshold, it is determined that it is not a false dust detection.
[0127] Figure 4 A flow chart of an obstacle type identification method for dynamic obstacle type obstacles provided in an embodiment of the present disclosure, such as Figure 4 As shown, the following steps are included:
[0128] S401, obtaining at least one obstacle type for which dust false detection is required;
[0129] S402, determining a target obstacle belonging to a dynamic obstacle type from obstacle data detected by the unmanned vehicle sensor;
[0130] S403, based on the detection source of the obstacle data corresponding to the target obstacle, determine whether the detection source is a single sensor, if so, proceed to step S404, if not, proceed to step S409;
[0131] S404: for each target obstacle, determine the position information of the target obstacle obtained from the obstacle data, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle coordinate system;
[0132] S405. Determine whether the target obstacle is located in the complete coverage area of the corresponding sensor of the unmanned vehicle according to the position coordinates; if it is located in the complete coverage area, such as Figure 5 As shown, go to step S406. If it is not in the full coverage area, go to step S409.
[0133] S406: Determine whether the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset time is less than a preset number threshold. If the number is less than the preset number threshold, proceed to step S407; if the number is greater than or equal to the preset number threshold, proceed to step S409;
[0134] S407, based on the obstacle data corresponding to the target obstacle, determine whether the area of the target obstacle is less than a preset area threshold, if the area is less than the preset area threshold, proceed to step 408, if the area is greater than or equal to the preset area threshold, proceed to step S409;
[0135] S408: Determine that the target obstacle meets the dust false detection condition, and determine that the target obstacle is dust. This process ends.
[0136] S409: Determine that it is not a false dust detection. This process ends.
[0137] In another embodiment provided by the present disclosure, obstacle data is stored in the form of queues; and different obstacle types are respectively stored in different obstacle queues.
[0138] The above method further comprises the following steps:
[0139] Step 1: After determining that the target obstacle is dust, change the target obstacle type to dust type, delete it from the current obstacle queue, and add it to the dust type obstacle queue.
[0140] In the disclosed embodiment, obstacle data of different obstacle types collected by the sensors of the autonomous driving vehicle are stored in the form of queues, which are composed of isolated static obstacle types, dynamic obstacle types, and dust type obstacles. When obtaining at least one obstacle type that needs to be detected for dust false detection, obtain target obstacles of isolated static obstacle types and / or dynamic obstacle types; determine the corresponding dust false detection detection strategy according to the type of the target obstacle, and perform dust false detection detection on the target obstacle according to the dust false detection detection strategy in combination with the obstacle data detected by the unmanned vehicle sensor; after determining that the target obstacle meets the dust false detection through the dust false detection detection of the corresponding dust false detection detection strategy, the target obstacle is dust and is falsely detected as a physical obstacle, so the type of the target obstacle is changed to dust type, and is deleted from the original obstacle queue and added to the dust type obstacle queue.
[0141] Based on the same disclosed concept, the embodiments of the present disclosure also provide an obstacle type identification device and an unmanned vehicle. Since the principles of the problems solved by these devices and the unmanned vehicle are similar to those of the aforementioned obstacle type identification method, the implementation of the device and the unmanned vehicle can refer to the implementation of the aforementioned method, and the repeated parts will not be repeated.
[0142] With the above Figure 1 Corresponding to the method shown, the embodiment of the present disclosure also provides a device for identifying obstacle types, such as Figure 6 As shown, including:
[0143] The obstacle acquisition module 601 is used to acquire at least one obstacle type that needs to be detected by dust false detection; wherein the dust false detection is used to detect the false detection behavior of falsely detecting dust as a physical obstacle;
[0144] A type identification module 602 is used to perform dust false detection detection on target obstacles having at least one obstacle type based on the dust false detection detection strategy corresponding to at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor;
[0145] The type determination module 603 is used to determine that the target obstacle is dust when the target obstacle meets the dust false detection condition.
[0146] In yet another embodiment provided by the present disclosure, the obstacle type includes an isolated static obstacle type.
[0147] The type identification module 602 is used to determine the target obstacle belonging to the isolated static obstacle type from the obstacle data detected by the unmanned vehicle sensor; for each target obstacle, determine whether there is an obstacle belonging to the dynamic obstacle type within a first distance threshold range around the target obstacle; if there is an obstacle belonging to the dynamic obstacle type, determine whether there is a trajectory point in the historical driving trajectory points that passes through the target obstacle based on the historical driving trajectory information of the obstacle belonging to the dynamic obstacle type; if there is a trajectory point that passes through the target obstacle, determine that the target obstacle meets the dust false detection.
[0148] In another embodiment provided by the present disclosure, before determining whether there is an obstacle belonging to the dynamic obstacle type within the first distance threshold range around the target obstacle, the type identification module 602 is further configured to: determine an obstacle belonging to the dust type from the obstacle data; for each target obstacle, determine whether there is an obstacle belonging to the dust type within the second distance threshold range around the target obstacle;
[0149] The type identification module 602 is used to determine, for each target obstacle, whether there is an obstacle of the dynamic obstacle type within a first distance threshold range around the target obstacle within a first preset time period based on the spatiotemporal information of the target obstacle obtained from the obstacle data and the spatiotemporal information of the obstacles of the dynamic obstacle type if there is an obstacle of the dust type.
[0150] In yet another embodiment provided by the present disclosure, the first distance threshold range is greater than the second distance threshold range.
[0151] In another embodiment provided by the present disclosure, the type determination module 603 is also used to determine that it is not a false dust detection if there is no trajectory point crossing the target obstacle; and / or determine that it is not a false dust detection if there is no obstacle belonging to the dust type; and / or determine that it is not a false dust detection if there is no target obstacle belonging to the dust type or the dynamic obstacle type in the obstacle data.
[0152] In yet another embodiment provided by the present disclosure, the obstacle type includes a dynamic obstacle type;
[0153] The type identification module 602 is used to: determine the target obstacles belonging to the dynamic obstacle type from the obstacle data detected by the unmanned vehicle sensor; for each target obstacle, determine the position information of the target obstacle obtained from the obstacle data, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle coordinate system; determine whether the target obstacle is located in the full coverage area of the corresponding sensor of the unmanned vehicle according to the position coordinates; if it is located in the full coverage area, determine the area of the target obstacle based on the obstacle data corresponding to the target obstacle, and determine that the target obstacle meets the dust false detection when the area is less than a preset area threshold.
[0154] In another embodiment provided by the present disclosure, the type identification module 602 is also used to: if located in a fully covered area, determine the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset time; if the number is less than a preset number threshold, determine the area of the target obstacle based on the sensor data corresponding to the target obstacle, and if the area is less than the preset area threshold, determine that the target obstacle meets the dust false detection.
[0155] In another embodiment provided by the present disclosure, the type identification module 602 is also used to: determine whether the detection source is a single sensor based on the detection source of the obstacle data corresponding to the target obstacle; if it is from a single sensor, then for each target obstacle, determine the position information of the target obstacle, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle coordinate system.
[0156] In another embodiment provided by the present disclosure, the type determination module 603 is also used to: if it does not come from a single sensor, determine that it is not a false dust detection; and / or if it is not located in a fully covered area, determine that it is not a false dust detection; and / or if the number of times is greater than or equal to a preset number threshold, determine that it is not a false dust detection; and / or if the area is greater than or equal to a preset area threshold, determine that it is not a false dust detection.
[0157] In another embodiment provided by the present disclosure, obstacle data is stored in the form of queues; and different obstacle types are respectively stored in different obstacle queues.
[0158] The type determination module 603 is further used for:
[0159] After determining that the target obstacle is dust, the target obstacle type is changed to a dust type, and is deleted from the current obstacle queue and added to a dust type obstacle queue.
[0160] An unmanned vehicle provided by an embodiment of the present disclosure includes an obstacle type recognition device as described in any of the above embodiments.
[0161] Through the description of the above implementation methods, those skilled in the art can clearly understand that the embodiments of the present disclosure can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present disclosure.
[0162] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present disclosure.
[0163] Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description of the embodiment, or can be changed accordingly and located in one or more devices different from the present embodiment. The modules in the above embodiment can be combined into one module, or can be further divided into multiple sub-modules.
[0164] The serial numbers of the above-mentioned embodiments of the present disclosure are only for description and do not represent the advantages or disadvantages of the embodiments.
[0165] Obviously, those skilled in the art can make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is also intended to include these modifications and variations.
Claims
1. A method for identifying obstacle types, characterized in that: include: Acquire at least one obstacle type for which dust false detection detection is required; wherein the dust false detection detection is used to detect false detection behavior of falsely detecting dust as a physical obstacle; Based on the dust false detection detection strategies corresponding to the at least one obstacle type, and the obstacle data detected by the unmanned vehicle sensor, respectively, performing dust false detection detection on the target obstacles having the at least one obstacle type; In the case where the target obstacle meets the dust false detection condition, it is determined that the target obstacle is dust.
2. The method according to claim 1, characterized in that The obstacle types include isolated static obstacle types; The dust false detection detection strategy corresponding to the at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor respectively performs dust false detection detection on the target obstacle having the at least one obstacle type, including: Determining a target obstacle belonging to the isolated static obstacle type from obstacle data detected by the unmanned vehicle sensor; For each target obstacle, determining whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle; If there is an obstacle belonging to the dynamic obstacle type, determining whether there is a trajectory point in the historical driving trajectory points that passes through the target obstacle based on the historical driving trajectory information of the obstacle belonging to the dynamic obstacle type; If there is a trajectory point that passes through the target obstacle, it is determined that the target obstacle meets the dust false detection.
3. The method according to claim 2, characterized in that Before determining whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle, the method further includes: Determining, from the obstacle data, obstacles that are of the dust type; For each target obstacle, determining whether there is an obstacle of dust type within a second distance threshold range around the target obstacle; The step of determining, for each target obstacle, whether there is an obstacle of a dynamic obstacle type within a first distance threshold range around the target obstacle includes: If there is an obstacle of the dust type, then for each target obstacle, it is determined whether there is an obstacle of the dynamic obstacle type within the first distance threshold range around the target obstacle.
4. The method according to claim 3, characterized in that The first distance threshold range is greater than the second distance threshold range.
5. The method according to claim 3 or 4, characterized in that Also includes: If there is no track point passing through the target obstacle, it is determined that it is not a dust false detection; and / or If there is no dust-type obstacle, it is determined that it is not a false dust detection; and / or If the target obstacle, the obstacle of the dust type, or the obstacle of the dynamic obstacle type does not exist in the obstacle data, it is determined that it is not a dust false detection.
6. The method according to claim 1, characterized in that The obstacle type includes a dynamic obstacle type; The dust false detection detection strategy corresponding to the at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor respectively performs dust false detection detection on the target obstacle having the at least one obstacle type, including: Determining a target obstacle belonging to the dynamic obstacle type from obstacle data detected by the unmanned vehicle sensor; For each target obstacle, determine the position information of the target obstacle obtained from the obstacle data, and convert the position information into the position coordinates of the target obstacle in the unmanned vehicle self-vehicle coordinate system; Determining, according to the position coordinates, whether the target obstacle is located in a complete coverage area of the corresponding sensor of the unmanned vehicle; If it is located in the fully covered area, the area of the target obstacle is determined based on the obstacle data corresponding to the target obstacle, and when the area is less than a preset area threshold, it is determined that the target obstacle meets the dust false detection.
7. The method according to claim 6, characterized in that If the area is located in the full coverage area, the method further includes: Determine the number of times the target obstacle is detected by the unmanned vehicle sensor within a preset time; The determining, based on the sensor data corresponding to the target obstacle, the area of the target obstacle, and determining that the target obstacle meets the dust false detection condition when the area is less than a preset area threshold, includes: If the number is less than a preset number threshold, the area of the target obstacle is determined based on the sensor data corresponding to the target obstacle, and when the area is less than a preset area threshold, it is determined that the target obstacle meets the dust false detection.
8. The method according to claim 7, characterized in that Also includes: Based on a detection source of obstacle data corresponding to the target obstacle, determining whether the detection source is a single sensor; For each target obstacle, determining the position information of the target obstacle and converting the position information into the position coordinates of the target obstacle in the self-driving vehicle coordinate system includes: If it comes from a single sensor, the position information of each target obstacle is determined, and the position information is converted into the position coordinates of the target obstacle in the coordinate system of the unmanned vehicle.
9. The method according to claim 8, characterized in that The method further comprises: If it is not from a single sensor, it is determined not to be a false dust detection; and / or If it is not located in the fully covered area, it is determined that it is not a false dust detection; and / or If the number is greater than or equal to a preset number threshold, it is determined that it is not a false dust detection; and / or When the area is greater than or equal to the preset area threshold, it is determined that it is not a false dust detection.
10. The method according to claim 1, characterized in that The obstacle data is stored in the form of a queue; and different obstacle types are stored in different obstacle queues respectively; The method further comprises: After determining that the target obstacle is dust, the target obstacle type is changed to a dust type, deleted from the current obstacle queue, and added to a dust type obstacle queue.
11. An obstacle type identification device, characterized in that: include: An obstacle acquisition module is used to acquire at least one obstacle type that needs to be tested for dust false detection; wherein the dust false detection is used to detect false detection behavior of falsely detecting dust as a physical obstacle; A type identification module, configured to perform dust false detection detection on target obstacles having the at least one obstacle type based on the dust false detection detection strategies corresponding to the at least one obstacle type and the obstacle data detected by the unmanned vehicle sensor; The type determination module is used to determine that the target obstacle is dust when the target obstacle meets the dust false detection condition.
12. An unmanned vehicle, characterized in that: The device comprises the obstacle type identification device as claimed in claim 11.