Obstacle detection method, laser radar system and computer readable storage medium

Abstract

The embodiment of the invention discloses an obstacle detection method, a laser radar system and a computer readable storage medium, and the method comprises the steps: obtaining point clouds collected by a first radar and a second radar in N collection periods according to a time sequence, and N is an integer greater than 1; obtaining a main grid map of the Nth acquisition period according to the point cloud acquired by the first radar in the Nth acquisition period, the point cloud acquired by the second radar in the Nth acquisition period and the main grid map of the (N-1) th acquisition period; and performing clustering analysis on the main grid map of the Nth acquisition period to obtain obstacle information in the view field area of the second radar. The sensing result of the first radar is used as priori knowledge to be fused with the point cloud of the second radar, and the historical obstacle information containing the time sequence is combined, so that the obstacle detection precision of the view field area of the second radar can be improved on the basis of saving the computing power.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of laser radar technology, and in particular, to an obstacle detection method, a laser radar system, and a computer-readable storage medium. Background Art

[0002] With the development of science and technology, LiDAR is widely used in the fields of autonomous driving, intelligent robot navigation, drones and other intelligent equipment. LiDAR is a sensor device that emits laser beams to detect the position, speed and other characteristic quantities of target objects. Its working principle is to first emit a detection laser beam to the target object, then compare the reflected signal received from the target object with the transmitted signal, and obtain relevant information about the target object after processing.

[0003] In the state of autonomous driving or assisted driving, the vehicle's perception of the surrounding environment is very important. Usually, a lidar system including multiple lidars is used to provide full-range perception capabilities from near to far. However, the current traditional obstacle perception algorithms and lidar systems are difficult to balance detection accuracy and computing power costs. Summary of the invention

[0004] In view of this, the embodiments of the present application provide an obstacle detection method, a laser radar system, and a computer-readable storage medium, which improve the perception accuracy of obstacles while effectively saving computing power.

[0005] In a first aspect, an embodiment of the present application provides an obstacle detection method, which is applied to a laser radar system, wherein the laser radar system includes a first radar and a second radar, and the method includes: obtaining a point cloud collected by the first radar and the second radar in N acquisition cycles in chronological order, wherein N is an integer greater than 1; obtaining a main grid map of the N acquisition cycle based on the point cloud collected by the first radar in the Nth acquisition cycle, the point cloud collected by the second radar in the Nth acquisition cycle, and the main grid map of the N-1th acquisition cycle, wherein the main grid map of the N-1th acquisition cycle is determined based on the point clouds collected by the first radar and the second radar in the previous N-1 acquisition cycles; and performing cluster analysis on the main grid map of the Nth acquisition cycle to obtain obstacle information within the field of view of the second radar.

[0006] In the above technical solution, the point clouds collected by the first radar and the second radar in the first N-1 acquisition cycles represent the historical obstacle information sensed by the first radar and the second radar. Since the field of view area of ​​the first radar and the field of view area of ​​the second radar overlap in time sequence, the main grid map of the N-1th acquisition cycle is used in combination with the point cloud collected by the first radar in the Nth acquisition cycle and the point cloud collected by the second radar in the Nth acquisition cycle to determine the main grid map of the Nth acquisition cycle. The historical obstacle information sensed by the first radar can be used as prior knowledge, so that the main grid map of the Nth acquisition cycle can more accurately reflect the obstacle information in the field of view area of ​​the second radar, which is conducive to completing the obstacle detection in the field of view area of ​​the second radar with high precision. On the one hand, using the historical obstacle information sensed by the first radar as prior knowledge can not only improve the perception accuracy of obstacles in the field of view area of ​​the second radar, but also effectively save computing power. On the other hand, by generating the main grid map of the Nth acquisition cycle, reducing the three-dimensional point cloud into a two-dimensional grid map, and identifying obstacles in two-dimensional space, the time complexity and space complexity of the target detection algorithm can be effectively reduced.

[0007] In some embodiments, the above-mentioned main grid map of the Nth acquisition period is obtained based on the point cloud obtained by the first radar in the Nth acquisition period, the point cloud obtained by the second radar in the Nth acquisition period, and the main grid map of the N-1th acquisition period, including: obtaining the first radar observation grid layer according to the point cloud obtained by the first radar in the Nth acquisition period; obtaining the second radar observation grid layer according to the point cloud obtained by the second radar in the Nth acquisition period; obtaining the main grid map observation layer of the Nth acquisition period according to the first radar observation grid layer and the second radar observation grid layer; obtaining the main grid map of the Nth acquisition period according to the main grid map observation layer of the Nth acquisition period and the main grid map of the N-1th acquisition period.

[0008] In the above technical solution, the first radar observation grid layer and the second radar observation grid layer are superimposed to obtain the main grid map observation layer, and the main grid map of the N-1th acquisition cycle is updated through the main grid map observation layer to obtain the main grid map of the Nth acquisition cycle, so that the main grid map has time sequence. The main grid map layer is updated in real time through the point cloud collected by the second radar in the Nth acquisition cycle and the point cloud collected by the second radar in the Nth acquisition cycle, which can reflect the current obstacle information in the field of view of the second radar. The introduced time sequence information is conducive to the subsequent accurate identification of obstacles in the field of view of the second radar.

[0009] In some embodiments, the above-mentioned obtaining the first radar observation grid layer according to the point cloud collected by the first radar in the Nth collection cycle includes: rasterizing the point cloud collected by the first radar in the Nth collection cycle to obtain a first grid, wherein the first grid is marked with a grid height and a height difference, the grid height is the median of the heights of the point cloud points in the grid, and the height difference is the absolute value of the difference between the maximum and minimum values ​​of the heights of the point cloud points in the grid; marking the obstacle attributes of the first grid according to the obstacle information of the point cloud collected by the first radar in the Nth collection cycle, wherein the obstacle attributes include an obstacle grid or a non-obstacle grid; determining and marking the area to which the first grid belongs according to the grid height, height difference and obstacle attributes of the first grid to obtain the first radar observation grid layer, wherein the area includes a perception area of ​​interest or a perception area of ​​no interest.

[0010] In the above technical solution, the grid layer corresponding to the point cloud collected by the first radar in the Nth collection cycle is divided into a focus perception area and a non-focus perception area through grid height, height difference and obstacle attributes, so as to distinguish obstacles in the first radar field of view from roadside fixed objects such as green belts or steps, which is conducive to accurately identifying obstacles in the second radar field of view and reducing interference.

[0011] In some embodiments, a second radar observation grid layer is obtained based on the point cloud collected by the second radar in the Nth collection cycle, including: rasterizing the point cloud collected by the second radar in the Nth collection cycle to obtain a second grid, wherein the second grid is marked with a grid height and a height difference; determining and marking the ground attributes of the second grid based on the grid height and the height difference of the second grid to obtain the second radar observation grid layer, wherein the ground attributes include ground grids or non-ground grids.

[0012] In the above technical solution, the grid height and height difference are combined to distinguish the ground and non-ground in the field of view of the second radar, which is helpful to avoid interference of the ground area on obstacle perception.

[0013] In some embodiments, determining and marking the area to which the first grid belongs based on the grid height, height difference and obstacle attributes of the first grid includes: determining and marking the area to which the first grid that meets the first condition belongs as a non-focus perception area, and determining and marking the area to which the first grid that does not meet the first condition belongs as a focus perception area; wherein the first condition is that the grid height is greater than or equal to a first threshold, the height difference is greater than or equal to a second threshold, and the obstacle attribute is a non-obstacle grid, and the first threshold and the second threshold are determined based on the point cloud collected by the first radar in the first N-1 collection cycles.

[0014] In the above technical solution, the area to which the first grid belongs is segmented by the first threshold and the second threshold, thereby focusing on the perception area and improving the pertinence and accuracy of obstacle detection.

[0015] In some embodiments, the aforementioned determining and marking the ground attributes of the second grid based on the grid height and height difference of the second grid includes: determining and marking the second grid that meets the second condition as a non-ground grid, and determining and marking the second grid that does not meet the second condition as a ground grid; wherein the second condition is that the grid height is greater than or equal to a third threshold, or the height difference is greater than or equal to a fourth threshold, and the third threshold and the fourth threshold are determined based on the point cloud collected by the second radar in the first N-1 collection cycles.

[0016] In the above technical solution, the ground attribute to which the second grid belongs is determined by the third threshold and the fourth threshold, thereby accurately dividing the ground grid and the non-ground grid, thereby improving the pertinence and accuracy of obstacle detection.

[0017] In some embodiments, the above-mentioned obtaining the main grid map observation layer based on the first radar observation grid layer and the second radar observation grid layer includes: superimposing the first radar observation grid layer and the second radar observation grid layer to obtain a fused grid map layer; determining and marking the obstacle attributes of the fused grid map layer based on the area to which the first radar observation grid layer belongs and the ground attributes of the second radar observation grid layer to obtain the main grid map observation layer.

[0018] In the above technical solution, the obstacle attributes of the fused grid map layer are determined dually by the regional attributes of the first radar observation grid layer and the ground attributes of the second radar observation grid layer, which is beneficial to improving the accuracy and reliability of obstacle detection.

[0019] In a second aspect, an embodiment of the present application provides a laser radar system, including: A memory for storing executable program codes; The processor is used to call and run the executable program code from the memory so that the laser radar system executes the obstacle detection method of the first aspect as described above.

[0020] In a third aspect, some embodiments of the present application provide a computer-readable storage medium, which stores a computer program. When the computer program is executed, the obstacle detection method of the first aspect is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0022] Figure 1 is a schematic diagram of an application environment of an obstacle detection method in some embodiments of the present application; Figure 2 is a schematic diagram of the structure of a laser radar system in some embodiments of the present application; Figure 3 is a flowchart of an obstacle detection method in some embodiments of the present application; Figure 4 is a schematic diagram of the field of view area of ​​the first radar and the second radar in some embodiments of the present application; Figure 5 Schematic diagram of the flow of obstacle detection methods in some other embodiments of the present application.

[0023] Reference numerals 300, driving environment, 301, obstacles; 200. Vehicles; 100. LiDAR system; 10. First radar, 20. Second radar, 40. Processor, 50. Memory, 60. Bus. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0025] Some obstacle detection schemes known to the inventors of this application mostly use target detection algorithms based on traditional point cloud clustering or perception algorithms based on deep learning to calculate and process three-dimensional point clouds to identify obstacle information. However, target detection algorithms based on traditional point cloud clustering have low accuracy and robustness, and cannot guarantee good timeliness; perception algorithms based on deep learning have high training costs, require more computing power during operation, and are more sensitive to data. For autonomous driving vehicles, the allocation of sensors and computing power will be emphasized. Forward and rearward perception tasks generally receive a larger operating computing power and cost ratio, while side blindness tasks often receive less computing power support, resulting in the timeliness and accuracy of side obstacle detection being difficult to meet expected requirements when the hardware configuration and processor computing power are limited.

[0026] In response to the above technical problems, the embodiments of the present application provide an obstacle detection method, a laser radar system and a computer-readable storage medium, which improve the perception accuracy of obstacles while effectively saving computing power.

[0027] The following uses the application of a laser radar system in an autonomous driving vehicle as an example to illustrate the application scenario of the laser radar system in this application. Figure 1 As shown, the laser radar system 100 is installed on the vehicle 200. The vehicle 200 detects obstacles 301 in the driving environment 300 through the laser radar system 100, so that the vehicle 200 can plan the driving path according to the detected information to avoid collision with the obstacle 301. It can be understood that the obstacle 301 can be other vehicles, pedestrians, bicycles or electric motorcycles, etc., and the vehicle 200 can be a passenger car, a truck, a cleaning car or a detection car, etc. Since the field of view of a single laser radar is limited, the laser radar system 100 includes multiple laser radars, which can be set according to the detection needs to perform multi-directional or omnidirectional detection of obstacles around the vehicle 200.

[0028] The obstacle detection method in the embodiment of the present application is executed by a processor of the laser radar system. Before introducing the obstacle detection method in the embodiment of the present application, the laser radar system provided in the embodiment of the present application is first introduced in detail.

[0029] See also Figure 2 The laser radar system 100 includes a first radar 10, a second radar 20, a processor 40 and a memory 50. The first radar 10, the second radar 20, the processor 40 and the memory 50 are communicatively connected via a bus 60.

[0030] The memory 50 is used to store executable program codes. The memory 50, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the obstacle detection method in the embodiment of the present application. In some embodiments, the memory 50 includes a program storage area and a data storage area, wherein the program storage area is used to store the application required by the operating system or at least one function, and the data storage area is used to store data created according to the use of the terminal. In some embodiments, the memory 50 may include a high-speed random access memory or a non-volatile memory, for example, at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 50 includes a remote memory remotely arranged relative to the processor 40, and the remote memory is used to connect to the terminal via a network. The above-mentioned network includes but is not limited to the Internet, an intranet, a local area network, a mobile communication network and a combination thereof.

[0031] The processor 40 is used to call and run the executable program code from the memory 50, so that the laser radar system 100 performs the obstacle detection method in any embodiment of the present application. In some embodiments, the processor 40 includes one or more of a central processing unit (CPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a controller, or a field programmable gate array (FPGA) device. The processor 40 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration.

[0032] In some embodiments, the laser radar system 100 also includes other hardware and software that support operation. The hardware may include a power supply, an antenna, a temperature control system, or a global positioning system (GPS) / (Inertial Measurement Unit, IMU) combined navigation system, etc. The software may include a radar driver, a system control software, a data processing software, or a simultaneous localization and mapping (SLAM) software, etc. Among them, the radar driver is used to drive the radar hardware to collect data; the system control software is used to control the settings of the radar's scanning mode, frequency, field of view and other parameters; the data processing software is used for signal filtering, time synchronization or coordinate transformation, etc.; the SLAM software is used for positioning and mapping.

[0033] Understandably, Figure 2Only one first radar 10 and one second radar 20 are used as examples. The number of first radars 10 can be one or more, and the number of second radars 20 can be one or more. The number of radars in the laser radar system and the field of view of each radar can be set according to actual conditions. Among them, the first radar is used to detect obstacles in the front or rear direction of the vehicle, and the second radar is used to detect obstacles on the side of the vehicle. In some embodiments, the detection distance and detection accuracy of the first radar are higher than those of the second radar.

[0034] In some embodiments, the laser radar system 100 includes two first radars and two second radars, and the horizontal field of view angles of the two first radars are the same and face opposite directions. In some embodiments, the laser radar system 100 includes one first radar and two second radars, wherein the first radar is used for forward medium and long-range detection, and the two second radars are used for close-range detection on the left and right sides, respectively.

[0035] In one embodiment, the horizontal field of view (HFOV) of the first radar is 120°, and the vertical field of view (VFOV) is 25°. The horizontal field of view (HFOV) of the second radar is 180°, and the vertical field of view (VFOV) is 90°.

[0036] The obstacle detection method provided in the embodiment of the present application is described in detail below in conjunction with the exemplary application and implementation of the laser radar system 100 provided in the embodiment of the present application. Figure 3 The obstacle detection method S100 includes but is not limited to the following steps: S10: Obtaining point clouds collected by the first radar and the second radar in N collection cycles in chronological order, where N is an integer greater than 1.

[0037] The above-mentioned acquisition cycle can be preset according to actual needs. In a possible implementation, the time required for the first radar or the second radar to collect a frame of point cloud is taken as an acquisition cycle. Correspondingly, the point cloud collected by the first radar in one acquisition cycle can be understood as 1 frame of point cloud collected by the first radar, and the point cloud collected by the second radar in one acquisition cycle can be understood as 1 frame of point cloud collected by the second radar. The system control software of the laser radar system can control the first radar and the second radar to scan at the same frame rate, generate corresponding point cloud frames each time scanning, and synchronize the point clouds collected by the first radar and the second radar. Exemplarily, the frame rate can be 600 frames per second. The point clouds collected by the first radar and the second radar in N acquisition cycles in sequence include multiple frames of point clouds collected by the first radar and arranged in time order, and multiple frames of point clouds collected by the second radar and arranged in time order. The point cloud collected by the first radar in the Nth acquisition cycle refers to the latest frame of point cloud generated by the first radar scanning at the current moment, and the point cloud collected by the second radar in the Nth acquisition cycle refers to the latest frame of point cloud generated by the second radar scanning at the current moment. When the laser radar system 100 is running, the processor 40 calculates and processes the point clouds collected by the first radar and the second radar within N collection cycles at a certain frequency. The current moment can refer to any moment when the processor 40 performs calculation and processing.

[0038] S20: Obtain a main grid map of the Nth acquisition cycle based on the point cloud collected by the first radar in the Nth acquisition cycle, the point cloud collected by the second radar in the Nth acquisition cycle, and the main grid map of the N-1th acquisition cycle, wherein the main grid map of the N-1th acquisition cycle is determined based on the point clouds collected by the first radar and the second radar in the previous N-1 acquisition cycles.

[0039] The point cloud collected by the first radar in the first N-1 acquisition cycles includes multiple frames of point clouds whose timing is before the point cloud collected in the Nth acquisition cycle; the point cloud collected by the second radar in the first N-1 acquisition cycle includes multiple frames of point clouds whose timing is before the point cloud collected in the Nth acquisition cycle. It can be understood that the main grid map is continuously updated with the acquisition cycle. Specifically, the update frequency of the main grid map is the same as the frame rate of the point cloud. The main grid map of the N-1th acquisition cycle refers to the main grid map corresponding to the point cloud collected by the first radar in the N-1th acquisition cycle or the point cloud collected by the second radar in the N-1th acquisition cycle. The main grid map of the Nth acquisition cycle refers to the main grid map corresponding to the point cloud collected by the first radar in the Nth acquisition cycle or the point cloud collected by the second radar in the Nth acquisition cycle. Since the point cloud collected by the first radar in the first N-1 acquisition cycles can reflect the historical obstacle information in the field of view of the first radar, and the point cloud collected by the second radar in the first N-1 acquisition cycles can reflect the historical obstacle information in the field of view of the second radar, the main grid map of the N-1th acquisition cycle can reflect the historical obstacle information in the field of view of both the first radar and the second radar.

[0040] It can be understood that, on the one hand, in some embodiments, the point cloud collected by the first radar in the Nth collection cycle and the point cloud collected by the second radar in the Nth collection cycle have an intersection in the field of view area, and the point cloud collected by the first radar in the Nth collection cycle can reflect the obstacle information in the partial field of view area of ​​the second radar. Therefore, the point cloud collected by the first radar in the Nth collection cycle can be used to identify obstacles in the field of view area of ​​the second radar. On the other hand, since the laser radar system is mounted on a mobile device, as the mobile device moves, the field of view area of ​​the first radar and the field of view area of ​​the second radar overlap in time sequence. The laser radar system is applied to Figure 1 Taking the application environment shown as an example, as the vehicle 200 moves forward, the field of view of the first radar will continue to fall into the field of view of the second radar. That is to say, for obstacles in front of the left or right of the vehicle 200, which are currently in the scanning range of the first radar, they will enter the scanning range of the second radar at the next moment or at a certain moment in the future. Thus, the current obstacle information in the field of view of the second radar can be determined with reference to the historical obstacle information in the field of view of the first radar. The main grid map of the N-1th acquisition cycle can effectively make up for the area in the point cloud collected by the second radar in the Nth acquisition cycle that does not intersect with the point cloud collected by the first radar in the Nth acquisition cycle, that is, effectively make up for the area in the field of view of the second radar that does not intersect with the field of view of the first radar.

[0041] It can be seen that Figure 4As shown, for the first area where the point cloud collected by the first radar in the Nth collection cycle and the point cloud collected by the second radar in the Nth collection cycle intersect in the field of view, the obstacle information in the first area can be determined by referring to the point cloud collected by the first radar in the Nth collection cycle and the point cloud collected by the second radar in the Nth collection cycle; for the second area where the point cloud collected by the first radar in the Nth collection cycle and the point cloud collected by the second radar in the Nth collection cycle do not intersect in the field of view, the obstacle information in the second area can be determined by referring to the main grid map of the N-1th collection cycle. In this way, the main grid map of the Nth collection cycle can accurately reflect the obstacle information in the field of view of the second radar.

[0042] S30: performing cluster analysis on the main grid map of the Nth acquisition cycle to obtain obstacle information within the field of view of the second radar.

[0043] Based on the overlap between the field of view of the first radar and the field of view of the second radar in time series, the historical obstacle information provided by the main grid map of the N-1th acquisition period is used as prior knowledge. The point cloud collected by the first radar in the Nth acquisition period and the point cloud collected by the second radar in the Nth acquisition period are combined to obtain the main grid map of the Nth acquisition period that can reflect the obstacle information in the field of view of the second radar. This is conducive to real-time and accurate identification of obstacles in the field of view of the second radar, and subsequent tracking of obstacles based on the time series information.

[0044] Perform cluster analysis on the main grid map of the Nth acquisition cycle to obtain obstacle information in the field of view of the second radar. Cluster analysis refers to using a cluster analysis algorithm to divide the obstacle grids in the main grid map of the Nth acquisition cycle into multiple groups based on the distance between the obstacle grids, and each group represents an obstacle. The cluster analysis algorithm can be a K-means algorithm, a hierarchical clustering algorithm, or a meanshift algorithm. It can be understood that a group represents an obstacle or multiple overlapping obstacles. In some embodiments, the center coordinates of the group and the contour shape on the map can be used as the corresponding obstacle information.

[0045] In some embodiments, the clustering results (the aforementioned multiple groups) are fitted with bounding boxes. Specifically, the minimum circumscribed rectangle of each group is generated as a bounding box. In this embodiment, the center coordinates and length and width of the bounding box can be used as the corresponding obstacle information. In some embodiments, the position of the obstacle point cloud in the bounding box can be optimized to obtain more accurate obstacle information. Among them, the position optimization can be performed using the point cloud ABB bounding box (Axis-Aligned Bounding Box) optimization algorithm.

[0046] Based on the cluster analysis results, the obstacle information in the field of view of the second radar is determined. On the one hand, the historical obstacle information sensed by the first radar is used as prior knowledge, which can effectively avoid the interference of non-dynamic obstacles on the cluster analysis of dynamic obstacles. On the other hand, by generating the main grid map of the Nth acquisition cycle, the three-dimensional point cloud is reduced to a two-dimensional grid map, and obstacles are identified in two-dimensional space, which can effectively reduce the time and space complexity of the recognition algorithm, thereby saving computing power.

[0047] In some embodiments, obtaining the main grid map of the Nth acquisition period according to the point cloud acquired by the first radar in the Nth acquisition period, the point cloud acquired by the second radar in the Nth acquisition period, and the main grid map of the N-1th acquisition period specifically includes: S21: obtaining a first radar observation grid layer according to the point cloud collected by the first radar in the Nth collection period; obtaining a second radar observation grid layer according to the point cloud collected by the second radar in the Nth collection period.

[0048] As the acquisition cycle is continuously updated, the point cloud of the Nth acquisition cycle is usually used to guide the mobile device (e.g., vehicle 200) to avoid obstacles or plan a path. That is, the point cloud of the Nth acquisition cycle is an observation point cloud. In some embodiments, the data used for current observation perception can be divided into observation layers. Exemplarily, the observation layer includes the point cloud of the Nth acquisition cycle or other data (e.g., a grid map) generated based on the point cloud of the Nth acquisition cycle.

[0049] The grid map obtained by rasterizing the point cloud collected by the first radar in the Nth acquisition cycle is the first radar observation grid layer. Here, the first radar observation grid layer refers to the grid map of the first radar used for current observation perception and the latest in time series. Similarly, the grid map obtained by rasterizing the point cloud collected by the second radar in the Nth acquisition cycle is the second radar observation grid layer. Here, the second radar observation grid layer refers to the grid map of the second radar used for current observation perception and the latest in time series.

[0050] S22: Obtaining a main grid map observation layer of the Nth acquisition period according to the first radar observation grid layer and the second radar observation grid layer.

[0051] The first radar observation grid layer and the second radar observation grid layer are merged to obtain the main grid map observation layer of the Nth acquisition cycle, so that the main grid map observation layer of the Nth acquisition cycle includes the area to which each grid recorded in the first radar observation grid layer belongs and the ground attributes of each grid recorded in the second radar observation grid layer. It can be understood that for the field of view area where the first radar and the second radar do not overlap, the first radar observation grid layer and the second radar observation grid layer are spliced; for the field of view area where the first radar and the second radar overlap, the regional attributes of the grid in the first radar observation grid layer and the ground attributes of the grid in the second radar observation grid layer can be integrated. As the first radar observation grid layer and the second radar observation grid layer are continuously updated with the point cloud acquisition cycle, the main grid map observation layer is also continuously updated with the point cloud acquisition cycle.

[0052] S23: Obtaining the main grid map of the Nth acquisition period according to the main grid map observation layer of the Nth acquisition period and the main grid map of the N-1th acquisition period.

[0053] The main grid map of the Nth acquisition period obtained by combining the main grid map observation layer of the Nth acquisition period and the main grid map of the N-1th acquisition period has time sequence, can reflect the obstacle information in the field of view of the second radar, and is conducive to the subsequent tracking of dynamic obstacles in the field of view of the second radar.

[0054] In some embodiments, Figure 4 As shown, the main grid map covers the area within 10 meters in front and behind and 20 meters in left and right of the vehicle 200, and the area includes a partial field of view area of ​​the first radar and at least a partial field of view area of ​​the second radar. When the laser radar system is started, the main grid map is a blank grid map. As the point cloud acquisition cycle is updated, the main grid map of the Nth acquisition cycle is generated according to the main grid map observation layer of the Nth acquisition cycle and the main grid map of the N-1th acquisition cycle. Exemplarily, the main grid map of the 0th acquisition cycle is the initial blank grid map, the main grid map of the 1st acquisition cycle is the main grid map observation layer of the 1st acquisition cycle, the main grid map of the 2nd acquisition cycle is obtained according to the main grid map observation layer of the 2nd acquisition cycle and the main grid map of the 1st acquisition cycle, the main grid map of the 3rd acquisition cycle is obtained according to the main grid observation layer of the 3rd acquisition cycle and the main grid map of the 2nd acquisition cycle, and so on.

[0055] Specifically, in some embodiments, the grid k in the observation layer of the main grid map of the Nth acquisition cycle and the grid k' in the main grid map of the N-1th acquisition cycle represent the same position in the field of view of the laser radar system, and the obstacle attributes of the grid k' are updated according to the obstacle attributes of the grid k and the obstacle attributes of the grid k'. When the obstacle attributes of each grid in the main grid map of the N-1th acquisition cycle are updated, the main grid map of the Nth acquisition cycle is obtained. In some embodiments, a static binary Bayesian filter is used to calculate the updated obstacle attributes based on the obstacle attributes of the grid k and the obstacle attributes of the grid k'.

[0056] In some embodiments, obtaining a first radar observation grid layer according to a point cloud acquired by a first radar in an Nth acquisition cycle includes: S211: rasterizing the point cloud acquired by the first radar in the Nth acquisition cycle to obtain a first grid.

[0057] Among them, rasterization refers to converting discrete point cloud data into regular raster data. Specifically, rasterization includes the following process: projecting the point cloud collected by the first radar in the Nth acquisition cycle to the xy plane to form a two-dimensional point cloud containing height values. Then divide the xy plane containing the two-dimensional point cloud into a grid of size r*r to obtain the grid layer corresponding to the point cloud collected by the first radar in the Nth acquisition cycle. Among them, r is a set value, which can be set according to actual conditions. For example, r is 0.1m. Among them, when converting the point cloud collected by the first radar in the Nth acquisition cycle of the three-dimensional state into a two-dimensional point cloud, the height value of each laser point in the point cloud collected by the first radar in the Nth acquisition cycle is retained as the height value of the two-dimensional point cloud, and the z-axis direction is perpendicular to the xy plane.

[0058] In some embodiments, the first grid is any grid in the grid layer corresponding to the point cloud acquired by the first radar. The first grid is marked with a grid height and a height difference, and the two-dimensional point cloud in the first grid includes a plurality of laser points. In some embodiments, the median of the height of the point cloud points in the grid can be used as the grid height of the grid. The grid height reflects the height of the grid in the laser radar system coordinate system. In some embodiments, the absolute value of the difference between the maximum and minimum values ​​of the height of the point cloud points in the grid can be used as the height difference of the grid. The height difference reflects the distance between the highest and lowest points of the object corresponding to the grid, that is, reflects the height of the object.

[0059] S212: Mark the obstacle attributes of the first grid according to the obstacle information of the point cloud acquired by the first radar in the Nth acquisition cycle.

[0060] The processor of the laser radar system uses a deep learning-based perception algorithm to detect obstacles for each frame of point cloud collected by the first radar, and obtains real-time obstacle information in front to avoid obstacles or plan routes. In some embodiments, the perception algorithm can be a PointPillars algorithm or other trained neural network algorithms.

[0061] After the point cloud collected by the first radar in the Nth collection cycle is calculated and processed by the perception algorithm, the point cloud collected by the first radar in the Nth collection cycle is marked with obstacle information. Exemplarily, the perception algorithm will identify the point cloud features to generate a detection frame, and the detection frame is marked with the obstacle category. Among them, the detection frame is characterized by position parameters (x, y, z, w,l, h, θ), where (x, y, z) is the center coordinate of the detection frame, w is the width of the detection frame, l is the length of the detection frame, h is the height of the detection frame, and θ is the posture of the detection frame. The point cloud framed by the detection frame is the obstacle of the corresponding category. For example, if the obstacle category of detection frame 1# is vehicle, then the point cloud framed by detection frame 1# is identified as a vehicle; if the obstacle category of detection frame 2# is pedestrian, then the point cloud framed by detection frame 2# is identified as a pedestrian; if the obstacle category of detection frame 3# is cyclist, then the point cloud framed by detection frame 3# is identified as a cyclist.

[0062] In the embodiment of the present application, the point clouds framed by each detection frame are uniformly marked as obstacles (for example, 1 is used to represent obstacles), and the other point clouds outside the detection frame are uniformly marked as non-obstacles (for example, 0 is used to represent non-obstacles), so that the point cloud collected by the first radar in the Nth acquisition cycle has obstacle information. Therefore, according to the obstacle information of the point cloud collected by the first radar in the Nth acquisition cycle, the obstacle attribute of the first grid is marked. Among them, the obstacle attribute includes an obstacle grid or a non-obstacle grid. It can be understood that the obstacle grid refers to the grid occupied by the obstacle, and the non-obstacle grid refers to the grid not occupied by the obstacle.

[0063] Specifically, in some embodiments, when the point cloud in the first grid is marked as an obstacle or more than a first percentage of the point cloud in the first grid is marked as an obstacle, the first grid is marked as an obstacle grid. The first percentage can be set by a person skilled in the art based on actual conditions, for example, the first percentage is 60% or 70%. In some embodiments, when the point cloud in the first grid is not marked as an obstacle or more than a second percentage of the point cloud in the first grid is not marked as an obstacle, the first grid is marked as a non-obstacle grid. The second percentage can be set by a person skilled in the art based on actual conditions, for example, the second percentage is 60% or 50%.

[0064] S213: Determine and mark the area to which the first grid belongs according to the grid height, height difference and obstacle attributes of the first grid to obtain a first radar observation grid layer.

[0065] Among them, the area includes a perception area of ​​interest or a perception area of ​​non-interest. It can be understood that the perception area of ​​interest refers to the area in the first radar observation grid layer where attention needs to be paid to detecting obstacles. Exemplarily, the perception area of ​​interest generally includes obstacles, passable ground or other objects in the road that affect traffic. The perception area of ​​non-interest refers to the area in the first radar observation layer where attention does not need to be paid to detecting obstacles. Exemplarily, the perception area of ​​non-interest generally includes roadside fixed objects such as green belts or steps.

[0066] The grid height and height difference can reflect the height size of the object corresponding to the first grid, and the obstacle attribute can reflect whether the object corresponding to the first grid is an obstacle. Combining the height size and obstacle attribute can determine whether the object corresponding to the first grid is a roadside fixed object such as a green belt or steps, that is, determine the area to which the first grid belongs. It can be understood that for any first grid in the grid layer corresponding to the point cloud collected by the first radar in the Nth acquisition cycle, the area to which it belongs is determined in the above manner. After the area to which each first grid in the grid layer corresponding to the point cloud collected by the first radar in the Nth acquisition cycle belongs is determined, the first radar observation grid layer is obtained.

[0067] In autonomous driving scenarios, perception algorithms are usually used to detect dynamic obstacles such as vehicles, pedestrians or cyclists. Fixed roadside objects such as green belts or steps are usually regarded as objects that do not need to be detected. When it is necessary to detect green belts or steps, the perception algorithm needs to be further trained and optimized. The algorithm is complex and consumes a lot of computing power. In the above method, the grid layer corresponding to the point cloud collected by the first radar in the Nth acquisition cycle is divided into a perception area of ​​interest and a perception area of ​​non-interest, thereby reducing the interference of fixed roadside objects on the perception of obstacles in the field of view of the second radar, which is conducive to the subsequent accurate perception of obstacles in the field of view of the second radar. In addition, compared with the use of perception algorithms to identify objects such as green belts or steps, the above method is simple in calculation and can effectively save computing power.

[0068] In some embodiments, based on the grid height, height difference and obstacle attributes of the first grid, the area to which the first grid belongs is determined and marked to obtain a first radar observation grid layer, including: determining and marking the area to which the first grid that meets the first condition belongs as a non-focus perception area, and determining and marking the area to which the first grid that does not meet the first condition belongs as a focus perception area.

[0069] In some embodiments, the value 0 can be used to represent the aforementioned non-focus perception area, and the value 1 can be used to represent the aforementioned focus perception area. That is, when the area to which the first grid belongs is 1, it means that it belongs to the focus perception area; when the area to which the first grid belongs is 0, it means that it belongs to the non-focus perception area. It can be understood that each grid in the main grid map of the Nth acquisition cycle is marked with obstacle attributes. The obstacle grid represents the corresponding object as a target obstacle, and the non-obstacle grid represents the corresponding object as a non-target obstacle. Exemplarily, in the main grid map of the Nth acquisition cycle, the obstacle grid is filled with black and the non-obstacle grid is filled with white. Thus, the obstacle information in the field of view of the second radar can be determined based on the distribution of the black obstacle grid.

[0070] In some embodiments, the first condition is that the grid height is greater than or equal to a first threshold, the height difference is greater than or equal to a second threshold, and the obstacle attribute is a non-obstacle grid.

[0071] When the grid height is greater than or equal to the first threshold, the possibility that the object corresponding to the first grid is in the ground area in the laser radar system coordinate system is excluded; when the height difference is greater than or equal to the second threshold, it means that the object corresponding to the first grid may be a roadside fixed object or obstacle such as a green belt, a tree, a telephone pole, a traffic light, etc. When the obstacle attribute is an obstacle grid, it means that there are obstacles such as cars, pedestrians or cyclists that can be identified by the perception algorithm in the area corresponding to the first grid; when the obstacle attribute is a non-obstacle grid, the object corresponding to the first grid may be a roadside fixed object such as a green belt or there is no obstacle in the area corresponding to the first grid.

[0072] As can be seen from the above, the first threshold is used to exclude the ground, and the second threshold is used to distinguish dynamic obstacles such as pedestrians and vehicles from static obstacles such as green belts, trees, and steps. Each time the lidar system is started, the processor calls the first threshold and the second threshold pre-stored in the memory, and the first threshold and the second threshold are fixed thresholds set in advance. In order to make the first threshold and the second threshold conform to the actual road conditions, in some embodiments, the first threshold and the second threshold are determined based on the point cloud collected by the first radar in the first N-1 acquisition cycles. Exemplarily, the first threshold is 30cm, and the second threshold ranges from 30cm to 50cm.

[0073] For example, in Figure 1In the autonomous driving scenario shown, the first radar scans the area in front of the vehicle 200, and usually uses a perception algorithm to identify obstacles in the point cloud collected by the first radar in the Nth collection cycle to perform obstacle avoidance or path planning. As the vehicle 200 moves forward, the area it passes through is a historically passable area of ​​the road. Based on the historical frame point cloud corresponding to the historically passable area (i.e., the point cloud collected by the first radar in the first N-1 collection cycles), the road condition, such as the ground height or the ground slope height, can be determined. Therefore, the first threshold and the second threshold can be determined based on the road condition, which is conducive to more accurately segmenting the area to which the first grid belongs, so that the first radar observes the information of the grid layer more accurately, thereby improving the pertinence and accuracy of obstacle detection.

[0074] In some embodiments, obtaining a second radar observation grid layer according to a point cloud acquired by the second radar in an Nth acquisition cycle includes: S214: Rasterizing the point cloud acquired by the second radar in the Nth acquisition cycle to obtain a second grid, wherein the second grid is marked with a grid height and a height difference.

[0075] Among them, the rasterization can refer to the description of rasterization in step S211. Similarly, the point cloud collected by the second radar in the Nth acquisition cycle is rasterized, including the following process: the point cloud collected by the second radar in the Nth acquisition cycle is projected to the xy plane to form a two-dimensional point cloud containing height values. Then the xy plane containing the two-dimensional point cloud is divided into grids of size r*r to obtain the grid layer corresponding to the point cloud collected by the second radar in the Nth acquisition cycle. Among them, r is the same as the r mentioned in step S211. Among them, when the point cloud collected by the second radar in the three-dimensional state in the Nth acquisition cycle is converted into a two-dimensional point cloud, the height value of each laser point in the point cloud collected by the second radar in the Nth acquisition cycle in the z-axis direction is retained as the height value of the two-dimensional point cloud. In one embodiment, the xy plane is a horizontal plane parallel to the driving direction of the vehicle.

[0076] The second grid is any grid in the grid layer corresponding to the point cloud collected by the second radar in the Nth collection cycle. The second grid is marked with a grid height and a height difference. It can be understood that the two-dimensional point cloud in the second grid includes multiple laser points. In some embodiments, the median of the height of the point cloud points in the second grid can be used as the grid height of the second grid, and the absolute value of the difference between the maximum and minimum values ​​of the height of the point cloud points in the second grid can be used as the height difference of the second grid.

[0077] S215: Determine and mark the ground attributes of the second grid according to the grid height and the height difference of the second grid to obtain a second radar observation grid layer.

[0078] The ground attributes include ground grids or non-ground grids. The ground grid refers to a grid representing the ground, and the non-ground grid refers to a grid representing the non-ground.

[0079] The grid height and height difference can reflect the height size of the object corresponding to the second grid. Based on the obvious difference in height size between the ground and other non-ground objects, here, combined with the height size of the object reflected by the second grid, determine whether the second grid is a ground grid or a non-ground grid. For any second grid in the grid layer corresponding to the point cloud collected by the second radar in the Nth acquisition cycle, its ground attributes are determined in the above manner. After the ground attributes of each second grid in the grid layer corresponding to the point cloud collected by the second radar in the Nth acquisition cycle are determined, the second radar observation grid layer is obtained. Here, the second radar observation grid layer refers to the grid map of the second radar used for current observation perception and the latest in time sequence.

[0080] By using the above method, each second grid in the grid layer corresponding to the point cloud collected by the second radar in the Nth collection cycle is marked with ground attributes, so that the obtained second radar observation grid layer can distinguish between ground and non-ground, which is conducive to the subsequent accurate perception of obstacles in the second radar field of view.

[0081] In some embodiments, the ground attributes of the second grid are determined and marked based on the grid height and height difference of the second grid to obtain a second radar observation grid layer, including: determining and marking the second grid that meets the second condition as a non-ground grid, and determining and marking the second grid that does not meet the second condition as a ground grid.

[0082] In some embodiments, a non-ground grid may be represented by a value of 1, and a ground grid may be represented by a value of 0. That is, when the ground attribute of the second grid is 1, it represents a non-ground grid; when the ground attribute of the second grid is 0, it represents a ground grid.

[0083] In some embodiments, the second condition is that the grid height is greater than or equal to a third threshold, or the height difference is greater than or equal to a fourth threshold.

[0084] When the grid height is greater than or equal to the third threshold, it means that the object corresponding to the second grid has a higher height in the laser radar system coordinate system, which may be an obstacle or other non-ground object. Therefore, the second grid that satisfies the grid height greater than or equal to the third threshold is marked as a non-ground grid. In some embodiments, the third threshold may be a fixed threshold set in advance, for example, the third threshold is 30cm or 40cm, etc. When the height difference is greater than or equal to the fourth threshold, it means that there is a certain height difference between the object corresponding to the second grid and it cannot be the ground. Therefore, the second grid that satisfies the height difference greater than or equal to the fourth threshold is marked as a non-ground grid. In some embodiments, the fourth threshold may be a fixed threshold set in advance, for example, the fourth threshold is 20cm or 30cm, etc.

[0085] In some embodiments, the third threshold and the fourth threshold are determined based on the point cloud collected by the second radar in the first N-1 collection cycles. For example, the third threshold is 40 cm and the fourth threshold is 30 cm; in other embodiments, the third threshold is 30 cm and the fourth threshold is 20 cm.

[0086] In some embodiments, obtaining the main grid map observation layer according to the first radar observation grid layer and the second radar observation grid layer includes: S221: Superimpose the first radar observation grid layer and the second radar observation grid layer to obtain a fused grid map layer.

[0087] Among them, superposition refers to splicing, aligning or overlapping the grids in the first radar observation grid layer and the grids in the second radar observation grid layer. Specifically, for the field of view area where the first radar and the second radar do not overlap, the first grid of the first radar observation grid layer and the second grid of the second radar observation grid layer are spliced; for the field of view area where the first radar and the second radar overlap, the regional attributes of the first grid in the first radar observation grid layer and the ground attributes of the second grid in the second radar observation grid layer can be integrated.

[0088] S222: Determine and mark obstacle attributes of the fused grid map layer according to the region to which the first radar observation grid layer belongs and the ground attributes of the second radar observation grid layer to obtain a main grid map observation layer.

[0089] Specifically, for the first grid in the fused grid map layer, its obstacle attribute is determined based on the area to which the first grid belongs. For example, when the first grid belongs to the perception area of ​​interest, the first grid is determined to be an obstacle grid; when the first grid belongs to the perception area of ​​no interest, the first grid is determined to be a non-obstacle grid.

[0090] For the second grid in the fused grid map layer, its obstacle attribute is determined based on the ground attribute of the second grid. For example, when the second grid is a ground grid, it is determined to be a non-obstacle grid; when the second grid is a non-ground grid, it is determined to be an obstacle grid.

[0091] For the third grid in the fused grid map (formed by the overlap of the first grid and the second grid), the obstacle attribute of the third grid is determined based on the area to which the first grid belongs and the ground attribute of the second grid. In some embodiments, the detection result of the first radar can be trusted. When the first grid belongs to the perception area of ​​interest, the third grid is determined to be an obstacle grid regardless of whether the second grid is a ground grid or a non-ground grid; when the first grid belongs to the perception area of ​​no interest, the third grid is determined to be a non-obstacle grid regardless of whether the second grid is a ground grid or a non-ground grid. In some embodiments, when the first grid belongs to the perception area of ​​interest and the second grid is a non-ground grid, the third grid is determined to be an obstacle grid, and when the first grid belongs to the perception area of ​​no interest, the third grid is determined to be a non-obstacle grid.

[0092] The obstacle attributes of the fused grid map layer are determined by the regional attributes of the first radar observation grid layer and the ground attributes of the second radar observation grid layer, thereby improving the accuracy of obstacle detection.

[0093] See also Figure 5 , Figure 5 FIG. 1 is a flow chart of an obstacle detection method provided by another embodiment of the present application. Figure 5 As shown, the method A10 comprises the following steps: A11: Obtain point clouds collected by the first radar and the second radar in N collection cycles in chronological order, where N is an integer greater than 1.

[0094] A12: rasterize the point cloud acquired by the first radar in the Nth acquisition cycle to obtain the first grid; mark the obstacle attribute of the first grid according to the obstacle information of the point cloud acquired by the first radar in the Nth acquisition cycle; determine and mark the area to which the first grid that meets the first condition belongs as a non-attention perception area, and determine and mark the area to which the first grid that does not meet the first condition belongs as an attention perception area, and obtain the first radar observation grid layer. The first condition is that the grid height is greater than or equal to the first threshold, the height difference is greater than or equal to the second threshold, and the obstacle attribute is a non-obstacle grid.

[0095] The point cloud collected by the first radar in the Nth collection cycle is rasterized, and fixed roadside objects such as green belts or steps are identified. In combination with the obstacle attributes, the observation layer corresponding to the point cloud collected by the first radar in the Nth collection cycle is divided into a focus perception area and a non-focus perception area, so that the obtained first radar observation grid layer can exclude fixed roadside objects such as green belts or steps, so as to reduce the interference of these fixed roadside objects on the perception of obstacles in the field of view of the second radar.

[0096] A13: rasterize the point cloud collected by the second radar in the Nth collection cycle to obtain a second grid, wherein the second grid is marked with a grid height and a height difference; determine and mark the second grid that meets the second condition as a non-ground grid, and determine and mark the second grid that does not meet the second condition as a ground grid, to obtain the second radar observation layer. The second condition is that the grid height is greater than or equal to the third threshold, or the grid height difference is greater than or equal to the fourth threshold, and the third threshold and the fourth threshold are determined based on the point cloud collected by the second radar in the first N-1 collection cycles.

[0097] The ground and non-ground are identified in the above manner, and the ground attributes of each second grid in the observation layer corresponding to the point cloud collected by the second radar in the Nth collection cycle are marked, so that the obtained second radar observation grid layer can distinguish between the ground and non-ground, which is beneficial to improve the pertinence and accuracy of obstacle detection.

[0098] A14: The first radar observation grid layer and the second radar observation grid layer are superimposed to obtain a fused grid map layer; according to the area to which the first radar observation grid layer belongs and the ground attributes of the second radar observation grid layer, obstacle attributes of the fused grid map layer are determined and marked to obtain the main grid map observation layer of the Nth acquisition cycle.

[0099] The obstacle attributes of the fused grid map layer are determined dually by the regional attributes of the first radar observation grid layer and the ground attributes of the second radar observation grid layer, which is beneficial to improving the accuracy and reliability of the detection results.

[0100] A15: The main grid map of the Nth acquisition period is obtained according to the main grid map observation layer of the Nth acquisition period and the main grid map of the N-1th acquisition period.

[0101] A16: Perform cluster analysis on the main grid map of the Nth acquisition cycle to obtain obstacle information within the field of view of the second radar.

[0102] In summary, the present application is based on the overlap of the field of view of the first radar and the field of view of the second radar in time sequence, and adopts the main grid map of the N-1th acquisition period obtained according to the point cloud collected by the first radar in the first N-1 acquisition period and the point cloud collected by the second radar in the first N-1 acquisition period to represent the historical obstacle information sensed by the first radar. Thus, the historical obstacle information sensed by the first radar is used as prior knowledge, and the main grid map of the Nth acquisition period is determined by combining the point cloud collected by the first radar in the Nth acquisition period and the point cloud collected by the second radar in the Nth acquisition period, so that the main grid map of the Nth acquisition period can reflect the obstacle information in the field of view of the second radar, which is conducive to the subsequent accurate identification of the obstacle information in the field of view of the second radar according to the main grid map of the Nth acquisition period. In this way, on the one hand, using the historical obstacle information sensed by the first radar as prior knowledge, compared with clustering analysis or deep learning detection of the point cloud collected by the second radar in the Nth acquisition period, it can not only accurately perceive the obstacles in the field of view of the second radar, but also effectively save computing power. On the other hand, by generating the main grid map of the Nth acquisition cycle, reducing the three-dimensional point cloud into a two-dimensional grid map, performing cluster analysis in two-dimensional space, and identifying obstacles, the time and space complexity of the recognition algorithm can be effectively reduced, thereby saving computing power.

[0103] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the obstacle detection method in any of the above method embodiments is implemented.

[0104] In some embodiments, the storage medium may be a ferroelectric random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EE PROM), a flash memory, a magnetic surface memory, an optical disk, or a compact disc read-only memory (CD-ROM); it may also be various devices including one or any combination of the above memories.

[0105] In some embodiments, a computer program may be in the form of a program, software, software module, script or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment.

[0106] In the description of the present application, the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previously associated objects are in an "or" relationship.

[0107] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, those skilled in the art can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and are not limited here. In addition, in this article, the term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or system including a series of elements includes not only those elements, but also includes other elements that are not clearly listed, or also includes elements inherent to such process, method, article or system. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method, article or system including the element. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art belonging to the technical field of this application.

[0108] The above are only preferred embodiments of the present invention, and the scope of the patent of the present invention is not limited thereto. In the absence of conflict, the technical features in the above different embodiments can be combined, the steps can be implemented in any order, and there are many other changes in different aspects of the present application as described above. Although the present application is described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that without departing from the concept of the present application, the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features can be replaced by equivalents. Any equivalent structure or equivalent process transformation made using the contents of the specification and drawings of the present invention, or directly or indirectly used in other related technical fields, are similarly included in the scope of patent protection of the present invention.

Claims

1. An obstacle detection method, applied to a laser radar system, wherein the laser radar system comprises a first radar and a second radar, characterized in that: The method comprises: Obtaining point clouds collected by the first radar and the second radar in N collection cycles in chronological order, where N is an integer greater than 1; Obtaining a main grid map of the Nth acquisition cycle according to the point cloud acquired by the first radar in the Nth acquisition cycle, the point cloud acquired by the second radar in the Nth acquisition cycle, and the main grid map of the N-1th acquisition cycle, wherein the main grid map of the N-1th acquisition cycle is determined according to the point clouds acquired by the first radar and the second radar in the first N-1 acquisition cycles; Perform cluster analysis on the main grid map of the Nth acquisition period to obtain obstacle information within the field of view of the second radar.

2. The method according to claim 1, characterized in that The method of obtaining the main grid map of the Nth acquisition period according to the point cloud acquired by the first radar in the Nth acquisition period, the point cloud acquired by the second radar in the Nth acquisition period, and the main grid map of the N-1th acquisition period includes: According to the point cloud collected by the first radar in the Nth collection period, a first radar observation grid layer is obtained; according to the point cloud collected by the second radar in the Nth collection period, a second radar observation grid layer is obtained; According to the first radar observation grid layer and the second radar observation grid layer, a main grid map observation layer of the Nth acquisition period is obtained; The main grid map of the Nth acquisition period is obtained according to the main grid map observation layer of the Nth acquisition period and the main grid map of the N-1th acquisition period.

3. The method according to claim 2, characterized in that The step of obtaining a first radar observation grid layer according to the point cloud acquired by the first radar in the Nth acquisition cycle includes: Rasterizing the point cloud acquired by the first radar in the Nth acquisition cycle to obtain a first grid, wherein the first grid is marked with a grid height and a height difference, the grid height is the median of the heights of the point cloud points in the grid, and the height difference is the absolute value of the difference between the maximum value and the minimum value of the heights of the point cloud points in the grid; Marking obstacle attributes of the first grid according to obstacle information of the point cloud acquired by the first radar in the Nth acquisition cycle, wherein the obstacle attribute includes an obstacle grid or a non-obstacle grid; According to the grid height, height difference and obstacle attributes of the first grid, the area to which the first grid belongs is determined and marked to obtain the first radar observation grid layer, wherein the area includes a perception area of ​​interest or a perception area of ​​no interest.

4. The method according to claim 3, characterized in that: The step of obtaining a second radar observation grid layer according to the point cloud acquired by the second radar in the Nth acquisition cycle includes: rasterizing the point cloud acquired by the second radar in the Nth acquisition cycle to obtain a second grid, wherein the second grid is marked with the grid height and the height difference; According to the grid height and the height difference of the second grid, the ground attributes of the second grid are determined and marked to obtain the second radar observation grid layer, wherein the ground attributes include ground grids or non-ground grids.

5. The method according to claim 3, characterized in that: The determining and marking the area to which the first grid belongs according to the grid height, the height difference and the obstacle attribute of the first grid includes: Determine and mark the area to which the first grid that meets the first condition belongs as the non-attention perception area, and determine and mark the area to which the first grid that does not meet the first condition belongs as the attention perception area; The first condition is that the grid height is greater than or equal to a first threshold, the height difference is greater than or equal to a second threshold, and the obstacle attribute is the non-obstacle grid, and the first threshold and the second threshold are determined based on the point cloud collected by the first radar in the first N-1 collection cycles.

6. The method according to claim 4, characterized in that The step of determining and marking the ground attributes of the second grid according to the grid height and the height difference of the second grid includes: Determine and mark the second grid that meets the second condition as the non-ground grid, and determine and mark the second grid that does not meet the second condition as the ground grid; The second condition is that the grid height is greater than or equal to a third threshold, or the height difference is greater than or equal to a fourth threshold, and the third threshold and the fourth threshold are determined based on the point cloud collected by the second radar in the first N-1 collection cycles.

7. The method according to claim 4, characterized in that The step of obtaining the main grid map observation layer according to the first radar observation grid layer and the second radar observation grid layer comprises: Superimposing the first radar observation grid layer and the second radar observation grid layer to obtain a fused grid map layer; According to the region to which the first radar observation grid layer belongs and the ground attributes of the second radar observation grid layer, the obstacle attributes of the fused grid map layer are determined and marked to obtain the main grid map observation layer.

8. A laser radar system, characterized in that: include: A memory for storing executable program codes; A processor, configured to call and run the executable program code from the memory so that the laser radar system executes the obstacle detection method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the obstacle detection method according to any one of claims 1 to 7 is implemented.

Images