Obstacle avoidance method and device of sweeper, terminal equipment and storage medium

Abstract

The application discloses an obstacle avoidance method and device of a cleaning vehicle, a terminal device and a storage medium, and relates to the automatic driving field.The method comprises the following steps: obtaining global point cloud information of a current frame of a laser radar, and performing conditional filtering to obtain coarse obstacle information; obtaining a virtual detection frame of an obstacle in the global point cloud information of the current frame; performing voxelization processing on the coarse obstacle information to obtain a plurality of first grid frames; calculating the similarity between the first grid frames and the virtual detection frame, extracting obstacle points in the first grid frames that meet a preset condition and that coincide with the virtual detection frame to obtain fine obstacle information, and performing obstacle avoidance planning based on the fine obstacle information.The application can not only solve the problem of long time consumption caused by multi-sensor matching, but also can effectively improve the obstacle avoidance efficiency and accuracy of the cleaning vehicle.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, specifically to a method, device, terminal equipment, and storage medium for obstacle avoidance in a sweeper vehicle. Background Technology

[0002] LiDAR, as an emerging 3D sensor, can obtain object information from multiple directions, thus achieving better perception results. However, LiDAR is not only costly and complex to manufacture, but also performs extremely poorly in rainy, foggy, or foggy conditions. In point clouds, water mist and highly reflective objects have similar intensities, making them difficult to distinguish. Therefore, how to use LiDAR for obstacle avoidance in foggy conditions has become a hot topic.

[0003] Currently, many obstacle avoidance solutions have emerged on the market, such as:

[0004] Patent CN108181904A proposes a robot obstacle avoidance scheme that determines object position information by the echo duration of ultrasonic waves. Although using millimeter-wave or ultrasonic radar can accurately locate objects and effectively solve the obstacle avoidance problem under water mist, it requires the fusion of multiple sensors, resulting in high cost, long time consumption, and difficulty in deployment.

[0005] The patent with publication number CN110488805A proposes to use a binocular camera to obtain point cloud depth information. Although it has the advantages of high efficiency and high performance, 2D detection based on binocular cameras is greatly affected by light and still cannot solve the problem of high-reflection obstacles under water mist.

[0006] In summary, the existing obstacle avoidance solutions have the following drawbacks: multi-sensor fusion results in high cost, long time consumption and difficulty in deployment; 2D detection is greatly affected by light and cannot solve the problem of high-reflectivity obstacles under water mist. Summary of the Invention

[0007] To address the shortcomings of the existing technology, this invention provides a method, device, terminal equipment, and storage medium for obstacle avoidance in a sweeper. By employing a single sensor, it can accurately screen highly reflective obstacles in rain and fog, thereby effectively improving the obstacle avoidance efficiency and accuracy of the sweeper.

[0008] To achieve the above objectives, the present invention provides an obstacle avoidance method for a sweeper, which involves mounting a lidar on the sweeper and defining the area within the lidar's field of view that is less than a first threshold distance from the sweeper as a first-level area.

[0009] The obstacle avoidance method includes the following steps:

[0010] Step 1: Obtain the global point cloud information of the current frame of the lidar, and perform conditional filtering on the global point cloud information of the current frame based on the range data of the first-level region and the intensity data of the obstacle to obtain the coarse obstacle information of the obstacle.

[0011] Step 2: Obtain the virtual detection box of the obstacle in the global point cloud information of the current frame;

[0012] Step 3: Voxelize the coarse obstacle information to obtain several first grid frames, wherein each first grid frame contains at least one obstacle point;

[0013] Step 4: Calculate the similarity between each first grid frame and the virtual detection frame, and extract all obstacle points in the first grid frame that overlap with the virtual detection frame and whose similarity meets the preset conditions to obtain detailed obstacle information of the obstacle;

[0014] Step 5: Based on the detailed obstacle information, plan the obstacle avoidance trajectory of the sweeper.

[0015] In one embodiment, step 2, obtaining the virtual detection box of the obstacle in the global point cloud information of the current frame, includes:

[0016] Based on the detection information of obstacles in the global point cloud information of historical frames, the position of obstacles from the far end to the near end is estimated, and the virtual detection box of the obstacle in the global point cloud information of the current frame is obtained.

[0017] The detection information includes the obstacle's position, velocity, and inherent dimensions.

[0018] In one embodiment, step 2 includes:

[0019] The area within the field of view of the lidar that is outside the primary region is defined as the secondary region;

[0020] When an obstacle enters the secondary region, the obstacle's position information, velocity information, and inherent size information are identified and acquired.

[0021] When an obstacle enters the primary region from the secondary region, the position of the obstacle in the global point cloud information of the corresponding frame and the relative velocity information of the obstacle are obtained, and the virtual detection box of the obstacle in the global point cloud information of the corresponding frame is obtained by combining the inherent size information of the obstacle.

[0022] Based on the relative velocity information of the obstacle and the initial virtual detection box, the virtual detection box of the obstacle in the global point cloud information of the current frame can be obtained by real-time position estimation in the global point cloud information of the subsequent frame.

[0023] In one embodiment, in step 4, the similarity index between the first grid frame and the virtual detection frame includes overlap and / or the number of overlapping points.

[0024] In one embodiment, the determination criterion for a first grid frame whose similarity meets a preset condition is as follows:

[0025] For any of the first grid frames, calculate the overlap ratio (IOU) between it and the virtual detection frame, and calculate the number of obstacle points N in the first grid frame that overlap with the virtual detection frame. If the IOU is greater than the second threshold and N is greater than the third threshold, then the similarity of the first grid frame is determined to meet the preset conditions; otherwise, the preset conditions are not met.

[0026] In one embodiment, step 5 includes:

[0027] Step 5.1: The fine obstacle information is voxelized to obtain a plurality of second grid frames, wherein each second grid frame contains at least one obstacle point;

[0028] Step 5.2: Calculate the distance between each of the second grid frames and the sweeper vehicle, and determine whether there are any second grid frames with a distance less than the fourth threshold:

[0029] If so, proceed to step 5.3;

[0030] Otherwise, maintain the current driving status of the sweeper.

[0031] Step 5.3: Determine whether the second grid frame, whose distance is less than the fourth threshold, is located in the primary region.

[0032] If so, control the sweeper to move in the opposite direction and change the preset angle to pass over the obstacle;

[0033] Otherwise, keep the sweeper in its current driving state.

[0034] In one embodiment, step 1, the conditional filtering process is as follows:

[0035] ;

[0036] ;

[0037] ;

[0038] ;

[0039] in, , These are the minimum and maximum coordinates of the first-level region on the x-axis. , These are the minimum and maximum coordinates of the first-level region on the y-axis. , These are the minimum and maximum coordinates of the first-level region on the z-axis. , These are the lower and upper limits of the obstacle reflection intensity information. The spatial coordinates of the obstacle points in the coarse obstacle information are... This refers to the reflection intensity information of obstacle points in the coarse obstacle information.

[0040] To achieve the above objectives, the present invention also provides an obstacle avoidance device for a sweeper, employing the above-described method for obstacle avoidance. The obstacle avoidance device includes:

[0041] The coarse obstacle extraction unit is used to conditionally filter the global point cloud information of the current frame based on the range data of the primary region and the intensity data of the obstacle to obtain coarse obstacle information of the obstacle.

[0042] The obstacle estimation unit is used to obtain the virtual detection box of the obstacle in the global point cloud information of the current frame;

[0043] A voxel segmentation unit is used to voxelize the coarse obstacle information to obtain a plurality of first grid frames, wherein each first grid frame contains at least one obstacle point.

[0044] The fine obstacle extraction unit is used to calculate the similarity between each first grid frame and the virtual detection frame, and to extract all obstacle points in the first grid frame that overlap with the virtual detection frame and whose similarity meets the preset conditions, so as to obtain the fine obstacle information of the obstacle.

[0045] The obstacle avoidance planning unit is used to plan the sweeper's driving trajectory for obstacle avoidance based on the detailed obstacle information.

[0046] To achieve the above objectives, the present invention also provides a terminal device, which is a computer, an unmanned vehicle, a drone, an unmanned driving device, or a mobile robot, and the terminal device is equipped with:

[0047] Memory, used to store programs;

[0048] A processor for executing the program stored in the memory, wherein when the program is executed, the processor is used to perform some or all of the steps of the method described above.

[0049] To achieve the above objectives, the present invention also provides a computer-readable storage medium storing computer-executable instructions; when executed by a processor, the computer-executable instructions are used to implement some or all of the steps of the method described above.

[0050] Compared with the prior art, the present invention has the following beneficial technical effects:

[0051] 1. This invention enables sweepers to accurately avoid obstacles by using LiDAR for single-sensor imaging. Compared with traditional multi-sensor fusion methods, it eliminates the problems of long time consumption and difficult deployment caused by multi-sensor matching.

[0052] 2. This invention obtains detailed obstacle information by matching the coarse obstacle information obtained from conditional filtering with the virtual detection box of the obstacle. This not only prevents the loss of object imaging information at the intersection of noise and obstacles, but also accurately filters out highly reflective objects with poor imaging effects, thereby effectively improving the obstacle avoidance efficiency and accuracy of the sweeper. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0054] Figure 1 This is a flowchart of the obstacle avoidance method in Embodiment 1 of the present invention;

[0055] Figure 2 This is a schematic diagram of the primary region and the secondary region in Embodiment 1 of the present invention;

[0056] Figure 3 This is a schematic diagram of the detection results obtained through coarse obstacle information in Embodiment 1 of the present invention;

[0057] Figure 4 This is a schematic diagram of the detection results obtained through fine obstacle information in Embodiment 1 of the present invention;

[0058] Figure 5 This is a structural block diagram of the obstacle avoidance device in Embodiment 2 of the present invention;

[0059] Figure 6 This is a structural block diagram of the terminal device in Embodiment 3 of the present invention.

[0060] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0062] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0063] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0064] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection, an electrical connection, a physical connection, or a wireless communication connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two elements or the interaction between two elements, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0065] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0066] Example 1

[0067] This embodiment discloses an obstacle avoidance method for a sweeper vehicle, mainly applicable to obstacle avoidance in rainy or foggy weather or under water spray conditions. The sweeper vehicle can be a robotic vacuum cleaner, a water truck, or a fire truck, or any vehicle that comes into close contact with water. By equipping the sweeper vehicle with a single 16-line LiDAR and acquiring global point cloud information from the LiDAR for obstacle identification and avoidance, the method utilizes the global point cloud information, which includes the information of every point in the entire frame of the point cloud image. x Axis value,y Axis value, z Axis value and reflection intensity information i .

[0068] Specifically, in the obstacle recognition and avoidance process, conditional filtering is used at the near-vehicle end to filter obstacles and then voxelization is performed to obtain coarse obstacle information. Simultaneously, 3D perception is used to acquire 3D obstacle detection information at the far-vehicle end, including the obstacle's speed, position, height, width, and other relevant physical parameters. Then, based on given relative motion speed relationships and object loss logic, the obstacle's position is calculated from the far-vehicle end to the near-vehicle end, resulting in a virtual detection box for the obstacle at the near-vehicle end. The coarse obstacle information is then matched with the virtual detection box to obtain fine obstacle information. This fine obstacle information is used to plan vehicle operations and prevent collisions. This process not only prevents the loss of object imaging information at the intersection of noise and obstacles but also accurately filters out highly reflective objects with poor imaging quality, effectively improving the obstacle avoidance efficiency and accuracy of the sweeper.

[0069] In this embodiment, the area within the LiDAR's field of view is divided into primary and secondary regions based on distance. Specifically, the area within the LiDAR's field of view where the distance to the sweeper's marker point is less than a first threshold is defined as the primary region, and the remaining area is defined as the secondary region. The primary region is the near-vehicle area, where obstacle clarity is low due to water mist and other factors, resulting in numerous incomplete areas in the point cloud image and making detection difficult. Therefore, the near-vehicle primary region is defined as a water mist area. The secondary region is the far-vehicle area, where obstacle outlines are clearer and perception results are more accurate due to less or no water mist. Therefore, the far-vehicle secondary region is defined as a water mist-free area. The sweeper's marker point can be a feature point such as the sweeper's center point, lower left point, or upper right point.

[0070] refer to Figure 1 The obstacle avoidance method in this embodiment specifically includes the following steps:

[0071] Step 1: Obtain the global point cloud information of the current frame from the LiDAR, and perform conditional filtering on the global point cloud information of the current frame based on the range data of the primary region and the intensity data of the obstacles to obtain coarse obstacle information. The global point cloud information includes the information of every point in the entire frame point cloud map. x Axis value, y Axis value, z Axis value and reflection intensity information i ;

[0072] Step 2: Based on the detection information of the obstacle in the global point cloud information of the historical frame, the position is estimated to obtain the virtual detection box of the obstacle in the global point cloud information of the current frame;

[0073] Step 3: Voxelize the coarse obstacle information to obtain several first grid frames. Each first grid frame contains at least one obstacle point. That is, the obstacle point set corresponding to the coarse obstacle information is divided into several subsets, and each subset corresponds to a first grid frame.

[0074] Step 4: Calculate the similarity between each first grid frame and the virtual detection frame, such as the degree of overlap and the number of overlapping points. Extract all obstacle points in the first grid frame that overlap with the virtual detection frame and meet the preset similarity conditions to obtain the detailed obstacle information of the obstacle.

[0075] Step 5: Perform voxelization processing based on fine obstacle information, and determine the obstacle avoidance planning of the sweeper's driving trajectory based on the distance between the second grid frame after voxelization of fine obstacle information and the sweeper's identification point.

[0076] In the specific implementation process, the process of conditionally filtering the global point cloud information of the current frame based on the range data of the primary region and the intensity data of obstacles in step 1 is as follows:

[0077] For any point in the global point cloud information ,in, These are the spatial coordinates of the point. For the reflection intensity information at this point, determine whether it simultaneously satisfies the following four-dimensional constraints:

[0078] ;

[0079] ;

[0080] ;

[0081] ;

[0082] in, , These are the minimum and maximum coordinates of the first-level region on the x-axis. , These are the minimum and maximum coordinates of the first-level region on the y-axis. , These are the minimum and maximum coordinates of the first-level region on the z-axis. , These are the lower and upper limits for obstacle reflection intensity information;

[0083] If satisfied, then the decision point is... This is an obstacle point in the coarse obstacle information.

[0084] Because the coarse obstacle information obtained from the primary region is incomplete, and highly reflective objects such as bright traffic cones and black cars are missed, this embodiment supplements the primary region by adding a secondary region. The secondary region is a water-free area for obtaining obstacle detection information. Obstacles include, but are not limited to, ordinary objects such as bollards, roadblocks, vehicles, and utility poles, as well as highly reflective objects such as bright traffic cones and black cars. The detection information includes the obstacle's coordinates, its inherent dimensions (length, width, height, side slope, etc.), and its speed. The coordinates of the obstacle can be the spatial coordinates of its center point, lower left point, upper right point, or other marker points.

[0085] Therefore, in step 2, the position of the obstacle from the far end to the near end can be estimated based on the detection information of the obstacle in the global point cloud information of the historical frame, and then the virtual detection box of the obstacle in the global point cloud information of the current frame can be estimated. The specific implementation process is as follows:

[0086] First, when an obstacle enters the secondary area, the system identifies and acquires the obstacle's position, velocity, and inherent size information.

[0087] Secondly, when an obstacle enters the primary region from the secondary region, the position of the obstacle in the global point cloud information of the corresponding frame and the relative velocity information of the obstacle are obtained, and the virtual detection box of the obstacle in the global point cloud information of the corresponding frame is obtained by combining the inherent size information of the obstacle.

[0088] Finally, based on the relative velocity information of the obstacle and the initial virtual detection box, the real-time position of the obstacle in the global point cloud information of the subsequent frame can be calculated to obtain the virtual detection box of the obstacle in the global point cloud information of the current frame.

[0089] In this embodiment, the relative velocity information of the obstacle refers to the magnitude and direction of the obstacle's velocity relative to the vehicle (i.e., the sweeper) when the vehicle is stationary as a reference point. Within this secondary region, in a Cartesian coordinate system based on the vehicle, if the velocity of the obstacle within the secondary region... Assuming the vehicle always travels in a straight line, Since the vehicle speed is the given value, then:

[0090] ;

[0091] ;

[0092] in, This represents the speed vector of the sweeper during normal operation. , They are respectively The orthogonal decomposition values ​​along the x-axis and y-axis satisfy the parallelogram rule, i.e. Figure 2 As shown;

[0093] In a Cartesian coordinate system based on obstacles, if the velocity of the obstacle in the secondary region... Substituting the calculated vehicle speed into the equation, and assuming the vehicle always travels in a straight line, Given the relative velocity of the obstacle, we have:

[0094] ;

[0095] ;

[0096] ;

[0097] in, This represents the velocity vector of the obstacle during its normal movement within the secondary zone. , They are respectively The orthogonal decomposition values ​​along the x-axis and y-axis satisfy the parallelogram rule, i.e. Figure 2 As shown;

[0098] If the speed of the obstacles in the secondary area Furthermore, the fixed obstacle moves from front to back of the vehicle, assuming the vehicle always travels in a straight line. Given the relative velocity of the obstacle, we have:

[0099] ;

[0100] ;

[0101] ;

[0102] Based on the vehicle's Cartesian coordinate system, kinematic derivations are performed within the secondary region to obtain relevant physical information. It is assumed that the time it takes for the obstacle marker to enter the secondary region is... The time spent leaving the secondary zone is The initial speed of the obstacle is Then we have:

[0103] ;

[0104] ;

[0105] in, Let be the upper boundary value of the secondary region on the y-axis, and this value is defined as the predicted value. Let the predicted objects existing within the secondary region be the observed values. Kalman filtering is then applied to the predicted and observed values ​​to obtain a new... Obtain the last frame of obstacles within the secondary area. Location information at any time As initial information, among which The relevant physical properties include the obstacle's length, width, height, and lateral slope. The obstacle moves at a constant velocity with negligible acceleration within the secondary region. The velocity information obtained through the above derivation... Calculate the position of the obstacle within each frame of the primary region, considering the time interval of global point cloud information between two adjacent frames. The time is 0.1s, which is used to record the entry of an obstacle into the first-level area. Location information at any time Then we have:

[0106] ;

[0107] ;

[0108] ;

[0109] Finally, the inherent dimensional information of the obstacle, such as its length, width, height, and side slope, is obtained from the obstacle detection information in the secondary region, and combined with the marker points on the obstacle in the current frame. By using the spatial coordinate information, the virtual detection box of the obstacle in the global point cloud information of the current frame can be obtained.

[0110] In this embodiment, the similarity between the first grid frame and the virtual detection frame in step 4 includes two indicators: overlap and number of overlapping points. The determination criterion for the first grid frame whose similarity meets the preset conditions is as follows:

[0111] For any first grid frame, calculate its overlap ratio (IOU) with the virtual detection frame, and calculate the number of obstacle points N in the first grid frame that overlap with the virtual detection frame. If the IOU is greater than the second threshold and N is greater than the third threshold, then the similarity of the first grid frame is determined to meet the preset conditions; otherwise, the preset conditions are not met.

[0112] It is worth noting that in practical applications, the similarity between the first grid frame and the virtual detection frame is not limited to using both the overlap degree and the number of overlap points at the same time; only one of them can be used.

[0113] In practice, the first grid frame obtained by voxelizing the coarse obstacle information has a size of 10cm × 10cm × 10cm. Of course, other sizes can also be chosen, depending on the hardware performance.

[0114] In this embodiment, the specific implementation process of step 5, which involves obstacle avoidance planning based on detailed obstacle information for the sweeper's trajectory, includes:

[0115] Step 5.1: Voxelize the fine obstacle information to obtain several second grid frames. Each second grid frame contains at least one obstacle point. That is, the obstacle point set corresponding to the fine obstacle information is divided into several subsets, and each subset corresponds to a second grid frame.

[0116] Step 5.2: Calculate the distance between each second grid frame and the sweeper, and determine whether there are any second grid frames with a distance less than the fourth threshold. The fourth threshold is preferably set to 50cm. The distance between the second grid frame and the sweeper specifically refers to the distance between the center point of the second grid frame and the marking point of the sweeper.

[0117] If so, proceed to step 5.3;

[0118] Otherwise, maintain the current driving status of the sweeper.

[0119] Step 5.3: Determine whether the second grid frame with a distance less than the fourth threshold is located in the primary area, that is, determine whether there is a second grid frame in the primary area that satisfies the condition of "the distance between it and the sweeper is less than the fourth threshold":

[0120] If so, control the sweeper to drive in the opposite direction at a preset angle after passing the obstacle. The preset angle can be set according to actual needs. When the sweeper is a small sweeper in the park, the preset angle can be set to 15°, 20°, etc. When the sweeper is a large sweeper on the road, the preset angle can be set to 45°, 60°, etc. For example, control the sweeper to drive in the opposite direction at 45° after passing the obstacle.

[0121] Otherwise, keep the sweeper in its current driving state.

[0122] In practice, the second grid frame, obtained by voxelizing the fine obstacle information, has dimensions of 30cm × 30cm × 30cm. Of course, other dimensions can also be chosen, depending on the hardware performance.

[0123] refer to Figure 3 The detection results are obtained using coarse obstacle information. Figure 4 The detection results are obtained through detailed obstacle information. Figure 3 , Figure 4 The white dots in the image are noise detection points. (Through...) Figure 3 and Figure 4The comparison shows that the effect of adding virtual object detection is significantly better than that of not adding virtual object detection. Therefore, this embodiment obtains fine obstacle information by feature matching between the coarse obstacle information obtained by conditional filtering and the virtual detection box of the obstacle. This not only prevents the loss of object imaging information at the intersection of noise and obstacles, but also accurately filters out highly reflective objects with poor imaging effects, thereby effectively improving the obstacle avoidance efficiency and accuracy of the sweeper.

[0124] Example 2

[0125] Based on the obstacle avoidance method in Embodiment 1, this embodiment discloses an obstacle avoidance device for a sweeper. (Reference) Figure 5 The obstacle avoidance device includes a coarse obstacle extraction unit, an obstacle calculation unit, a voxel segmentation unit, a fine obstacle extraction unit, and an obstacle avoidance planning unit. This obstacle avoidance device is used to execute some or all of the steps of the obstacle avoidance method in Embodiment 1, thereby achieving precise obstacle avoidance for the sweeper. Specifically:

[0126] The coarse obstacle extraction unit is used to perform conditional filtering on the global point cloud information of the current frame based on the range data of the primary region and the intensity data of the obstacle to obtain coarse obstacle information of the obstacle.

[0127] The obstacle estimation unit is used to obtain the virtual detection box of the obstacle in the global point cloud information of the current frame;

[0128] The voxel segmentation unit is used to voxelize the coarse obstacle information to obtain a plurality of first grid frames, wherein each first grid frame contains at least one obstacle point.

[0129] The fine obstacle extraction unit is used to calculate the similarity between each first grid frame and the virtual detection frame, and extract all obstacle points in the first grid frame that overlap with the virtual detection frame and whose similarity meets the preset conditions, to obtain the fine obstacle information of the obstacle;

[0130] The obstacle avoidance planning unit is used to plan the sweeper's trajectory for obstacle avoidance based on the detailed obstacle information.

[0131] In this embodiment, the specific working process and working principle of the coarse obstacle extraction unit, obstacle estimation unit, voxel segmentation unit, fine obstacle extraction unit and obstacle avoidance planning unit are the same as those in Embodiment 1, so they will not be described again in this embodiment.

[0132] Example 3

[0133] like Figure 6The diagram illustrates a terminal device disclosed in this embodiment, comprising a transmitter, a receiver, a memory, and a processor. The transmitter transmits instructions and data, the receiver receives instructions and data, the memory stores computer-executed instructions, and the processor executes the computer-executed instructions stored in the memory to implement some or all of the steps performed by the obstacle avoidance method in Embodiment 1. Its specific implementation process is the same as the obstacle avoidance method in Embodiment 1.

[0134] It is important to note that the aforementioned memory can be either standalone or integrated with the processor. When the memory is set up independently, the terminal device also includes a bus for connecting the memory and the processor.

[0135] In practical applications, terminal devices include computers, unmanned vehicles, drones, driverless equipment, or mobile robots.

[0136] Example 4

[0137] This embodiment discloses a computer-readable storage medium storing computer-executable instructions. When the processor executes the computer-executable instructions, it implements some or all of the steps of the obstacle avoidance method in Embodiment 1 above.

[0138] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A method for obstacle avoidance by a sweeper, characterized in that, The lidar is mounted on the sweeper, and the area within the lidar's field of view that is less than the distance to the sweeper is defined as the first-level area; The obstacle avoidance method includes the following steps: Step 1: Obtain the global point cloud information of the current frame of the lidar, and perform conditional filtering on the global point cloud information of the current frame based on the range data of the first-level region and the intensity data of the obstacle to obtain the coarse obstacle information of the obstacle. Step 2, obtaining the virtual detection box of the obstacle in the global point cloud information of the current frame, including: calculating the position of the obstacle from the far end to the near end of the vehicle based on the detection information of the obstacle in the global point cloud information of the historical frame, and obtaining the virtual detection box of the obstacle in the global point cloud information of the current frame, wherein the detection information includes the obstacle's position information, velocity information and inherent size information; Step 3: Voxelize the coarse obstacle information to obtain several first grid frames, wherein each first grid frame contains at least one obstacle point; Step 4: Calculate the similarity between each first grid frame and the virtual detection frame, and extract all obstacle points in the first grid frame that overlap with the virtual detection frame and whose similarity meets the preset conditions to obtain detailed obstacle information of the obstacle; Step 5: Based on the detailed obstacle information, plan the obstacle avoidance trajectory of the sweeper.

2. The obstacle avoidance method for a sweeper according to claim 1, characterized in that, Step 2 includes: The area within the field of view of the lidar that is outside the primary region is defined as the secondary region; When an obstacle enters the secondary region, the obstacle's position information, velocity information, and inherent size information are identified and acquired. When an obstacle enters the primary region from the secondary region, the position of the obstacle in the global point cloud information of the corresponding frame and the relative velocity information of the obstacle are obtained, and the virtual detection box of the obstacle in the global point cloud information of the corresponding frame is obtained by combining the inherent size information of the obstacle. Based on the relative velocity information of the obstacle and the initial virtual detection box, the virtual detection box of the obstacle in the global point cloud information of the current frame can be obtained by real-time position estimation in the global point cloud information of the subsequent frame.

3. The obstacle avoidance method for a sweeper according to claim 1 or 2, characterized in that, In step 4, the similarity index between the first grid frame and the virtual detection frame includes overlap and / or the number of overlapping points.

4. The obstacle avoidance method for a sweeper according to claim 3, characterized in that, For the first grid frame whose similarity meets the preset conditions, the determination criteria are as follows: For any of the first grid frames, calculate the overlap ratio (IOU) between it and the virtual detection frame, and calculate the number of obstacle points N in the first grid frame that overlap with the virtual detection frame. If the IOU is greater than the second threshold and N is greater than the third threshold, then the similarity of the first grid frame is determined to meet the preset conditions; otherwise, the preset conditions are not met.

5. The obstacle avoidance method for a sweeper according to claim 1 or 2, characterized in that, Step 5 includes: Step 5.1: The fine obstacle information is voxelized to obtain a plurality of second grid frames, wherein each second grid frame contains at least one obstacle point; Step 5.2: Calculate the distance between each of the second grid frames and the sweeper vehicle, and determine whether there are any second grid frames with a distance less than the fourth threshold: If so, proceed to step 5.3; Otherwise, maintain the current driving status of the sweeper. Step 5.3: Determine whether the second grid frame, whose distance is less than the fourth threshold, is located in the primary region. If so, control the sweeper to move in the opposite direction and change the preset angle to pass over the obstacle; Otherwise, keep the sweeper in its current driving state.

6. The obstacle avoidance method for a sweeper according to claim 1 or 2, characterized in that, In step 1, the conditional filtering process is as follows: in, , These are the minimum and maximum coordinates of the first-level region on the x-axis. , These are the minimum and maximum coordinates of the first-level region on the y-axis. , These are the minimum and maximum coordinates of the first-level region on the z-axis. , These are the lower and upper limits of the obstacle reflection intensity information. The spatial coordinates of the obstacle points in the coarse obstacle information are... This refers to the reflection intensity information of obstacle points in the coarse obstacle information.

7. An obstacle avoidance device for a sweeper vehicle, characterized in that, The obstacle avoidance device for a sweeper vehicle, comprising the method described in any one of claims 1 to 6, includes: The coarse obstacle extraction unit is used to conditionally filter the global point cloud information of the current frame based on the range data of the primary region and the intensity data of the obstacle to obtain coarse obstacle information of the obstacle. The obstacle estimation unit is used to obtain the virtual detection box of the obstacle in the global point cloud information of the current frame; A voxel segmentation unit is used to voxelize the coarse obstacle information to obtain a plurality of first grid frames, wherein each first grid frame contains at least one obstacle point. The fine obstacle extraction unit is used to calculate the similarity between each first grid frame and the virtual detection frame, and to extract all obstacle points in the first grid frame that overlap with the virtual detection frame and whose similarity meets the preset conditions, so as to obtain the fine obstacle information of the obstacle. The obstacle avoidance planning unit is used to plan the sweeper's driving trajectory for obstacle avoidance based on the detailed obstacle information.

8. A terminal device, said terminal device being a computer, unmanned vehicle, drone, autonomous driving equipment, or mobile robot, characterized in that, The terminal device is equipped with: Memory, used to store programs; A processor for executing the program stored in the memory, wherein when the program is executed, the processor is configured to perform some or all of the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions; when executed by a processor, the computer-executable instructions are used to implement some or all of the steps of the method as described in any one of claims 1 to 6.

Images