Depth point cloud and wheel type odometer-based anti-falling method for robot to go up and down elevator

Abstract

The invention provides a method for preventing a robot from falling in and out of an elevator based on a depth point cloud and a wheel type odometer, which comprises the following steps: firstly, acquiring depth image information, carrying out two times of straight-through filtering on point cloud data, screening out an anti-falling detection area, then carrying out preprocessing on the point cloud in the area, and carrying out radius-based outlier elimination, so as to obtain an anti-falling detection result; dividing the detection area into grids according to transverse and longitudinal resolutions, traversing all the grids to carry out anti-falling detection calculation, and outputting a calculation result; and finally, triggering secondary motion verification of the wheel type speedometer according to a calculation result, if the calculation result is true and the motion of the speedometer exceeds a threshold value at the moment, issuing a final anti-falling triggering result, otherwise, issuing an anti-falling non-triggering result. The detection capability of the robot on the potential falling environment when the robot goes up and down the elevator is effectively improved, the problem that the robot goes up and down the elevator fails due to frequent triggering of falling prevention in the movement process is solved, and the operation safety and stability of the robot are improved.

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to a method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer. Background Art

[0002] Typically, robots using single-point laser ranging sensors or depth cameras for fall prevention during elevator rides use sensors with high accuracy, but in practice, they often experience false alarms due to the complexity of the scene. The latter uses a 3D point cloud for height detection. When a large depression appears in front of the robot's elevator, it often triggers the fall prevention function prematurely, causing the robot to stop and fail to reach the elevator. Furthermore, fixed-mounted depth cameras have blind spots, preventing accurate observation of the elevator terminal and posing a risk of falls. Summary of the Invention

[0003] In order to solve the problems of the prior art, the present invention provides a method for preventing robots from getting on and off elevators based on depth point clouds and wheel odometers, which effectively improves the robot's ability to detect potential falling environments when getting on and off elevators, and reduces the problem of elevator failure caused by frequent triggering of anti-fall measures during movement, thereby improving the robot's operational safety and stability.

[0004] The present invention provides a method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer, comprising the following steps:

[0005] S1, obtains depth image information, converts the depth image into a depth point cloud and rotates the coordinate axis;

[0006] S2, perform two straight-through filtering on the point cloud data to screen out the anti-fall detection area, and then pre-process the point cloud in the area;

[0007] S3: Divide the detection area into grids according to horizontal and vertical resolutions, traverse all grids to perform anti-fall detection calculations, and output the calculation results. Specifically, the pre-processed point cloud data is further divided into small grids according to different horizontal and vertical resolutions. The point cloud data within the grid is logically judged. That is, if the depth information of the point cloud falling within the grid is lower than the ground threshold and the number of point clouds is higher than the threshold, the grid is judged as a potential fall risk and counted. All grids are traversed in a loop to count the number of grids with potential fall risks. When the number of grids exceeds a certain threshold, it is determined that anti-fall is triggered and the status is released.

[0008] S4, triggering the secondary motion check of the wheel odometer according to the calculation result. If the calculation result is true and the odometer movement exceeds the threshold at this time, the final anti-fall trigger result is released, otherwise the anti-fall non-trigger result is released.

[0009] As a further improvement, in step S1, the depth image information is obtained by a depth camera carried by a mobile robot.

[0010] For further improvement, the specific process of step S1 is to obtain the depth image information of the depth camera carried by the mobile robot, convert the depth image information into the universal PCL point cloud format, and rotate the data at the same time so that the point cloud data can accurately represent the position data returned by the reflected object at each depth point in each frame.

[0011] As a further improvement, the data preprocessing process in step S2 includes downsampling processing and radius-based outlier removal.

[0012] As a further improvement, the secondary motion verification process of the wheel odometer is triggered in step S4. Specifically, the current odometer data (x0, y0) is recorded at the moment of triggering the anti-fall protection. At this time, the robot is still allowed to move forward, and the Euclidean distance s between the current odometer data (x, y) and (x0, y0) is calculated in real time. When s exceeds the threshold, the robot is allowed to release the anti-fall trigger result and stop.

[0013] The present invention also provides a device for executing a method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer, which includes at least a processor and a memory, wherein the memory stores computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory, including data input, data processing, and data output modules, to execute the above-mentioned method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer.

[0014] The present invention also provides a computer-readable storage medium storing a computer program or instruction. When the computer program or instruction is executed, the above-mentioned method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer is implemented.

[0015] The beneficial effects of the present invention are: effectively improving the robot's ability to detect potential falling environments when getting on and off the elevator, reducing the problem of failure in getting on and off the elevator caused by frequent triggering of anti-fall protection during movement, and improving the robot's operational safety and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] An embodiment of the present invention provides a method for preventing a robot from falling when going up or down an elevator based on a depth point cloud and a wheel odometer, comprising the following steps:

[0020] Acquire the depth image information of the depth camera carried by the mobile robot, convert the depth image information into the universal PCL point cloud format, and rotate the data so that the point cloud data can accurately represent the position data returned by the reflected object at each depth point in each frame (that is, converted into the format of x, y, z).

[0021] Set up an anti-fall detection area, which is used to detect whether there are potential dangerous areas such as depressions or protrusions in front of the robot that may affect the smooth operation of the robot.

[0022] The point cloud information in the anti-fall detection area is first preprocessed, including downsampling and removal of outliers based on radius. This operation not only reduces the amount of point cloud data without affecting the overall shape and characteristics of the point cloud, but also improves the efficiency of subsequent calculations.

[0023] The pre-processed point cloud data is then divided into small grids according to different horizontal and vertical resolutions. Logical judgment is performed on the point cloud data within the grid. That is, if the depth information of the point cloud falling within the grid is lower than a certain threshold of the ground and the number of point clouds is higher than a certain threshold, the grid will be judged as a potential fall risk and counted. All grids are looped through to count the number of grids with potential fall risks. When the number of grids exceeds a certain threshold, it is determined to trigger anti-fall protection and the status is released. In this way, it can be accurately determined whether there is a fall area in the area in front of the robot.

[0024] Since the depth camera is installed on the robot body, there is a detection blind spot. In the special scenario where the robot is going up and down the elevator, there is a large gap in front of the elevator that can easily trigger the anti-fall function. In order to avoid the problem that the robot cannot accurately stop in the center of the elevator due to the triggering of the anti-fall function, the secondary motion verification of the wheel odometer is introduced. That is, when the anti-fall state is triggered during the robot going up the elevator, but there is still a certain redundant distance between the robot and the front end of the elevator, the robot can be allowed to move forward a certain distance (such as 5cm). When the distance is exceeded, the robot stops moving and reports an alarm. This ensures that the robot stops accurately in the center of the elevator under the condition that the anti-fall detection is accurate and effective.

[0025] A specific implementation step is as follows:

[0026] The embodiment of the invention provides a method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer, comprising steps S1 to S4:

[0027] S1, the depth image is converted into a depth point cloud and the coordinate axis is rotated.

[0028] In an embodiment of the present invention, depth image information of a depth camera carried by a mobile robot is obtained, the depth image information is converted into a universal PCL point cloud format, and the data is rotated so that the point cloud data can accurately represent the position data returned by the reflected object at each frame of the depth point (i.e., converted into the format of x, y, z).

[0029] S2, perform two straight-through filtering on the point cloud data to screen out the anti-fall detection area, and then remove outliers based on the radius of the point cloud in the area.

[0030] The anti-fall detection area is mainly determined by the robot parameters. For example, the width of the robot (0.6m) is used as the width of the detection area, and the length of the detection area is 0.3m to 1.2m in front of the robot.

[0031] In an embodiment of the present invention, an anti-fall detection area is provided, which is used to detect whether there are potential dangerous areas such as depressions or protrusions in front of the robot that may affect the smooth operation of the robot.

[0032] The point cloud information in the anti-fall detection area is first preprocessed, including downsampling and removal of outliers based on radius. This operation not only reduces the amount of point cloud data without affecting the overall shape and characteristics of the point cloud, but also improves the efficiency of subsequent calculations.

[0033] S3, then divides the detection area into grids according to the horizontal and vertical resolutions, traverses all grids to perform anti-fall detection calculations, and outputs the calculation results.

[0034] In an embodiment of the present invention, the pre-processed point cloud data is further divided into small grids according to different horizontal and vertical resolutions, and a logical judgment is performed on the point cloud data within the grid, that is, if the depth information of the point cloud falling within the grid is lower than a certain threshold of the ground and the number of point clouds is higher than a certain threshold (these two thresholds are set artificially based on experience, such as 0.1m below the ground threshold and 60% of the total number of point clouds), the grid is determined to be a potential fall risk and counted, and all grids are looped through to count the number of grids with potential fall risks. When the number of grids exceeds a certain threshold, it is determined to trigger anti-fall and the status is released. In this way, it can be accurately determined whether there is a fall area in the area in front of the robot.

[0035] S4: Trigger the wheel odometer secondary motion check based on the calculation result. If the calculation result is true and the odometer movement exceeds the 5 cm threshold at this time, the final anti-fall trigger result is issued. Otherwise, the anti-fall not triggered result is issued. The process of triggering the wheel odometer secondary motion check specifically records the current odometer data (x0, y0) at the moment of anti-fall triggering. At this time, the robot is still allowed to move forward. The Euclidean distance s between the current odometer data (x, y) and (x0, y0) is calculated in real time. When s exceeds the threshold, the robot is triggered to issue an anti-fall trigger result and stop.

[0036] In an embodiment of the present invention, due to the presence of a detection blind spot in the depth camera installed on the robot body and the special scenario where the robot is in the elevator, there is a large area in front of the elevator that can easily trigger the anti-fall function. In order to avoid the problem that the robot cannot accurately stop in the center of the elevator due to the triggering of the anti-fall function, a secondary motion verification of the wheeled odometer is introduced. That is, when the anti-fall state is triggered during the robot going up the elevator, but the actual robot still has a certain redundant distance from the front end of the elevator, the robot can be allowed to move forward a certain distance (such as 5 cm). When the distance is exceeded, the robot stops moving and reports an alarm message. This ensures that the robot stops accurately in the center of the elevator under the condition that the anti-fall detection is accurate and effective.

[0037] An embodiment of the present invention provides a method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheeled odometer, comprising obtaining image information returned by a depth camera in the space in which the robot is moving; screening out an anti-fall detection area and performing data preprocessing such as through filtering and outlier removal to reduce controller calculations; performing grid segmentation on the point cloud data within the detection area with different resolutions in the horizontal and vertical directions, traversing all grids, and counting the grids when the point cloud height within the grid is lower than a certain distance from the plane. If the cumulative number of horizontal and vertical grids exceeds a certain threshold, it is determined that anti-fall is triggered; since the robot is in a special scenario of going up and down an elevator and is very likely to trigger anti-fall, which may cause the robot to fail to stop accurately in the center of the elevator, a wheeled odometer is introduced for secondary motion verification. When the anti-fall detection is triggered when going up and down the elevator, the robot is still allowed to move a certain distance (such as 5 cm). If the threshold is exceeded, the robot movement is stopped and the anti-fall triggering result is reported. The robot's elevator anti-fall method based on deep point cloud and wheel odometry effectively improves the robot's ability to detect potential falling environments when getting on and off the elevator, reduces the problem of elevator failure caused by frequent triggering of anti-fall during movement, and improves the robot's operational safety and stability.

[0038] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented through hardware associated with computer program instructions. The program can be stored in any computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0039] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, the above is only a preferred embodiment of the present invention. Since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited to this. Any technical personnel familiar with this technical field is within the technical scope disclosed by the present invention. For ordinary technical personnel in this technical field, changes or replacements that can be easily thought of should be covered within the protection scope of the present invention without departing from the principle of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for preventing robots from falling when going up and down elevators based on depth point cloud and wheel odometer, characterized in that The following steps are involved: S1, obtains depth image information, converts the depth image into a depth point cloud and rotates the coordinate axis; S2, filter out the anti-fall detection area and then pre-process the point cloud in the area; S3: Divide the detection area into grids according to the horizontal and vertical resolutions, traverse all grids to perform anti-fall detection calculations, and output the calculation results; Specifically, the pre-processed point cloud data is divided into small grids according to different horizontal and vertical resolutions. The point cloud data in the grid is logically judged. That is, if the depth information of the point cloud falling in the grid is lower than the ground threshold and the number of point clouds is higher than the threshold, the grid is judged as a potential fall risk and counted. All grids are traversed in a loop to count the number of grids with potential fall risks. When the number of grids exceeds a certain threshold, it is determined to trigger the fall protection and the status is released. S4, triggering the secondary motion check of the wheel odometer according to the calculation result. If the calculation result is true and the odometer movement exceeds the threshold at this time, the final anti-fall trigger result is released, otherwise the anti-fall non-trigger result is released.

2. The method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheeled odometer according to claim 1 is characterized in that: In step S1, the depth image information is acquired by a depth camera carried by a mobile robot.

3. The method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheeled odometer according to claim 1 or 2, characterized in that: The specific process of step S1 is to obtain the depth image information of the depth camera carried by the mobile robot, convert the depth image information into the universal PCL point cloud format, and rotate the data at the same time so that the point cloud data can accurately represent the position data returned by the object reflected by each depth point in each frame.

4. The method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheeled odometer according to claim 1 is characterized in that: The data preprocessing process in step S2 includes downsampling and radius-based outlier removal.

5. The method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheeled odometer according to claim 1 or 4, characterized in that: The process of screening out the anti-fall detection area in step S2 is specifically to perform two straight-through filtering on the point cloud data.

6. The method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheeled odometer according to claim 1, characterized in that: Step S4 triggers the secondary motion verification process of the wheel odometer. Specifically, the current odometer data (x0, y0) is recorded at the moment of triggering the anti-fall protection. At this time, the robot is still allowed to move forward. The Euclidean distance s between the current odometer data (x, y) and (x0, y0) is calculated in real time. When s exceeds the threshold, the robot is allowed to release the anti-fall trigger result and stop.

7. A device for implementing a method for preventing robots from falling when going up and down elevators based on depth point clouds and wheel odometers, characterized in that: The method comprises at least a processor and a memory, wherein the memory stores computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory, including data input, data processing, and data output modules, and executes the method for preventing robots from getting on and off elevators based on depth point clouds and wheel odometers as described in claim 1.

8. A computer-readable storage medium, characterized in that: A computer program or instruction is stored, and when the computer program or instruction is executed, the method for preventing a robot from falling when going up and down an elevator based on a depth point cloud and a wheel odometer as described in claim 1 is implemented.

Images