Feature point-based mobile robot accurate parking method

Abstract

The application discloses a feature point-based accurate parking method of a mobile robot, and belongs to the technical field of indoor positioning of the mobile robot. In order to solve the problem that the prior art cannot accurately park at a target point at a short distance, the method comprises the following steps: 1, the mobile robot is guided by a laser radar, and autonomously navigates to reach the edge of an obstacle inflation area required to be parked; 2, after the mobile robot reaches the edge of the obstacle inflation area, the autonomous navigation is abandoned, the laser radar extracts feature points and reads data values of the feature points, and the data values of the feature points are transmitted to an industrial computer in the mobile robot; 3, the industrial computer analyzes the acquired feature point distance data values and the extracted feature point distance value information under the same angle, and then adjusts the robot pose; 4, after the robot pose is adjusted, the laser radar continues to guide the forward movement; the distance between the indoor mobile robot and the obstacle is judged through the first feature point d1 data value; and the forward movement is stopped when the distance between the two is a preset value.

Description

Technical Field

[0001] This invention belongs to the field of indoor positioning technology for mobile robots, specifically relating to a method for accurate docking of mobile robots based on feature points. Background Technology

[0002] In recent years, the functionality of mobile robots has become increasingly powerful, benefiting fields such as transportation, healthcare, and industry. A mobile robot is a comprehensive system integrating environmental perception, dynamic decision-making and planning, and behavioral control and execution. Localization technology, as one of the key technologies for mobile robots, uses LiDAR, odometry, and inertial measurement units to detect the robot's pose and position in real time. This allows for more precise control by issuing commands, enabling the robot to travel along the desired path. The diversity of localization algorithms directly determines the application environment and robustness of the localization, affecting the accuracy and reliability of the mobile robot during movement. Therefore, multiple accurate localization methods are needed during the movement of mobile robots.

[0003] For the localization of mobile robots, scholars at home and abroad have proposed a variety of algorithms and processing methods, such as visual localization, lidar localization, and multi-sensor fusion localization.

[0004] Chinese patent publication number "CN 114489054 A", entitled "Method and Robot for Controlling Robot to Stop at Target Points", proposes a method for controlling a robot to stop at target points, which uses at least two target points for stopping orientation. During movement, the robot selects one of the at least two stopping orientations and stops near the target point. While this method can reduce the difference in movement path angles, it cannot stop close to the target point when there are obstacles there. The stopping distance is determined by the expansion radius of the obstacle, making accurate stopping at the target point impossible. Summary of the Invention

[0005] To address the problem of existing technologies failing to accurately dock at target points at close range, this invention provides a method for accurate docking of mobile robots based on feature points.

[0006] The technical solution of this invention to solve the technical problem is:

[0007] A method for accurate docking of mobile robots based on feature points, characterized by the following steps:

[0008] Step 1: Guided by LiDAR, the mobile robot autonomously navigates to the edge of the obstacle expansion zone where it needs to stop;

[0009] Step 2: After the mobile robot reaches the edge of the obstacle expansion zone, it abandons autonomous navigation. The LiDAR extracts feature points and reads the data values ​​of the feature points, and transmits the data values ​​of the feature points to the industrial control computer inside the mobile robot.

[0010] Step 3: The industrial control computer analyzes the acquired feature point distance data and the feature point distance information extracted at the same angle, and then adjusts the robot pose.

[0011] Step 4: After the robot's pose is adjusted, it continues to move forward under the guidance of the LiDAR; the distance between the indoor mobile robot and the obstacle is determined by the data value of the first feature point d1; the robot stops moving forward when the distance between the two is a preset value.

[0012] The feature point extraction method in step 2 is as follows: the point emitted by the center of the lidar on the obstacle is taken as the first feature point A. To the left of the first feature point, the second feature point B is extracted at a fixed angle. To the right of the first feature point, the third feature point C is extracted at the same fixed angle. After the feature points are determined, the distance data value measured by the lidar of the feature points is read.

[0013] The method for analyzing the distance data of feature points described in step 3 is as follows: Let the distance data measured by the lidar of the first feature point A be d1, the distance data measured by the lidar of the second feature point B be d2, and the distance data measured by the lidar of the third feature point C be d3. Let the angle of the second feature point B be α and the angle of the third feature point C be β.

[0014] Data analysis is divided into two stages: the first stage is coarse adjustment, and the second stage is fine adjustment;

[0015] The coarse adjustment involves comparing and analyzing the values ​​of d2 and d3 to adjust them so that the robot's front is parallel to the obstacle. The robot's pose adjustment satisfies the following formula:

[0016] |d2-d3|≤1

[0017] The fine adjustment involves comparing and analyzing the ratios of d2 and d1 to the cosine of angle α, and comparing and adjusting the ratios of d3 and d1 to the cosine of angle β. This allows the robot to further correct its pose based on the coarse adjustment, achieving a more precise parallel position. The robot's pose adjustment satisfies the following formula:

[0018]

[0019] The fine-tuning phase is completed when the result is |d2-d3|≤0.01.

[0020] The beneficial effects of this invention are as follows: The feature-point-based method for accurate docking of a mobile robot allows the robot to overcome expansion radius constraints after approaching an obstacle. Feature point extraction confirms the robot's pose, followed by coarse and fine adjustments to regulate the robot's pose and achieve the correct posture. This ensures the robot continues to move forward in the correct posture, approaching the obstacle at close range and achieving true target-point docking. Attached Figure Description

[0021] Figure 1 This is a structural diagram of an indoor mobile robot that implements the functions of an embodiment of the present invention. 1 represents the robot body; 2 represents the lidar; and 3 represents the display.

[0022] Figure 2 A flowchart illustrating the method for accurate docking of an indoor mobile robot based on feature points, as provided in an embodiment of the present invention.

[0023] Figure 3 The following are models of indoor mobile robot application scenarios provided in the embodiments of the present invention: (a) is a model of an indoor mobile robot reaching the vicinity of an obstacle, and (b) is a model of an indoor mobile robot being unable to move forward due to its expansion radius.

[0024] Figure 4 This is a geometric diagram illustrating the forward docking of the feature point extraction algorithm provided in this embodiment of the invention.

[0025] Figure 5 This is a geometric diagram illustrating the feature point extraction algorithm provided in this embodiment of the invention during lateral docking. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings.

[0027] The present invention provides a method for accurate docking of a mobile robot based on feature points, implemented using a mobile robot 1 equipped with a lidar 2 and a display 3, as follows: Figure 1 As shown, a lidar 2 is placed on one side of the upper surface of the mobile robot 1, and a display 3 is placed on the other side.

[0028] like Figure 2 As shown, a method for accurate docking of a mobile robot based on feature points is described. This method specifically includes the following steps:

[0029] Step 1: Guided by LiDAR 2, the indoor mobile robot 1 starts from its starting point and autonomously plans its path to the target point near the obstacle where it needs to stop. Display 3 shows the real-time path planning status of the indoor mobile robot. A model of the indoor mobile robot application scenario is shown below. Figure 3 As shown, Figure 3 (a) is a model of an indoor mobile robot reaching the vicinity of an obstacle. Figure 3(b) is a model of an indoor mobile robot that cannot move forward due to its expansion radius.

[0030] Step 2: Upon reaching the vicinity of the obstacle, autonomous navigation is abandoned. The LiDAR extracts feature points and reads their data values, transmitting these values ​​to the industrial control computer inside the mobile robot.

[0031] Feature point extraction method: Point A, emitted from the center of the LiDAR on the obstacle, is taken as the first feature point. To the left of the first feature point, a second feature point B is extracted at a fixed angle. To the right of the first feature point, a third feature point C is extracted at the same fixed angle. After the feature points are determined, the LiDAR data values ​​of the feature points are read.

[0032] Forward parking:

[0033] After the indoor mobile robot autonomously navigates to the vicinity of an obstacle, it stops its autonomous navigation function. It then activates the feature point docking algorithm, the geometric principle of which is shown in the diagram below. Figure 4 As shown. First, real-time LiDAR data is read, and the LiDAR data at the 0-angle position is extracted as the first feature point A, and the distance value is read and recorded as d1. Then, to the left of the first feature point, the angle α is set to 30 degrees as the second feature point B, and the distance value is read and recorded as d2. To the right of the first feature point, the angle β is set to 30 degrees as the third feature point C, and the distance value is read and recorded as d3.

[0034] Lateral parking:

[0035] After the indoor mobile robot autonomously navigates to the vicinity of an obstacle, it stops its autonomous navigation function. It then activates the feature point docking algorithm, the geometric principle of which is shown in the diagram below. Figure 5 As shown. First, real-time LiDAR data is read. When the left side of the indoor mobile robot approaches the obstacle, the LiDAR data at the 90-degree angle is extracted as the first feature point A, and the distance value is recorded as d1. When the right side of the indoor mobile robot approaches the obstacle, the LiDAR data at the 270-degree angle is extracted as the first feature point A, and the distance value is recorded as d1. Then, to the left of the first feature point, the angle α is set to 30 degrees as the second feature point B, and the distance value is recorded as d2. To the right of the first feature point, the angle β is set to 30 degrees as the third feature point C, and the distance value is recorded as d3.

[0036] Step 3: Analyze the acquired feature point data values. Analyze the distance values ​​of the extracted feature points at the same angle;

[0037] Forward parking:

[0038] In the embodiment of the present invention, after the eigenvalue extraction of the indoor mobile robot is completed, a rough pose adjustment is performed. In the rough adjustment stage, the indoor mobile robot satisfies the following formula:

[0039] |d2 - d3| ≤ 1

[0040] When d2 < d3, the indoor mobile robot swings to the right for adjustment; when d2 > d3, the indoor mobile robot swings to the left for adjustment. It is adjusted until |d2 - d3| ≤ 1, and at this time, the rough adjustment stage is completed.

[0041] After the rough adjustment stage is completed, the indoor mobile robot enters the fine adjustment stage. This adjustment stage is based on the following formula:

[0042]

[0043] When d2 < d3, the indoor mobile robot swings to the right for adjustment; when d2 > d3, the indoor mobile robot swings to the left for adjustment. It is adjusted until |d2 - d3| ≤ 0.01, and at this time, the fine adjustment stage is completed.

[0044] Lateral docking:

[0045] In the embodiment of the present invention, after the eigenvalue extraction of the indoor mobile robot is completed, a rough pose adjustment is performed. In the rough adjustment stage, the indoor mobile robot satisfies the following formula:

[0046] |d2 - d3| ≤ 1

[0047] When d2 < d3, the indoor mobile robot swings to the left for adjustment; when d2 > d3, the indoor mobile robot swings to the right for adjustment. It is adjusted until |d2 - d3| ≤ 1, and at this time, the rough adjustment stage is completed.

[0048] After the rough adjustment stage is completed, the indoor mobile robot enters the fine adjustment stage. This adjustment stage is based on the following formula:

[0049]

[0050] When d2 < d3, the indoor mobile robot swings to the left for adjustment; when d2 > d3, the indoor mobile robot swings to the right for adjustment. It is adjusted until |d2 - d3| ≤ 0.01, and at this time, the fine adjustment stage is completed.

[0051] Step 4: When the distance value information meets the set requirements, the lidar continues to guide the forward movement. When docking forward, a Y-direction speed is given, and when docking laterally, an X-direction speed is given. The distance between the indoor mobile robot and the obstacle is judged through the data value of the first feature point d1. It stops moving forward when the preset distance between the two is 8 cm.

Claims

1. A method for accurate docking of mobile robots based on feature points, characterized by: The method specifically includes the following steps: Step 1: Guided by LiDAR, the mobile robot autonomously navigates to the edge of the obstacle expansion zone where it needs to stop; Step 2: After the mobile robot reaches the edge of the obstacle expansion zone, it abandons autonomous navigation. The LiDAR extracts feature points and reads the data values ​​of the feature points, and transmits the data values ​​of the feature points to the industrial control computer inside the mobile robot. The feature point extraction method is as follows: the point emitted by the center of the lidar on the obstacle is taken as the first feature point A. To the left of the first feature point, the second feature point B is extracted at a fixed angle. To the right of the first feature point, the third feature point C is extracted at the same fixed angle. After the feature points are determined, the distance data value measured by the lidar of the feature points is read. Step 3: The industrial control computer analyzes the acquired feature point distance data and the feature point distance information extracted at the same angle, and then adjusts the robot pose. The method for analyzing the feature point distance data is as follows: Let the distance data measured by the lidar for the first feature point A be denoted as... The distance data measured by the lidar at the second feature point B is... The distance data measured by the lidar at the third feature point C is... The angle for extracting the second feature point B is α, and the angle for extracting the third feature point C is β. Data analysis is divided into two stages: the first stage is coarse adjustment, and the second stage is fine adjustment; Step 4: After the robot's pose is adjusted, it continues to move forward under the guidance of the LiDAR; passing the first feature point. The data value determines the distance between the indoor mobile robot and the obstacle; it stops moving when the distance between the two is a preset value. The coarse adjustment is as follows: This process... and Numerical comparisons and adjustments are made to ensure the robot is parallel to the obstacle. The robot's pose adjustment satisfies the following formula: | |≤1； The fine-tuning is as follows: This process is for and Comparative analysis with the ratio of angle α and cosine. and By comparing and analyzing the ratio of angle β to cosine, the robot's pose is further corrected based on the initial coarse adjustment, achieving a more precise parallel position. The robot's pose adjustment is based on the following formula: ； Until adjusted to | |≤0.01, at this point the fine-tuning phase is complete.

Images