Intelligent stair climbing method of stair climbing robot based on deformable wheels
By designing an intelligent stair climbing algorithm system, using environmental perception and trajectory planning optimization, the stability and efficiency problems of the deformed wheel mobile robot in the stair environment are solved, and efficient and smooth stair climbing capabilities and high-precision stair detection are achieved.
Patent Information
- Application Number
- CN202411034444.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-31
AI Technical Summary
Existing mobile robots based on deformation wheels lack effective control methods and trajectory optimization algorithms when facing stairs or steps, resulting in low stability and efficiency of climbing stairs and insufficient environmental perception accuracy.
An intelligent stair climbing algorithm system was designed, including an environment perception subsystem, wheel deformation control and balance control subsystem, and compensated for the lidar point cloud through IMU information, and used Ceres optimizer and mean filtering to perform point cloud processing to generate a stair map The relative position of the robot is calculated, the deformation wheel radius and the rotation angle of the rim are calculated, and the trajectory planning algorithm is used to optimize the wheel centroid trajectory.
It realizes efficient climbing of stairs and steps without changing the direction of the robot, improves the efficiency and stability of the robot's tasks, provides high-precision stair detection and smooth trajectory planning, and improves the success rate of overcoming steps. .
Smart Images

Figure CN119002477A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an intelligent stair climbing method of a mobile robot based on deformable wheels, belonging to the technical field of intelligent equipment. Background Art
[0002] With the increasing cost of labor, reducing labor costs can greatly increase operating profits. The widespread use of mobile robots can not only reduce labor costs, but also improve work efficiency. This phenomenon has attracted the attention of the industry and academia to the field of mobile robots. A variety of mobile robots have been launched one after another. The mobile robot based on deformable wheels combines the advantages of wheeled robots and bionic foot robots, and can efficiently pass through obstacles such as stairs. It has attracted the attention of many scholars in recent years. However, there are currently few studies on the construction of intelligent stair climbing systems for mobile robots based on deformable wheels, and there are many research gaps: first, there is almost no research on the control of mobile robots based on deformable wheels, mainly involving the deformation control of deformable wheels and the balance control of two-wheeled mobile robots based on deformable wheels; then, there is very little research on the trajectory optimization algorithm for deformable wheels to overcome steps; finally, the research accuracy of the environment perception algorithm for stair scenes cannot meet the deformation requirements of the deformable wheels when overcoming steps. In summary, it is very necessary to build an intelligent stair climbing algorithm system for the designed mobile robot based on deformable wheels, which includes environmental perception, wheel deformation control, balance control of two-wheeled mobile robots based on deformable wheels, and trajectory optimization algorithm. Summary of the invention
[0003] In order to solve some challenges faced by existing mobile robots when facing stairs or step-like obstacles, the present invention builds an intelligent stair climbing algorithm system for the designed two-wheeled mobile robot based on deformable wheels. The system mainly includes an environmental perception subsystem, a wheel deformation control and a two-wheeled mobile robot balance control subsystem based on deformable wheels, and a wheel center of mass trajectory optimization subsystem. The intelligent stair climbing algorithm including the above subsystems improves the stability and trajectory smoothness of the robot when climbing stairs.
[0004] The technical solution adopted by the present invention to solve the above problems is:
[0005] The intelligent stair-climbing method of the stair-climbing robot based on deformable wheels described in the present invention specifically comprises the following steps:
[0006] Step 1: Use IMU information to compensate for the motion distortion of the lidar point cloud to obtain a high-precision and high-reliability lidar point cloud;
[0007] Step 2: Project, plane fit and dimension the step point cloud through Ceres optimizer and mean filter method to obtain the height and depth information of each step;
[0008] Step 3: Use the compensated point cloud to perform scan matching and check whether the point cloud data is degraded. If so, solve the degradation problem and finally generate a map of the stairs and the relative position of the robot.
[0009] Step 4, providing position information and stair size information for the deformation time and deformation amount of the wheels of the mobile robot; the stair size information includes the step depth D and the step height H;
[0010] Step 5: Obtain the deformation wheel radius r and wheel rim rotation angle θ required for the step of this size through inverse kinematics calculation;
[0011] Step 6: Pass (r, θ) and the distance L of the robot relative to the stairs obtained in the stair detection algorithm to the trajectory planning algorithm;
[0012] Step 7: Plan the rotation angle of the wheel at any time in the planning algorithm And the rotation angle q of the two BLDC motors that control the deformation wheel (r, θ) des (q 1des ,q 2des ).
[0013] Furthermore, the specific steps of the inverse kinematics calculation in step 5 are:
[0014] Step 501, solving the deformation wheel radius r and the rim rotation angle θ required to overcome the step of any size;
[0015]
[0016] In formula (1), D represents the depth of the step, H represents the height of the step, and r0 represents;
[0017] Step 502: According to the radius r and the rim rotation angle θ of the required deformed wheel, the rotation angles (q1, q2) of the solid line and the dashed line sun gears are respectively solved:
[0018]
[0019] In formulas (2) and (3), the radius r of the deformed wheel and the rotation angle θ of the wheel rim are related to P 3y and ξ, so using (P 3y ,ξ) represents the two degrees of freedom of the deformation wheel (r,θ);
[0020] Step 503: From formula (2) and formula (3), it can be obtained that:
[0021]
[0022] Step 504: Expected (P 3y,ξ) corresponding to the solid and dashed sun gear rotation angle (q1,q2) formula:
[0023]
[0024] Furthermore, the specific steps of the staircase detection algorithm in step 6 are:
[0025] Step A: Use IMU information to perform motion distortion compensation on the LiDAR point cloud, so as to obtain a LiDAR point cloud with high accuracy and high reliability;
[0026] Step B: The dedistorted point cloud is used to establish a stair point cloud map and detect the size of the stair steps;
[0027] Step C: Output the staircase mapping result through degradation detection;
[0028] Step D, extracting stair point clouds from the laser radar point clouds, setting a threshold for the horizontal angle of the laser radar point clouds, and determining the point clouds that meet the threshold range as the point clouds of the area of interest;
[0029] Step E: Project the point cloud in the region of interest to the Z axis of the laser radar coordinate system, perform XZ plane fitting of the point cloud, and use the Ceres optimizer to solve the straight line fitting parameters of the point cloud in the XZ plane. If the fitting parameters meet the threshold requirements, the point cloud in the region of interest is identified as a stair point cloud.
[0030] Step F: Perform size detection on the identified stair point cloud.
[0031] Furthermore, the specific steps of the trajectory planning algorithm in step 6 are:
[0032] Step a: Analyze the kinematics of the deformable wheel overcoming the step, and mark several landmark points O that the deformable wheel must pass through to overcome the step i , calculate the position expression of each landmark point, and stipulate that the speed and acceleration of the deformable wheel when passing the first and last landmark points are zero;
[0033] Step b: connect two adjacent landmark points with a fifth-order Bezier curve, and the optimizer solves the optimal trajectory under the constraint conditions;
[0034] Step c: Analyze the possible interference of the deformation wheel in overcoming the step process, and establish a mathematical model to avoid interference. Constrain the movement and shape of the deformation wheel in the form of inequalities or equations, which is essentially the rotation angle of the deformation wheel. And the radius r of the deformed wheel and the rotation angle θ of the wheel rim are constrained;
[0035] Step d: In order to make the trajectory of the deformable wheel overcoming the step as smooth as possible and constrain the center of mass of the deformable wheel to reciprocate up and down, an objective function about acceleration is established, and it is expected that the sum of the average acceleration and the maximum acceleration in the entire trajectory optimization process is minimized;
[0036] Step e: using the particle swarm optimization algorithm as the optimizer, solving the wheel center of mass trajectory that can minimize the objective function under the motion constraint, and solving the unknown parameters at the same time; in the particle swarm optimization algorithm, the number of unknown parameters is expressed as the search dimension of the particle swarm algorithm.
[0037] The beneficial effects of the present invention are:
[0038] 1. The present invention can not only efficiently perform tasks on flat ground, but also can change the shape of the wheels when encountering stairs and continue to perform tasks efficiently by climbing the stairs;
[0039] 2. The robot system of the present invention can achieve going up and down stairs, bidirectional climbing of steps and stair obstacles by only changing the rim angle direction of the wheels without changing the robot's forward direction, thereby improving the efficiency of the robot in performing tasks;
[0040] 3. The stair detection algorithm of the present invention provides a high-precision stair step size for the robot's wheel deformation, which helps the mobile robot system to overcome the steps and continue to perform the task;
[0041] 4. The trajectory planning algorithm of the present invention plans a smooth trajectory for the deformable wheel to overcome steps or stairs, which helps to improve the stability and success rate of the robot in overcoming steps. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 It is a schematic diagram of the inverse kinematics algorithm;
[0043] Figure 2 is a flow chart of the intelligent perception algorithm of the present invention;
[0044] Figure 3 This is a control diagram of a stair-climbing robot system based on deformable wheels;
[0045] Figure 4 This is a schematic diagram of stair climbing based on a deformable wheel stair climbing robot system;
[0046] Figure 5 It is a structural diagram of the deformable wheel climbing robot system. DETAILED DESCRIPTION
[0047] Specific implementation method 1: Figures 1 to 4 As shown, an intelligent stair climbing method of a stair climbing robot based on deformable wheels specifically includes:
[0048] Step 1: Use IMU information to compensate for the motion distortion of the lidar point cloud to obtain a high-precision and high-reliability lidar point cloud;
[0049] Step 2: Project, plane fit and dimension the step point cloud through Ceres optimizer and mean filter method to obtain the height and width information of each step;
[0050] Step 3: Use the compensated point cloud to perform scan matching and check whether the point cloud data is degraded. If so, solve the degradation problem and finally generate a map of the stairs and the relative position of the robot.
[0051] Step 4, providing position information and stair size information for the deformation time and deformation amount of the wheels of the mobile robot; the stair size information includes the step depth D and the step height H;
[0052] Step 5: Obtain the deformation wheel radius r and wheel rim rotation angle θ required for the step of this size through inverse kinematics calculation;
[0053] Step 6: Pass (r, θ) and the distance L of the robot relative to the stairs obtained in the stair detection algorithm to the trajectory planning algorithm;
[0054] Step 7: Plan the rotation angle of the wheel at any time in the planning algorithm And the rotation angle q of the two BLDC motors that control the deformation wheel (r, θ) des (q 1des ,q 2des ).
[0055] The rotation angle q of the deformation wheel des (q 1des ,q 2des ) is the desired rotation angle in the deformation control of the deformation wheel. The rotation angle of the deformation wheel is controlled by the PD controller, and The actual rotation angle of the BLDC that controls the deformation of the deformation wheel is fed back to the PD controller.
[0056] Expected wheel rotation angle generated by trajectory planning This is the desired state of the wheel, and the IMU measures the vehicle body's rotation angle γ around the y-axis and the angle around the z-axis. The reference angle used to control the balance and steering of the mobile robot is passed into the balance control algorithm. The difference from the balance control algorithm of the traditional two-wheeled balance mobile robot is that after the wheel is deformed, the line connecting the center of mass of the wheel and the contact point between the wheel rim and the ground is no longer perpendicular to the ground, so the actual angle of the wheel is It also affects the balance of the mobile robot, so it needs to be passed to the balance control algorithm. Under the action of the balance control algorithm, the balance of the mobile robot during the stair climbing process is achieved.
[0057] like Figure 5 As shown, the deformable wheel stair climbing robot system includes a deformable wheel on each side; a rotation drive system for each deformable wheel; a battery to provide energy; a control board with Orange Pi as the core; a radar and an IMU.
[0058] Specific implementation method 2: Figures 1 to 4 As shown, the specific steps of inverse kinematics calculation in step 5 are:
[0059] Step 501, solving the deformation wheel radius r and the rim rotation angle θ required to overcome the step of any size;
[0060]
[0061] In formula (1), W represents the depth of the step, H represents the height of the step, and r0 represents;
[0062] Step 502: According to the radius r of the required deformed wheel and the rim rotation angle θ, solve the rotation angles (q1, q2) of the red and green sun gears:
[0063]
[0064] In formulas (2) and (3), the radius r of the deformed wheel and the rotation angle θ of the wheel rim are related to P 3y and ξ, so using (P 3y ,ξ) represents the two degrees of freedom of the deformation wheel (r,θ);
[0065] Step 503: From formula (2) and formula (3), it can be obtained that:
[0066]
[0067] Step 504: Expected (P 3y ,ξ) corresponding to the red and green sun gear rotation angle q1,q2 formula:
[0068]
[0069] Specific implementation method three: Figures 1 to 4 As shown, the specific steps of the staircase detection algorithm in step 6 are:
[0070] Step A: Use IMU information to perform motion distortion compensation on the LiDAR point cloud, so as to obtain a LiDAR point cloud with high accuracy and high reliability;
[0071] Step B: The dedistorted point cloud is used to establish a stair point cloud map and detect the size of the stair steps;
[0072] Step C: Output the staircase mapping result through degradation detection;
[0073] Step D, extracting stair point clouds from the laser radar point clouds, setting a threshold for the horizontal angle of the laser radar point clouds, and determining the point clouds that meet the threshold range as the point clouds of the area of interest;
[0074] Step E: Project the point cloud in the region of interest to the Z axis of the laser radar coordinate system, perform XZ plane fitting of the point cloud, and use the Ceres optimizer to solve the straight line fitting parameters of the point cloud in the XZ plane. If the fitting parameters meet the threshold requirements, the point cloud in the region of interest is identified as a stair point cloud.
[0075] Step F: Perform size detection on the identified stair point cloud.
[0076] Specific implementation method four: Figures 1 to 4 As shown, the specific steps of the trajectory planning algorithm in step 6 are:
[0077] Step a: Analyze the kinematics of the deformable wheel overcoming the step, and mark several landmark points O that the deformable wheel must pass through to overcome the step i , calculate the position expression of each landmark point, and stipulate that the speed and acceleration of the deformable wheel when passing the first and last landmark points are zero;
[0078] Step b: connect two adjacent landmark points with a fifth-order Bezier curve, and the optimizer solves the optimal trajectory under the constraint conditions;
[0079] Step c: Analyze the possible interference of the deformation wheel in overcoming the step process, and establish a mathematical model to avoid interference. Constrain the movement and shape of the deformation wheel in the form of inequalities or equations, which is essentially the rotation angle of the deformation wheel. And the radius r of the deformed wheel and the rotation angle θ of the wheel rim are constrained;
[0080] Step d: In order to make the trajectory of the deformable wheel overcoming the step as smooth as possible and constrain the center of mass of the deformable wheel to reciprocate up and down, an objective function about acceleration is established, and it is expected that the sum of the average acceleration and the maximum acceleration in the entire trajectory optimization process is minimized;
[0081] Step e: using the particle swarm optimization algorithm as the optimizer, solving the wheel center of mass trajectory that can minimize the objective function under the motion constraint, and solving the unknown parameters at the same time; in the particle swarm optimization algorithm, the number of unknown parameters is expressed as the search dimension of the particle swarm algorithm.
[0082] The above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent replacement and improvement made to the above embodiments without departing from the content of the technical solution of the present invention, based on the technical essence of the present invention, within the spirit and principles of the present invention, still fall within the protection scope of the technical solution of the present invention.
Claims
1. An intelligent stair climbing method for a stair climbing robot based on deformable wheels, characterized in that: Specifically include: Step 1: Use IMU information to compensate for the motion distortion of the lidar point cloud to obtain a high-precision and high-reliability lidar point cloud; Step 2: Project, plane fit and dimension the step point cloud through Ceres optimizer and mean filter method to obtain the height and depth information of each step; Step 3: Use the compensated point cloud to perform scan matching and check whether the point cloud data is degraded. If so, solve the degradation problem and finally generate a map of the stairs and the relative position of the robot. Step 4, providing position information and stair size information for the deformation time and deformation amount of the wheels of the mobile robot; the stair size information includes the step depth D and the step height H; Step 5: Obtain the deformation wheel radius r and wheel rim rotation angle θ required for the step of this size through inverse kinematics calculation; Step 6: Pass (r, θ) and the distance L of the robot relative to the stairs obtained in the stair detection algorithm to the trajectory planning algorithm; Step 7: Plan the rotation angle of the wheel at any time in the planning algorithm And the rotation angle q of the two BLDC motors that control the deformation wheel (r, θ) des (q 1des ,q 2des ).
2. The intelligent stair climbing method of a stair climbing robot based on deformable wheels according to claim 1, characterized in that: The specific steps of inverse kinematics calculation in step 5 are: Step 501, solving the deformation wheel radius r and the rim rotation angle θ required to overcome the step of any size; In formula (1), D represents the depth of the step, H represents the height of the step, and r0 represents; Step 502: According to the radius r of the required deformed wheel and the rim rotation angle θ, solve the rotation angles (q1, q2) of the solid line and dotted line sun gears: In formulas (2) and (3), the radius r of the deformed wheel and the rotation angle θ of the wheel rim are related to P 3y and ξ, so using (P 3y ,ξ) represents the two degrees of freedom of the deformation wheel (r,θ); Step 503: From formula (2) and formula (3), it can be obtained that: Step 504: Expected (P 3y ,ξ) corresponding to the solid and dashed sun gear rotation angle (q1,q2):
3. The intelligent stair climbing method of a stair climbing robot based on deformable wheels according to claim 1, characterized in that: The specific steps of the stair detection algorithm in step 6 are: Step A: Use IMU information to perform motion distortion compensation on the LiDAR point cloud, so as to obtain a LiDAR point cloud with high accuracy and high reliability; Step B: The dedistorted point cloud is used to establish a stair point cloud map and detect the size of the stair steps; Step C: Output the staircase mapping result through degradation detection; Step D, extracting stair point clouds from the laser radar point clouds, setting a threshold for the horizontal angle of the laser radar point clouds, and determining the point clouds that meet the threshold range as the point clouds of the area of interest; Step E: Project the point cloud in the region of interest to the Z axis of the laser radar coordinate system, perform XZ plane fitting of the point cloud, and use the Ceres optimizer to solve the straight line fitting parameters of the point cloud in the XZ plane. If the fitting parameters meet the threshold requirements, the point cloud in the region of interest is identified as a stair point cloud. Step F: Perform size detection on the identified stair point cloud.
4. The intelligent stair climbing method of a stair climbing robot based on deformable wheels according to claim 1, characterized in that: The specific steps of the trajectory planning algorithm in step 6 are: Step a: Analyze the kinematics of the deformable wheel overcoming the step, and mark several landmark points O that the deformable wheel must pass through to overcome the step i , calculate the position expression of each landmark point, and stipulate that the speed and acceleration of the deformable wheel when passing the first and last landmark points are zero; Step b: connect two adjacent landmark points with a fifth-order Bezier curve, and the optimizer solves the optimal trajectory under the constraint conditions; Step c: Analyze the possible interference of the deformation wheel in overcoming the step process, and establish a mathematical model to avoid interference. Constrain the movement and shape of the deformation wheel in the form of inequalities or equations, which is essentially the rotation angle of the deformation wheel. And the radius r of the deformed wheel and the rotation angle θ of the wheel rim are constrained; Step d: In order to make the trajectory of the deformable wheel overcoming the step as smooth as possible and constrain the center of mass of the deformable wheel to reciprocate up and down, an objective function about acceleration is established, and it is expected that the sum of the average acceleration and the maximum acceleration in the entire trajectory optimization process is minimized; Step e: using the particle swarm optimization algorithm as the optimizer, solving the wheel center of mass trajectory that can minimize the objective function under the motion constraint, and solving the unknown parameters at the same time; in the particle swarm optimization algorithm, the number of unknown parameters is expressed as the search dimension of the particle swarm algorithm.
Citation Information
Patent Citations
Quadruped robot autonomous stair climbing method and system based on local environment perception
CN117270551A
Method to Use Depth Sensors on the Bottom of Legged Robot for Stair Climbing
US20240053761A1
Path planning method for wall-climbing robot
WO2023130673A1
Cited By
Offshore platform unmanned inspection robot and control method thereof
CN119748473A
An unmanned inspection robot for offshore platforms and a control method thereof
CN119748473B
