Robotic dog calibration method

Abstract

The invention relates to a robot dog calibration method. The robot dog calibration method comprises the following steps: measuring the height h of the robot dog at the positions of four limbs and performing zero setting; forwardly and reversely rotating the shoulder joint angle of the robot dog to find the high point of the shoulder The knee joint and the hip joint are controlled to move according to different angles respectively, different shoulder joint heights H are obtained, the optimal angle of the knee joint and the optimal angle of the hip joint and the length of a rod connecting the knee joint and the hip joint and the length of a rod connecting the hip joint and the foot end are obtained by taking the minimum average value of the difference values H and h as the target, and zero adjustment is conducted. The robot dog calibration method has the advantages that tools are not needed basically, and the length, the angle and the like can be calibrated.

Description

Technical Field

[0001] The present invention relates to the technical field of robots, and particularly to a calibration method for a robotic dog. Background Art

[0002] As a multi-legged robot, a robotic dog has a stronger load-bearing capacity and better stability than a bipedal robot, and a simpler structure than a hexapod or octopod robot. It can be used in fields such as military material transportation, hazardous environment detection, education, and entertainment.

[0003] To ensure the motion accuracy of the robotic dog, it is usually necessary to calibrate the motion parameters of the robotic dog. The existing calibration mainly focuses on calibrating the initial angle of the robotic dog. In the case of disassembly, over-limit use, impact, etc., the initial angle often changes, so calibration is required.

[0004] Generally, the calibration of a robotic dog is achieved by manually setting mechanical alignment at a special angle or by using a mechanical device to perform special position alignment. These methods either have insufficient accuracy and are difficult to debug, or require the configuration of tools and cannot be applied to all scenarios; calibrating only the angle also has certain limitations. The lengths of the thighs and calves and assembly errors will all affect the motion accuracy of the robotic dog to a certain extent. Therefore, it is also necessary to calibrate these parameters. Summary of the Invention

[0005] Based on this, the purpose of the present invention is to provide a calibration method for a robotic dog, which basically does not require tools and can calibrate lengths, angles, etc.

[0006] A calibration method for a robotic dog includes the following steps: measuring the height h of the robotic dog at the positions of its four limbs and zeroing it; rotating the shoulder joint angle of the robotic dog forward and backward to find the high point of the shoulder joint; respectively controlling the knee joint and hip joint to move at different angles to obtain different shoulder heights H, taking the minimum mean value of the difference between H and h as the target, and obtaining the optimal angles of the knee joint and hip joint and the lengths of the rod connecting the knee joint and hip joint and the rod connecting the hip joint and the foot end, and performing zero position adjustment.

[0007] The calibration method for a robotic dog according to the present invention can perform calibration in real time and quickly, improve the motion accuracy of the robotic dog, is convenient for calibration, and basically does not require tools; has high calibration accuracy and calibrates lengths, angles, etc.

[0008] Further, measuring the height h of the robotic dog at the positions of its four limbs and zeroing it includes the following steps: setting an attitude sensor in the middle of the top of the body of the robotic dog, and the attitude sensor can measure the inclination angles of the body of the robotic dog on the x, y, and z axes, which are α x , α y , α z; Taking the attitude sensor as the origin, measure the distances of the attitude sensor from the highest point in the vertical direction of the foot end on the body of the robot dog on the x and y axes, denoted as n and m respectively; calculate the heights h of the four points as h = h0 + n * sin∝ x -m * sin∝ x cos∝ y , where h0 is the theoretical height without deviation angle; zero-adjust the robot dog based on the naked eye or a zero-position alignment tool, and control the error within 1 degree.

[0009] Further, finding the high point of the shoulder joint by rotating the shoulder joint of the robot dog forward and backward includes the following steps: continuously fine-tune the shoulder joint angle forward and backward to find the high point direction. When the captured value h of the attitude sensor tends to increase, select this direction for fine adjustment. When the captured value h of the attitude sensor tends to decrease, fine-tune in the reverse direction and reduce the fine-tuning angle to improve the fine-tuning accuracy. Stop and record the position at this time and set it as the 0 point of the shoulder joint when the fine-tuning angle is about 0.01 degrees.

[0010] Further, controlling the knee joint and hip joint to move at different angles to obtain different shoulder joint heights H includes the following steps: calculate the shoulder joint height H = L2 * cosθ2 + L3 * cos(θ3 - θ2) during the movement of the knee joint and hip joint angles, where L2 is the length of the rod connecting the knee joint and the hip joint, L3 is the length of the rod connecting the hip joint and the foot end, θ2 is the angle between the rod connecting the knee joint and the hip joint and the vertical direction, and θ3 is the rod connecting the hip joint and the foot end and the rod connecting the knee joint and the hip joint.

[0011] Further, when calculating the shoulder joint height H during the movement of the knee joint and hip joint angles, define the initial angles θ2 and θ3 according to the difference of 0.1 degrees and the range of 1 degree, and define the initial rod lengths L2 and L3 according to the difference of 0.1 mm and the range of 1 mm. After n movements, record the differences between the H values and the h values measured by the sensor respectively; take the minimum average difference of H and h as the target to obtain the initial values of θ2, θ3, L2, and L3; then reduce the difference, and perform iteration again according to the difference of 0.01 degrees and the range of 0.1 degrees, and according to the difference of 0.01 mm and the range of 0.1 mm, to obtain the optimal values of θ2, θ3, L2, and L3.

[0012] Further, multiple legs are moved simultaneously during the process of controlling the knee joint and the frontal hip joint to move at different angles.

[0013] Furthermore, while controlling the knee and hip joints to move at different angles, two adjacent legs are calibrated simultaneously, with each leg performing two calculations. Because the robot dog maintains stability on three legs in certain situations, it's possible that one foot might not touch the ground, leading to incorrect calculations. Therefore, calibrating two adjacent legs simultaneously improves both accuracy and speed.

[0014] Furthermore, the method further includes the following steps: starting the robot dog's motion mode, fine-tuning the gait compensation parameters and the reduction ratio, and reducing the fluctuation amplitude of the robot dog's motion.

[0015] Furthermore, starting the robot dog's motion mode, fine-tuning the gait compensation parameters and the reduction ratio, and reducing the fluctuation amplitude of the robot dog's motion include the following steps: starting the robot dog's stationary stepping mode, wherein the height of the robot dog's body fluctuates periodically, and the deflection angle detected by the attitude sensor shows a harmonic change; by fine-tuning the gait compensation parameters and the reduction ratio, the fluctuation amplitude of the deflection angle detected by the attitude sensor is reduced.

[0016] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the robot dog calibration method of the present invention;

[0018] Figure 2 A structural diagram of the robot dog body and posture sensor of the present invention;

[0019] Figure 3 This is a front view of the robot dog of the present invention;

[0020] Figure 4 It is a side view of the robot dog of the present invention. DETAILED DESCRIPTION

[0021] Referring to the calibration methods for industrial robots, the calibration parameters for connecting rod transmissions are generally zero position, rod length, and speed ratio. Therefore, this method focuses on calibrating these three parameters. However, unlike the laser calibration method for industrial robots, the robot dog has four foot ends 15, so simple laser capture cannot be used for position calibration. Therefore, it is necessary to use a posture sensor 2 or other measurement method capable of measuring the height of the robot dog's limbs to measure the height of the robot dog's limbs.

[0022] See also Figures 1-3 The robot dog calibration method of the present invention comprises the following steps:

[0023] S10: Measure the height h of the robot dog at the limbs and perform zero adjustment.

[0024] In this embodiment, an attitude sensor 2 is provided in the middle of the top of the body 11 of the robot dog. The attitude sensor 2 can measure the inclination angles of the body 11 of the robot dog on the x, y, and z axes, which are α x , α y , α z .

[0025] Taking the attitude sensor 2 as the origin, the distances of the attitude sensor 2 from the highest points in the vertical directions of the upper foot ends 15 on the body 11 of the robot dog are measured on the x and y axes, denoted as n and m respectively, and the heights of four points can be calculated respectively

[0026] h = h0 + n * sin∝ x -m * sin∝ x cos∝ y

[0027] where h0 is the theoretical height without deviation angle.

[0028] Subsequently, manual zero adjustment is performed to zero the heights of the positions of the four limbs of the robot dog based on the naked eye or a zero position alignment tool, and the error is controlled to be less than 1 degree.

[0029] S20: Rotate the shoulder joint 12 of the robot dog forward and backward to find the high point of the shoulder joint 12.

[0030] Due to the special position of the shoulder joint 12, it can be found that the zero position of the shoulder joint 12 is theoretically the point that controls the height of the fuselage. Theoretically, no matter which direction it rotates, its height will decrease. Therefore, according to this logic, zero position capture is performed, that is, continuously fine-tune the angle of the shoulder joint 12 forward and backward to find the high point direction. When the sensor capture value h tends to increase, select this direction for fine adjustment. When the sensor capture value h tends to decrease, fine-tune in the reverse direction and reduce the fine-tuning angle, that is, improve the fine-tuning accuracy. When the fine-tuning angle is about 0.01 degrees, stop and record the position at this time and set it as the 0 point of the shoulder joint 12.

[0031] S30: Control the knee joint 13 and the hip joint 14 to move at different angles respectively to obtain different heights H of the shoulder joint 12. Taking the minimum average value of the difference between H and h as the goal, find the optimal angles of the knee joint 13 and the hip joint 14 and the lengths of the rod connecting the knee joint 13 and the hip joint 14 and the rod connecting the hip joint 14 and the foot end 15, and perform zero position adjustment.

[0032] After the shoulder joint 12 is calibrated, the first zeroing self-calibration is performed, that is, controlling the joint angles to perform attitude adjustment. Its adjustment logic is: Set the shoulder joint 12 to the zero position. Referring to the double-link coordinate geometry algorithm, control the knee joint 13 and the hip joint 14 to move at different angles respectively. At this time, the height H of the shoulder joint 12 will change, and n such H values are obtained

[0033] H = L2 * cosθ2 + L3 * cos(θ3 - θ2)

[0034] Wherein, L2 is the length of the rod connecting the knee joint 13 and the hip joint 14, L3 is the length of the rod connecting the hip joint 14 and the foot end 15, θ2 is the angle between the rod connecting the knee joint 13 and the hip joint 14 and the vertical direction, and θ3 is the angle between the rod connecting the hip joint 14 and the foot end 15 and the extension line of the rod connecting the knee joint 13 and the hip joint 14.

[0035] Adopt the enumeration method based on the nested algorithm. Define the initial angles θ2 and θ3 according to the difference of 0.1 degrees and the range of 1 degree, and define the initial rod lengths L2 and L3 according to the difference of 0.1 mm and the range of 1 mm. After n movements, record the difference between the H value and the h value measured by the sensor respectively. For example, if the initial angle of θ2 is 45 degrees, then within the range of 44 - 46, take a value every 0.1 degree and substitute it as the initial value of θ2 respectively. Take the minimum average value of the difference between H and h as the goal to obtain the initial values of θ2, θ3, L2, and L3. Then it is considered that the values of θ2, θ3, L2, and L3 at this time are closer to the actual values. Then reduce the difference, and perform iteration again according to the difference of 0.01 degrees and the range of 0.1 degree, and according to the difference of 0.01 mm and the range of 0.1 mm to obtain the optimal values of θ2, θ3, L2, and L3.

[0036] To achieve rapid calibration, the process of controlling the knee joint 13 and the hip joint 14 to move at different angles is carried out simultaneously for multiple legs. Since the support of three legs has stability in some cases, it is possible that a certain foot end 15 does not touch the ground, resulting in incorrect calculation. Therefore, control two adjacent legs to perform calibration simultaneously in turn, that is, the two left legs perform simultaneously, then the two right legs perform simultaneously, then the two front legs perform simultaneously, and then the two rear legs perform simultaneously. Each leg will perform two calculations.

[0037] After the above calibration is completed, use the calibrated θ2, θ3, L2, and L3 for zero position adjustment, that is, make H equal to h0, and control the error within 0.02. At this time, the first zeroing self - calibration is completed.

[0038] S40: Start the motion mode of the robotic dog, finely adjust the gait compensation parameters and the reduction ratio, and reduce the fluctuation amplitude of the robotic dog's movement.

[0039] Finally, perform dynamic tuning calibration. After the above calibration is completed, the robotic dog should be in the horizontal angle of the attitude sensor 2, that is, the inclination angles of its x, y, and z axes are α x , α y , α zAll are 0. At this time, dynamic calibration is performed, and the machine dog's in-place stepping mode is started. At this time, the height of the fuselage will show periodic fluctuations. When it is fed back to the sensor, the deflection angle shows a harmonic change. Dynamic tuning calibration is to reduce the fluctuation amplitude and ensure the stability of the machine dog by finely adjusting the gait compensation parameters and reduction ratio. The basic algorithm operation logic is to take the fluctuation amplitude removal as the target value, control the calibration parameters to change, record the parameters when the fluctuation amplitude decreases, and finally iterate to obtain the optimal value.

[0040] The machine dog calibration method of the present invention can perform calibration in real time and quickly, improve the motion accuracy of the machine dog, is convenient for calibration, and basically does not require tools; has high calibration accuracy, calibrates length, angle, speed ratio, etc.; can effectively improve the stability of the machine.

[0041] The above embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it cannot be understood as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and the present invention also intends to include these changes and modifications.

Claims

1. A method for calibrating a robotic dog, characterized in that: It includes the following steps: Measure the height h of the quadruped robot at the positions of its four limbs and zero it. Rotate the shoulder joint of the quadruped robot forward and backward to find the highest point of the shoulder joint. Control the knee joint and the hip joint to move at different angles respectively, obtain different shoulder joint heights H, take the minimum average difference between H and h as the goal, calculate the optimal angles of the knee joint and the hip joint and the lengths of the rod connecting the knee joint and the hip joint and the rod connecting the hip joint and the foot end, and perform zero position adjustment.

2. The method for calibrating a robotic dog according to claim 1, wherein: Measuring the height h of the quadruped robot at the positions of its four limbs and zeroing it includes the following steps: A posture sensor is arranged in the middle of the top of the body of the robotic dog, and the posture sensor can measure the inclination angles of the body of the robotic dog on the x, y, and z axes, which are α x , α y , α z ; Taking the attitude sensor as the origin, measure the distances of the attitude sensor from the highest point in the vertical direction of the foot end on the body of the quadruped robot on the x and y axes, denoted as n and m respectively. Calculate the heights of the four points respectively: h = h0 + n * sin∝ x - m * sin∝ x cos∝ y , where h0 is the theoretical height without declination angle; Zero the quadruped robot based on the naked eye or a zero position alignment tool, and control the error within 1 degree.

3. The method for calibrating a robotic dog according to claim 2, wherein: Rotating the shoulder joint of the quadruped robot forward and backward to find the highest point of the shoulder joint includes the following steps: Continuously fine-tune the shoulder joint angle forward and backward to find the direction of the highest point. When the captured value h of the attitude sensor tends to increase, select this direction for fine adjustment. When the captured value h of the attitude sensor tends to decrease, fine-tune in the reverse direction and reduce the fine-tuning angle to improve the fine-tuning accuracy. Stop and record the position at this time and set it as the zero point of the shoulder joint when the fine-tuning angle is about 0.01 degrees.

4. The method for calibrating a robotic dog according to claim 3, wherein: Controlling the knee joint and the hip joint to move at different angles respectively to obtain different shoulder joint heights H includes the following steps: Calculate the shoulder joint height H = L2 * cosθ2 + L3 * cos(θ3 - θ2) during the movement of the knee joint and the hip joint angles, where L2 is the length of the rod connecting the knee joint and the hip joint, L3 is the length of the rod connecting the hip joint and the foot end, θ2 is the angle between the rod connecting the knee joint and the hip joint and the vertical direction, and θ3 is the angle between the rod connecting the hip joint and the foot end and the extension line of the rod connecting the knee joint and the hip joint.

5. The method for calibrating a robotic dog according to claim 4, wherein: When calculating the shoulder joint height H during the movement of the knee joint and the hip joint angles, define the initial angles θ2 and θ3 according to the difference of 0.1 degrees and the range of 1 degree, and define the initial rod lengths L2 and L3 according to the difference of 0.1 mm and the range of 1 mm. After n movements, record the differences between the H values and the h values measured by the sensor respectively; take the minimum average difference between H and h as the goal to calculate the initial values of θ2, θ3, L2, and L3; then reduce the difference, and perform iteration again according to the difference of 0.01 degrees and the range of 0.1 degrees, and according to the difference of 0.01 mm and the range of 0.1 mm, to obtain the optimal values of θ2, θ3, L2, and L3.

6. The method for calibrating a robotic dog according to claim 5, wherein: During the process of controlling the knee joint and the frontal hip joint to move at different angles, multiple legs are carried out simultaneously.

7. The method for calibrating a robotic dog according to claim 6, characterized in that: During the process of controlling the knee joint and the frontal hip joint to move at different angles, two adjacent legs are calibrated simultaneously in sequence, and each leg is calculated twice.

8. The method for calibrating a robotic dog according to claim 7, wherein: It also includes the following steps: Start the motion mode of the quadruped robot, fine-tune the gait compensation parameters and the reduction ratio, and reduce the fluctuation amplitude of the quadruped robot's movement.

9. The method for calibrating a robotic dog according to claim 8, wherein: Starting the motion mode of the quadruped robot, fine-tuning the gait compensation parameters and the reduction ratio, and reducing the fluctuation amplitude of the quadruped robot's movement includes the following steps: Start the in-place walking mode of the robotic dog. The body height of the robotic dog fluctuates periodically, and the detected deflection angle of the attitude sensor shows a harmonic change. By finely adjusting the gait compensation parameters and the reduction ratio, the fluctuation amplitude of the deflection angle detected by the attitude sensor is reduced.

Images