Method and device for driving motion of a quadruped robot based on traction means

Abstract

This invention discloses a motion driving method and apparatus for a quadruped robot based on a traction device, relating to the field of legged robot technology. The method includes: acquiring traction force data generated by the traction device based on sensors, wherein the sensors include a three-dimensional force sensor and an angle encoder; calculating the target forward speed and target turning angular velocity of the quadruped robot based on the traction force data; and converting the target forward speed and target turning angular velocity into robot joint control commands through a force-position hybrid control strategy to drive the robot to follow the motion. This invention, through motion control based on the rate of change of force and a force-position hybrid response, enables the robot to intuitively, smoothly, and safely follow a load in complex and obstructed environments such as mountainous areas, improving the efficiency, reliability, and naturalness of human-robot collaborative operations, and providing an effective solution for the practical application of quadruped robots in complex scenarios.

Description

Technical Field

[0001] This invention relates to the field of legged robot technology, and in particular to a method and apparatus for driving the motion of a quadruped robot based on a traction device. Background Technology

[0002] When carrying out material transportation tasks in complex, unstructured terrains such as mountains and hills, the environment typically has the following characteristics: rugged and uneven paths, numerous visual obstructions, and variable and potentially severe lighting conditions. In such scenarios, traditional transportation methods face significant challenges: manual handling is inefficient and highly dangerous; animal-powered transport has poor controllability and is limited by terrain; and wheeled or tracked vehicles often cannot reach the mission point due to insufficient mobility.

[0003] With the development of robotics technology, quadruped robots, with their excellent terrain adaptability, are considered an ideal platform for solving the problem of transporting heavy objects in mountainous areas. Currently, the main technical approaches to achieving robot following include: the robot identifies and tracks the operator ahead using sensors such as LiDAR and visual cameras. However, in mountainous environments, the operator and robot are often obstructed by undulating terrain, trees, or rocks, leading to the loss of the tracking target. In addition, conditions such as fog, rain, snow, nighttime, or dappled sunlight under trees can severely interfere with the normal operation of visual sensors, making such solutions lack environmental robustness and difficult to guarantee reliability. Summary of the Invention

[0004] The main objective of this invention is to provide a method and apparatus for driving the motion of a quadruped robot based on a traction device, which aims to solve the technical problems of insufficient reliability of existing driving methods in complex environments with limited vision, and the lack of perception of the operator's dynamic intentions in existing physical traction schemes, resulting in stiff robot response, unnatural interaction, and low collaborative efficiency.

[0005] To achieve the above objectives, this invention proposes a motion driving method for a quadruped robot based on a traction device. The quadruped robot is equipped with a traction device for detecting traction force data. The method includes the following steps: The traction force data generated by the traction device is acquired based on sensors, including a three-dimensional force sensor and an angle encoder. Calculate the target forward speed and target turning angular velocity of the quadruped robot based on the traction force data; The force-position hybrid control strategy converts the target forward speed and target turning angular velocity into robot joint control commands, driving the robot to follow the motion.

[0006] Furthermore, the traction force data includes the magnitude of the traction force, the traction force direction angle, and the rate of change of the magnitude of the traction force.

[0007] Further, the calculation of the target forward speed and the target steering angular velocity based on the traction force data includes: the calculation method of the target forward speed is: Vt(k)=Vt(k - 1)+α×(Vtarget(k) Vt(k 1)), Vtarget(k)=Kv×F(k); a dead zone threshold Vdeadband is set. When ∣Vtarget(k)∣<Vdeadband, it is determined as a tremor interference, and Vtarget(k)=0 is set; where k = 0, 1, 2……R, R is the number of time periods, k is the time period serial number, Vt(k) is the target forward speed of the k-th time period, Vt(k 1) is the target forward speed of the (k - 1)-th time period, Kv is the speed interpolation coefficient, Vt(0)=0, F(k) is the magnitude of the traction force of the k-th time period,; ωt(k) is the target steering angular velocity of the k-th time period, Kω is the steering angular velocity interpolation coefficient, and θ(k) is the traction force direction angle of the k-th time period.

[0008] Further, the force-position hybrid control strategy includes: dynamically adjusting the force feedback weight and the position feedback weight according to the change rate of the magnitude of the traction force; when the absolute value of the change rate is greater than the preset sensitivity threshold, increasing the force feedback weight; when the absolute value of the change rate is less than or equal to the threshold, increasing the position feedback weight.

[0009] Further, it also includes an abnormal protection step: when it is detected that the magnitude of the traction force exceeds the safety threshold, or the change rate of the traction force direction angle exceeds the limit steering rate, controlling the quadruped robot to execute a safety strategy of decelerating to a stop.

[0010] Further, the traction force magnitude threshold is an adaptively adjustable threshold, and its adjustment method is: dynamically correcting the traction force magnitude threshold according to the current load weight of the quadruped robot and the friction coefficient of the movement road surface through a preset algorithm.

[0011] Further, the steering gain coefficient Kω is dynamically adjusted according to the absolute value of the traction force direction angle θ(k), where the value of Kω is inversely correlated with the magnitude of |θ(k)|.

[0012] The present invention also proposes a traction device for a quadruped robot, including a traction module, a fixing module, and a sensor module The traction module is for an operator to hold; The fixing module is connected to the traction module through a movable mechanism and is installed on the robot body The sensor module is integrated in the fixing module and is used to detect the traction force and the direction angle; The active mechanism is a soft chain mechanism that provides a range of motion of not less than 180°.

[0013] Furthermore, the gripping part of the traction module is equipped with a gripping status detection unit; the device also includes a protective shell.

[0014] The present invention also proposes a quadruped robot, wherein the traction device is disposed on the robot body; The main control unit, located within the robot body, is used to receive data from the traction device and execute the quadruped robot motion drive method based on the traction device to generate motion commands. The joint actuators are connected to the main control unit and the joints of the robot body, and are used to drive the robot to move according to motion commands.

[0015] This invention enables robots to follow loads intuitively, smoothly, and safely in complex, occluded environments such as mountainous areas through motion control based on the rate of change of force and a hybrid force-position response. This improves the efficiency, reliability, and naturalness of human-robot collaboration, and provides an effective solution for the practical application of quadruped robots in complex scenarios. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form a module of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0017] 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, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the quadruped robot motion driving method based on a traction device according to the present invention. Figure 2 This is a schematic diagram of the module structure of a traction device for a quadruped robot according to an embodiment of the present invention; The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of the present invention and are not intended to limit the present invention.

[0020] To better understand the technical solution of the present invention, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0021] refer to Figure 1, Figure 1 It is a schematic flow chart provided for the embodiment of the motion driving method of a quadruped robot based on a traction device in the present invention; As Figure 1 shown, the present invention proposes a motion driving method for a quadruped robot based on a traction device. The quadruped robot is provided with a traction device for detecting traction force data. The method includes the following steps: S10. Obtain the traction force data generated by the traction device based on the sensor.

[0022] In this embodiment, the sensor data of the traction device is read at a frequency of 500 Hz (control period ΔT = 0.002 s). The magnitude of the traction force F(k) in the current period k is calculated from the three-dimensional force sensor data, and the current traction force direction angle θ(k) is read from the angle encoder. The front of the robot is defined as 0°, the left side is positive, and the right side is negative.

[0023] S20. Calculate the target forward speed and target steering angular velocity of the quadruped robot according to the traction force data.

[0024] The calculation of the target forward speed and target steering angular velocity according to the traction force data includes: The calculation method of the target forward speed is: Vt(k)=Vt(k - 1)+α×(Vtarget(k) Vt(k - 1)), Vtarget(k)=Kv×F(k); Set the dead zone threshold Vdeadband. When ∣Vtarget(k)∣<Vdeadband, it is determined as hand tremor interference, and let Vtarget(k)=0; where k = 0, 1, 2... R, R is the number of time periods, k is the time period serial number, Vt(k) is the target forward speed in the k-th time period, Vt(k - 1) is the target forward speed in the k - 1-th time period, Kv is the speed interpolation coefficient, Vt(0)=0, F(k) is the magnitude of the traction force in the k-th time period,; ωt(k) is the target steering angular velocity in the k-th time period, Kω is the steering angular velocity interpolation coefficient, and θ(k) is the traction force direction angle in the k-th time period.

[0025] Through the filtering superposition algorithm and dead zone processing, the high-frequency fluctuations caused by hand tremors or minor disturbances are effectively filtered, avoiding sudden changes in speed and steering angular velocity, making the device move more smoothly, and improving the operation comfort. And both the target forward speed and the steering angular velocity adopt the filtering superposition form, with a unified algorithm structure, greatly reducing the redundancy of code implementation, facilitating development, debugging, and subsequent maintenance.

[0026] The algorithm described above ensures that the rate of change of the robot's speed is proportional to the rate of change of the force applied by the operator. When the operator gently increases the pulling force, Vt(k) increases gradually, and the robot accelerates smoothly; when the operator quickly releases the force, Fs(k) becomes negative and has a large absolute value, and Vt(k) decreases rapidly, causing the robot to decelerate gently. This fundamentally avoids the abrupt start and stop phenomena caused by the sudden jump to a fixed speed upon reaching a threshold, as seen in traditional methods.

[0027] In another embodiment, the determination of key control parameters such as the speed interpolation coefficient Kv and the steering gain coefficient Kω is based on systematic experiments and data fitting conducted on actual robot platforms and in typical work scenarios.

[0028] The specific implementation steps are as follows: Construct an experimental system that includes the quadruped robot, the traction device, and the test environment. The operator performs a series of standard traction actions on representative terrains (such as flat ground, slopes, and gravel roads), including acceleration, deceleration, constant speed walking, and turning with different curvatures.

[0029] Secondly, during the experiment, multiple sets of time-series data were recorded synchronously and with high precision, including: the magnitude F(t) and direction θ(t) of the pulling force applied by the operator, the robot's actual motion speed Va(t) and turning angular velocity ωa(t), and the robot's state information, including body posture and joint torque. Using the aforementioned algorithm model (e.g., Vt(k) = Vt(k-1) + Kv × Fs(k) × ΔT and ωt(k) = Kω × θ(k)) as the parameterized structure, the recorded Fs(t) and θ(t) were used as inputs, and the recorded Va(t) and ωa(t) were used as the desired outputs. The least squares method or optimization algorithm was used to fit the parameter set {Kv, Kω, ...} to minimize the overall error between the model output and the robot's actual motion data. The optimal parameter set obtained from the fitting was then fixed and applied to the control algorithm of the main control unit.

[0030] S30: The target forward speed and target turning angular velocity are converted into robot joint control commands through a force-position hybrid control strategy, driving the robot to achieve compliant following motion.

[0031] In this embodiment, the quadruped robot is equipped with a main control unit and a low-level whole-body dynamics controller.

[0032] Furthermore, the main control unit inputs Vt(k) and ωt(k) as high-level motion commands to the low-level whole-body dynamics controller of the quadruped robot. The low-level whole-body dynamics controller adopts a force-position hybrid control strategy based on model predictive control, and it is equipped with a force control loop and a position control loop. The force control loop is used to track the desired foot contact force, and the position control loop is used to track the desired body motion trajectory.

[0033] Furthermore, in the dynamic equations of the whole-body dynamics controller, the currently measured traction force F(k) is taken as an explicit external force input, and when calculating the required support force at each foot, the influence of this external force on the fuselage dynamics is actively compensated.

[0034] Furthermore, the control strategy of this invention differs from simple position tracking in that the robot's motion response is achieved through active dynamic adaptation to traction forces in force-position hybrid control, enabling the robot system to exhibit adaptive physical interaction characteristics. When the robot is subjected to sudden, unexpected lateral traction forces, the integrated force control mechanism can be rapidly activated, promptly adjusting the robot's center of gravity and foot force distribution, thereby effectively suppressing external disturbances and ensuring the overall posture stability of the robot during complex traction interactions.

[0035] Specifically, in this embodiment, the following control strategy is implemented in the underlying whole-body dynamics controller: the force feedback weight and position feedback weight are parameterized and defined, the force feedback weight coefficient is set as α, the position feedback weight coefficient is set as β, and the sum of α and β is constrained to be 1. The force feedback weight coefficient α is a function of the rate of change of the tensile force |Fs(k)|, and is specifically determined by the following mapping relationship: α(k)=f(|Fs(k)|), where f is a preset monotonically non-decreasing function, and its output value is restricted to the interval [αmin, αmax]. αmin and αmax are the preset minimum force feedback weight coefficient and maximum force feedback weight coefficient, respectively, and the corresponding position feedback weight coefficient is β(k)=1-α(k).

[0036] Furthermore, within each solution cycle of the force-position hybrid control strategy, the controller performs two calculations in parallel: first, based on the force feedback loop, it calculates a first control output Uf aimed at tracking the desired foot force including the compensation term of the traction force F(k); second, based on the position feedback loop, it calculates a second control output Up aimed at tracking the desired fuselage trajectory derived from the target forward velocity Vt(k) and the target turning angular velocity ωt(k).

[0037] Finally, the controller performs a weighted fusion of the two outputs using the dynamically adjusted weighting coefficients to generate a comprehensive control command Ucmd(k) acting on the robot's joint actuators. The calculation formula is as follows: Ucmd(k)=α(k)×Uf(k)+β(k)×Up(k); The dynamic adjustment process and effect of this strategy are as follows: When the rate of change of the pulling force |Fs(k)| increases, the force feedback weight coefficient α(k) increases accordingly, causing the comprehensive control command Ucmd(k) to focus more on the first control output Uf(k). This allows the control system to prioritize a rapid torque response and dynamic compliance to the operator's force application intention, thereby achieving agile following. When |Fs(k)| decreases, α(k) decreases while the position feedback weight coefficient β(k) increases. The comprehensive control command Ucmd(k) focuses more on the second control output Up(k), allowing the control system to prioritize accurate and smooth tracking of the desired motion trajectory, effectively filtering out interference, thereby achieving stable and efficient compliant motion. The dynamic weight adjustment mechanism, in conjunction with the strategy of using the pulling force as an external force for feedforward compensation, jointly ensures the adaptive optimization of the quadruped robot between rapid response and smooth motion.

[0038] In this embodiment, a quadruped robot motion driving method based on a traction device also includes a safety and anomaly handling mechanism, which is achieved by adaptive threshold adjustment and multi-dimensional safety monitoring.

[0039] Specifically, the adaptive threshold adjustment is used to dynamically set the traction force threshold Fthreshold for determining a valid traction intention, and corrects the value of Fthreshold based on real-time estimation of the quadruped robot system's state. Specifically, it estimates the robot's current load weight based on joint current information and estimates the ground friction coefficient through foot-to-ground slippage analysis. Based on these real-time parameters, Fthreshold is dynamically increased or decreased using a preset mapping algorithm. For example, under conditions of increased load or decreased road surface friction coefficient, the unit automatically increases Fthreshold. The technical effect of this dynamic adjustment process is that it effectively filters out minor force sensor noise fluctuations caused by changes in the robot's own inertia or foot slippage, thereby avoiding misidentification of these interferences as valid traction commands from the operator, significantly improving the robustness and reliability of the system's intention recognition under different loads and terrains.

[0040] Furthermore, multi-dimensional security monitoring is used to monitor the security of the traction interaction process in real time, and its monitoring objects include: The magnitude of the pulling force F(k); The rate of change of the traction force direction angle, where the rate of change of the traction force direction angle is determined by [θ(k)]. θ(k 1)] / ΔT is calculated.

[0041] In this embodiment, multi-dimensional safety monitoring presets a safety threshold Fmax and a limit turning rate ωmax for the two monitored objects. When F(k) > Fmax, it is determined that the robot is stuck or encounters a dangerous pull. When the rate of change of the direction angle exceeds ωmax, it is determined that the operator suddenly pulls hard or the device malfunctions, and is therefore classified as an abnormal state.

[0042] Among them, the limit turning rate ωmax is a pre-set safety threshold parameter. Specifically, the setting process of this parameter includes the following steps: Based on the multibody dynamics model of the quadruped robot, combined with its current motion posture, the contact constraints between the foot and the ground, and the torque-velocity characteristics of each joint actuator, calculation and analysis are performed to obtain the maximum theoretical angular velocity value that ensures the robot remains stable during dynamic turning; Based on the effective range and measurement accuracy of the angle sensor inside the traction device, as well as the impact resistance and fatigue characteristics of mechanical components such as the soft chain mechanism, the maximum angular velocity range that the device can withstand under long-term reliable working conditions is calibrated through experimental testing; By conducting multiple sets of human-machine interactive traction experiments, angular velocity data of the operator performing turning actions in different work scenarios are collected.

[0043] Furthermore, by comprehensively comparing the aforementioned upper limit of angular velocity based on theoretical stability, the limit of angular velocity based on device tolerance, and the empirical threshold of angular velocity based on interaction safety, the minimum value is selected. A safety factor is then introduced based on this minimum value to obtain the limiting turning rate ωmax. This comprehensive method ensures that the setting of ωmax conforms to the motion stability constraints of the robot body while remaining within the hardware safety tolerance of the traction device. Simultaneously, it can identify and suppress dangerous or unintentional sharp turns from the human-machine interaction perspective, thereby establishing a reliable safety protection mechanism for the traction following system and significantly reducing the risk of robot instability, device damage, or human-machine collisions caused by excessively high turning angular velocities.

[0044] In this embodiment, when the multi-dimensional safety monitoring unit triggers an abnormal state, the main control unit will immediately execute a safety strategy: overriding the normal motion control commands, generating a target forward speed command curve that smoothly decreases from the current value to zero; at the same time, controlling all joint actuators to switch to soft stop mode, gradually reducing the output torque of each joint according to the preset torque reduction curve, so that the robot can achieve smooth speed decay and controlled torque reduction in abnormal situations such as tension overload or excessive steering, achieving a smooth and safe stop, effectively avoiding the risk of instability caused by body overturning and joint impact due to emergency braking.

[0045] refer to Figure 2 , Figure 2 This is a schematic diagram of the module structure of a traction device for a quadruped robot according to an embodiment of the present invention.

[0046] like Figure 2 As shown, the present invention also proposes a traction device for a quadruped robot, including a traction module 10, a fixing module 20, and a sensor module 30.

[0047] Towing module 10, for operator gripping; The fixed module 20 is connected to the traction module 10 via a movable mechanism and is installed on the robot body; Sensor module 30, integrated into the fixing module 20, is used to detect tensile force and direction angle; The active mechanism is a soft chain mechanism with an active range of not less than 180°.

[0048] Furthermore, the gripping part of the traction module is equipped with a gripping status detection unit; the device also includes a protective shell.

[0049] The present invention also proposes a quadruped robot, including a traction device, a main control unit, and joint actuators; The traction device is mounted on the robot body; The main control unit, located within the robot body, is used to receive data from the traction device and execute the quadruped robot motion drive method based on the traction device to generate motion commands. The joint actuators are connected to the main control unit and the joints of the robot body, and are used to drive the robot to move according to motion commands.

[0050] Preferably, the core control method of the present invention runs on the main control unit. The main control unit adopts a high-performance embedded SoC system-on-a-chip, which integrates a general-purpose processor (CPU), an AI acceleration unit (BPU), and a microcontroller (MCU). The CPU is responsible for upper-layer intent parsing and instruction generation, the BPU can be used for parallel computation of complex force-position hybrid control models, and the MCU ensures the real-time and deterministic output of joint control commands.

[0051] This invention enables robots to follow loads intuitively, smoothly, and safely in complex, occluded environments such as mountainous areas through motion control based on the rate of change of force and a hybrid force-position response. This improves the efficiency, reliability, and naturalness of human-robot collaboration, and provides an effective solution for the practical application of quadruped robots in complex scenarios.

[0052] The above are only some embodiments of the present invention and do not limit the patent scope of the present invention. All equivalent structural transformations made under the technical concept of the present invention using the contents of the present invention specification and drawings, 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 driving the motion of a quadruped robot based on a traction device, characterized in that, The quadruped robot is equipped with a traction device for detecting traction force data; the method includes the following steps: The traction force data generated by the traction device is acquired based on sensors, including a three-dimensional force sensor and an angle encoder. The target forward speed and target turning angular velocity of the quadruped robot are calculated based on the traction force data. The target forward speed and target turning angular velocity are converted into robot joint control commands through a force-position hybrid control strategy, driving the robot to follow the movement.

2. The method as described in claim 1, characterized in that, The traction force data includes the magnitude of the traction force, the traction force direction angle, and the rate of change of the magnitude of the traction force.

3. The method as described in claim 2, characterized in that, The calculation of the target forward speed and the target steering angular velocity based on the traction force data includes: The calculation method of the target forward speed is: Vt(k)=Vt(k - 1)+α×(Vtarget(k) Vt(k 1)), Vtarget(k)=Kv×F(k); Set a dead zone threshold Vdeadband. When ∣Vtarget(k)∣<Vdeadband, it is determined as a hand tremor interference, and Vtarget(k)=0 is set; The target turning angular velocity is calculated as follows: ωt(k) = Kω × θ(k); Where k = 0, 1, 2, ..., R, R is the number of time periods, k is the time period number, and Vt(k) is the target's forward velocity in the k-th time period. 1) is the target forward speed in the (k-1)th time period, Kv is the speed interpolation coefficient, Vt(0)=0, F(k) is the magnitude of the traction force in the kth time period, ωt(k) is the target turning angular velocity in the kth time period, Kω is the turning angular velocity interpolation coefficient, and θ(k) is the traction force direction angle in the kth time period.

4. The method as described in claim 2 or 3, characterized in that, The force-position hybrid control strategy includes: The force feedback weight and position feedback weight are dynamically adjusted according to the rate of change of the pulling force; when the absolute value of the rate of change is greater than a preset sensitivity threshold, the force feedback weight is increased; when the absolute value of the rate of change is less than or equal to the threshold, the position feedback weight is increased.

5. The method as described in claim 2, characterized in that, Also includes: When the magnitude of the pulling force exceeds the safety threshold, or the rate of change of the traction direction angle exceeds the limit turning rate, the quadruped robot is controlled to decelerate to a stop.

6. The method as described in claim 3, characterized in that, The threshold value of the pulling force is an adaptively adjustable threshold value. The adjustment method is as follows: update the threshold value of the pulling force based on the current load weight of the quadruped robot and the friction coefficient of the moving surface.

7. The method as described in claim 3, characterized in that, The steering gain coefficient Kω is dynamically adjusted based on the absolute value of the traction direction angle θ(k), wherein the value of Kω is inversely correlated with the magnitude of |θ(k)|.

8. A traction device for a quadruped robot, characterized in that, The traction device is adapted to the method described in any one of claims 1-7, and includes a traction module, a fixing module, and a sensor module. The traction module is for the operator to hold; The fixed module is connected to the traction module via a movable mechanism and is mounted on the robot body. The sensor module, integrated into the fixing module, is used to detect the tensile force and direction angle; The active mechanism is a soft chain mechanism that provides a range of motion of not less than 180°.

9. The traction device as described in claim 8, characterized in that, The gripping part of the traction module is equipped with a gripping status detection unit; the device also includes a protective shell.

10. A quadruped robot, characterized in that, Includes the robot body and the traction device, main control unit, and joint actuator as described in claim 8; The traction device is mounted on the robot body; The main control unit is disposed within the robot body and is used to receive data from the traction device and execute the method of any one of claims 1-7 to generate motion commands; The joint actuator is connected to each joint of the main control unit and the robot body, and is used to drive the robot to move according to the motion command.

Images