A control method and system for a flexible joint of a humanoid robot

Abstract

The present application relates to the technical field of robot control, and particularly relates to a control method and system for a flexible joint of a humanoid robot. The method comprises the following steps: determining a residual torsion state corresponding to a flexible transmission assembly according to an angle difference between a driving end position and an output end position of the flexible joint in a stop state; in response to the flexible joint receiving a motion instruction, determining whether there is a pre-tightening offset affecting initial response before the joint is started by the residual torsion state; when there is a pre-tightening offset, controlling the driving end to perform an adjusting action opposite to a target motion direction, so that the flexible transmission assembly releases a residual torsion amount, and an initial control reference of the flexible joint is established. The present application detects the residual torsion state before the flexible joint is started and actively releases the pre-tightening offset, so that the control reference of the driving end and the output end is corrected, and the response speed, motion accuracy and stability of the flexible joint of the humanoid robot are improved.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and in particular to a control method and system for flexible joints of humanoid robots. Background Technology

[0002] Existing flexible joints typically use flexible transmission components to transmit torque between the drive and output ends. However, due to the elastic deformation characteristics of these components, residual torsion can easily occur during joint pausing or load changes, creating a preload offset between the drive and output ends. This leads to problems such as initial response lag, positional deviation, and motion shock when the joint restarts. Furthermore, existing control methods usually perform direct control based on the drive input, failing to fully consider the impact of the release of elastic energy stored within the flexible transmission components on the output response. This makes it difficult to achieve precise compensation during startup and dynamic adjustment during movement.

[0003] Therefore, how to identify and release the residual torsional state before the flexible joint is activated, establish an accurate initial control benchmark, and correct the driving torque according to the motion response has become an urgent problem to be solved in the current control process of flexible joints of humanoid robots. Summary of the Invention

[0004] Therefore, it is necessary to provide a control method and system for flexible joints of humanoid robots to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objective, a control method for a flexible joint of a humanoid robot is provided, the method comprising the following steps: Step S1: Determine the residual torsional state of the flexible transmission component based on the angle difference between the driving end position and the output end position of the flexible joint in the stopped state; Step S2: In response to the flexible joint receiving an action command, determine whether there is a preload offset that affects the initial response before the joint starts by checking the residual torsional state; Step S3: When there is preload offset, the control drive end performs an adjustment action opposite to the target motion direction, so that the flexible transmission component releases the residual torsion and establishes the initial control reference of the flexible joint. Step S4: Control the flexible joint to execute motion commands based on the initial control reference, and adjust the torque adjustment amount of the motion according to the output response during the operation of the motion commands.

[0006] This invention provides a control system for a flexible joint of a humanoid robot, used to execute the control method for a flexible joint of a humanoid robot as described above. The control system for a flexible joint of a humanoid robot includes: The confirmation module is used to determine the residual torsional state of the flexible transmission component based on the angle difference between the drive end position and the output end position of the flexible joint in the stopped state. The judgment module is used to determine whether there is a preload offset that affects the initial response before the joint starts by using the residual torsional state when the flexible joint receives a motion command. The adjustment module is used to control the drive end to perform an adjustment action opposite to the target motion direction when there is preload offset, so that the flexible transmission component releases the residual torsion and establishes the initial control reference of the flexible joint. The control module is used to control the flexible joint to execute motion commands based on the initial control reference, and to correct and adjust the torque adjustment amount of the motion according to the output response during the operation of the motion command.

[0007] The beneficial effects of this invention are that by detecting the residual torsional state between the drive end and the output end before the flexible joint is activated, and performing reverse micro-adjustment according to the pre-tightening offset direction, the elastic torsional amount accumulated inside the flexible transmission component is actively released, thereby eliminating the initial response deviation caused by the flexible transmission gap, elastic deformation, and long-term holding. At the same time, by adjusting the correlation between the slow-release driving force and the degree of elastic deformation release, the input torque at the drive end can be smoothly transmitted to the output end, reducing the impact and response lag at the moment of activation. During the execution of the action, the torque adjustment amount is dynamically corrected according to the output response, improving the position control accuracy, motion stability, and motion consistency of the flexible joint, enabling the humanoid robot joint to achieve smoother and more accurate motion control. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the steps of a control method for a flexible joint in a humanoid robot. Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S3. Figure 3 This is a diagram of a flexible joint architecture according to one embodiment; Figure 4 This is a schematic diagram of flexible joint control in one embodiment; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0009] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0010] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0011] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0012] To achieve the above objectives, please refer to Figures 1 to 4 A control method for flexible joints of a humanoid robot, the method comprising the following steps: Step S1: Determine the residual torsional state of the flexible transmission component based on the angle difference between the driving end position and the output end position of the flexible joint in the stopped state; Step S2: In response to the flexible joint receiving an action command, determine whether there is a preload offset that affects the initial response before the joint starts by checking the residual torsional state; Step S3: When there is preload offset, the control drive end performs an adjustment action opposite to the target motion direction, so that the flexible transmission component releases the residual torsion and establishes the initial control reference of the flexible joint. Step S4: Control the flexible joint to execute motion commands based on the initial control reference, and adjust the torque adjustment amount of the motion according to the output response during the operation of the motion commands.

[0013] In some embodiments, reference may be made to Figure 3 Taking a bipedal humanoid robot with a humanoid robotic arm as an example, this humanoid robot uses flexible joint structures at the shoulder, elbow, and wrist joints. These flexible joints include a drive motor, a harmonic reduction gear, a flexible transmission assembly, and an output joint shaft. During actions such as grasping and swinging, changes in external load and joint inertia cause torsional deformation in the flexible transmission assembly. Even after the robot stops moving, some elastic torsional energy may still be stored within the flexible transmission assembly, resulting in an angular deviation between the drive end and the output end.

[0014] Taking the humanoid robot's "arm-raising and object-grabbing" action as an example, when the robot arm is stationary, the control system first acquires the positions of the drive-end encoder and the output-end encoder of the shoulder joint's flexible joint. The drive-end encoder is mounted on the drive motor side to detect the motor shaft angle; the output-end encoder is mounted on the shoulder joint's output shaft side to detect the actual posture position of the robotic arm. By calculating the angle difference between the drive-end and output-end positions, the residual torsion of the flexible transmission component is obtained. For example, if a certain angular lag is detected between the drive-end and output-end, it indicates that the flexible transmission component still retains the torsional deformation caused by the previous action, and the current residual torsion state is determined based on the direction of the angle difference.

[0015] When the robot receives the "lift arm to grasp" action command, the control system determines whether there is a pre-tension offset before starting based on the target motion direction and the aforementioned residual torsional state. For example, if the arm was under a large load in the previous action, causing the flexible transmission component of the shoulder joint to produce elastic torsion in the downward direction, and the current action requires driving the robotic arm to lift upward, the residual torsional direction will counteract the initial driving force of the motor, preventing the output from immediately following the rotation of the drive end. Therefore, it is determined that there is a pre-tension offset in the flexible joint that affects the initial response.

[0016] It should be noted that, due to the differences in the motion joints of each robot, there is no specific limit to the threshold for judging pre-tightening offset. The judgment depends on the fact that when the absolute value of the angle difference is greater than a preset threshold (such as 0.05°) and the duration exceeds a preset time, pre-tightening offset is determined to exist.

[0017] When a preload offset is detected, the control drive motor first performs a small adjustment movement opposite to the lifting direction. For example, when the robotic arm needs to be lifted upwards, the control shoulder joint motor first generates a short-term reverse rotation, releasing the torsional energy stored inside the flexible transmission component by adjusting the torque in the opposite direction, thus gradually reducing the angle difference between the drive end and the output end. When the angle difference is detected to enter the set stable range, the control system records the current position of the shoulder joint drive end, the position of the output end, and the corresponding torque state, and uses this state as the initial control reference for subsequent arm lifting movements.

[0018] Subsequently, the robot performs a lifting and grasping motion based on this initial control baseline. During the robotic arm's movement, the control system continuously acquires the position feedback and motion response status of the shoulder joint output end, and determines the deviation between the actual robotic arm posture and the target motion trajectory. When the flexible transmission component undergoes elastic deformation again due to changes in the weight of the grasped object or motion inertia, causing a lag in the output response, the control system corrects the torque adjustment amount corresponding to the reverse adjustment action based on the output response deviation. For example, when an increase in the robotic arm load is detected, causing a slowdown in the output response, the subsequent adjustment torque compensation amount is increased; when an excessively fast output response is detected, the adjustment torque compensation amount is decreased, thereby ensuring that the shoulder joint flexible transmission component always adapts to the current load state.

[0019] In other embodiments, reference may also be made to Figure 4 Taking the flexible actuator of the shoulder joint of a humanoid robot as an example, the flexible joint includes a drive motor, a harmonic reduction gear, a flexible transmission gear, and an output joint shaft. The drive motor serves as the power input, and its output torque is reduced and amplified by the harmonic reduction gear before being transmitted to the flexible transmission gear. The flexible transmission gear then drives the output joint shaft to rotate, thereby adjusting the posture of the robotic arm linkage. Because the flexible transmission gear has a certain torsional stiffness, when the output joint shaft is subjected to external loads or inertia, the flexible transmission gear will undergo corresponding torsional deformation, creating an angle difference between the drive end angle and the output end angle.

[0020] Specifically, such as Figure 4 As shown in the dashed area on the left, the dashed box corresponds to the drive end structure inside the flexible joint. The drive motor is connected to a drive end encoder to detect the current rotation angle θm of the drive motor shaft. A harmonic reduction assembly is positioned between the drive motor and the flexible transmission assembly to reduce the rotational speed and increase the output torque. The flexible transmission assembly adopts a torsional elastic structure, with its two ends connected to the reduction output end and the output joint shaft, respectively, to generate a flexible transmission effect between the drive end and the load end. The output end encoder is located on the output joint shaft side to detect the actual rotation angle θo of the output joint shaft. By synchronously acquiring the drive end angle θm and the output end angle θo, the angle difference Δθ between the drive end and the output end can be calculated, where Δθ characterizes the current residual torsional state of the flexible transmission assembly.

[0021] When the robot is stationary, if there is no residual torsion within the flexible joint, the driving angle θm and the output angle θo should be approximately consistent. However, after the robot completes actions such as grasping or swinging, due to the load on the end effector, the flexible transmission component may store some elastic torsional energy, causing a deviation between the driving angle θm and the output angle θo. In this case, the torsional direction of the flexible transmission component is determined by calculating Δθ = θm - θo and considering the direction of the angle difference. For example, if the driving angle is greater than the output angle, it indicates that the flexible transmission component is in a state of torque accumulation, with unreleased elastic restoring force. If the driving angle is less than the output angle, it indicates that the flexible transmission component generates residual torsion in the opposite direction.

[0022] like Figure 4 As shown on the right, the output joint axis is connected to a mechanical link, which can be considered a load structure with length l and mass m, rotating around the output joint axis. The output joint axis corresponds to the output angle θo, and the motion command corresponds to the target motion angle θtarget. When the robot receives a new motion command, the control system determines the direction of the robot arm's movement based on the relationship between the current output angle θo and the target angle θtarget. Simultaneously, it combines the residual torsional state detected on the left to determine whether the current residual torque τs will affect the initial motion response.

[0023] For example, when the robotic arm was subjected to a downward load during its previous movement, causing the flexible transmission component to undergo torsional deformation in the corresponding direction, and the current movement requires the robotic arm to move upward, the residual torque τs will counteract the initial driving force generated by the drive motor, preventing the output joint axis from immediately responding to the target movement. Therefore, the control system determines that there is a preload offset in the flexible joint and controls the drive motor to perform a slight adjustment movement opposite to the target movement direction, causing the flexible transmission component to release the torsional energy stored inside.

[0024] During the reverse adjustment process, the drive motor generates a reverse driving torque, causing the flexible transmission component to gradually recover from its initial torsional state to a stable state. Simultaneously, the changes in the drive end angle θm and the output end angle θo are detected in real time. When the angle difference Δθ is detected to gradually decrease and enter a stable range, the control system records the current drive end position, output end position, and corresponding torque state, and uses this state as the initial control reference for the flexible joint to perform subsequent actions.

[0025] The torsional stiffness Kt of the flexible transmission component determines the relationship between the angle difference Δθ and the transmitted torque τs. During the robot's execution, the control system continuously collects changes in the output angle and load. When a deviation between the output response and the target trajectory is detected, the drive torque output is adjusted according to the degree of deviation. This allows the flexible joint to dynamically correct the compensation amount based on the actual load state, thereby reducing positional lag caused by excessive torsion of the flexible transmission component and improving the accuracy and stability of the humanoid robot's joint motion response.

[0026] Preferably, the method for determining whether there is a preload offset affecting the initial response before joint activation by assessing the residual torsional state in step S2 includes: The relative angular displacement between the drive end and the output end is obtained when the flexible joint is in a stopped state, and the torsional energy storage corresponding to the flexible transmission component is determined based on the relative angular displacement. The release stability corresponding to the residual torsional state is determined based on the relationship between the torsional energy storage and the joint stop holding time. When the residual torsional state persists in the stopped state and the torsional energy storage has not reached a stable convergence state, it is determined that there is a pre-tightening offset inside the flexible transmission component. By analyzing the direction of the angle change corresponding to the preload offset and the initial direction of the target motion, it can be determined whether the preload offset will cause a response deviation during the joint initiation phase.

[0027] In some embodiments, taking a flexible drive mechanism for the shoulder joint of a bipedal humanoid robot as an example, the shoulder joint is used to drive the robot arm to complete lifting, rotation and load grasping actions.

[0028] After the robot completes a grasping action and enters a stopped state, the control system acquires the motor shaft angle position collected by the encoder at the shoulder joint drive end and the robotic arm joint axis angle position collected by the encoder at the output end. Based on the difference between the drive end angle position and the output end angle position, the relative angular displacement of the flexible joint in the stopped state is obtained. For example, when the robot stops after grasping a heavy object, due to the continuous load on the robotic arm end, the output end is deflected by the load torque, while the drive end remains in the locked motor position, creating a certain angular difference between the drive end and the output end.

[0029] Based on the relative angular displacement and the torsional stiffness parameters of the flexible transmission component, the torsional energy stored inside the flexible transmission component is calculated. For example, when a relative angular displacement of 2° is detected between the drive end and the output end of the shoulder joint, the control system calculates the corresponding elastic recovery energy based on the elastic characteristics of the flexible transmission component and uses this energy as a characterization parameter of the current residual torsional state.

[0030] During the robot's stop and hold process, the control system continuously collects data on the changes in torsional energy stored at multiple time points and analyzes the release stability of the residual torsional state in conjunction with the stop and hold time. For example, in the initial stage after stopping, the elastic energy stored inside the flexible transmission component will be slowly released due to the elastic recovery of the material, causing the torsional energy stored to gradually decrease; if the torsional energy stored continues to change and has not entered a stable convergence state after the set hold time, it indicates that there is still unreleased elastic torsion inside the flexible transmission component.

[0031] When the control system detects that the shoulder joint continues to exhibit residual torsion while stationary, and the corresponding torsional energy has not reached a stable convergence state, it determines that there is pre-tension offset within the flexible transmission component. For example, after the robot arm stops gripping a heavy object, the output end of the shoulder joint is subjected to gravitational torque, causing the flexible transmission component to maintain a torsional state in the downward direction for a long time. In this case, it is considered that the residual torsion will affect the initiation of the next action.

[0032] The control system determines whether the preload offset will cause a deviation in the startup response based on the direction of the angle change corresponding to the preload offset and the initial motion direction corresponding to the next action command. For example, when the robot's next action command is to raise the arm, and the residual torsion direction inside the flexible transmission component is downward, part of the torque output by the drive motor during the startup phase is first used to offset the residual torsional energy, preventing the output end from immediately following the drive end's movement. Therefore, it is determined that the preload offset will affect the initial joint response. Conversely, when the residual torsion direction is consistent with the target motion direction, or when the residual torsion amount is lower than the preset response influence range, it is determined that the preload offset will not significantly affect the startup process.

[0033] It is important to note that in the above embodiments, the release stability is not determined solely by the torsional energy storage at a single moment, but rather by whether the torsional energy storage gradually becomes less significantly different during the continuous stop-and-hold process. Specifically, the control system compares the changes in torsional energy storage at adjacent time points. When the decrease in energy storage gradually decreases with the stop-and-hold time and remains relatively small across multiple consecutive time points, the residual torsional release process is considered to be gradually stabilizing. If the energy storage continues to decrease significantly, the residual torsion is considered to still be in the release process, thus obtaining the corresponding release stability.

[0034] Meanwhile, the formula for calculating torsional energy storage can be found as Es = ½ × Kt × (Δθ)², where Kt is the torsional stiffness and Δθ is the relative angular displacement.

[0035] Preferably, the release stability corresponding to the residual torsional state is determined based on the relationship between the torsional energy storage and the joint stop holding time, including: Obtain the change in torsional energy storage of the flexible joint in a stationary state, and extract the decay trend of torsional energy storage. Based on the attenuation trend analysis, the energy change rate and release amplitude changes during the release process of the internal elastic energy storage of the flexible transmission component are analyzed. Based on the changes in energy change rate and release amplitude, the steady-state parameters of the residual torsional release process are obtained; Based on the correlation between steady-state parameters and joint holding time, the release stability of residual torsion before the flexible joint is activated is obtained.

[0036] In some embodiments, taking a flexible drive mechanism for the shoulder joint of a bipedal humanoid robot as an example, the shoulder joint is used to drive the robot arm to complete lifting, rotation, and grasping actions. The shoulder joint includes a drive motor, a reduction transmission assembly, a flexible transmission assembly, and an output joint shaft. The flexible transmission assembly is disposed between the drive end and the output end, and is used to absorb impact loads during robot movement and improve joint compliance. When the robot completes the grasping action and enters a stopped state, due to the arm's own weight and end-effector load, the flexible transmission assembly will generate a certain degree of elastic torsion, causing residual torsional energy to accumulate internally. The release process of this energy will affect the response state when the next action is initiated.

[0037] After the robotic arm completes its grasping action and remains stationary, the control system acquires the change in torsional energy stored in the flexible transmission component of the shoulder joint. First, an encoder mounted on the drive end obtains the rotation angle of the shoulder joint drive motor, and an angle detection element mounted on the output end obtains the actual joint angle of the robotic arm. The degree of torsion of the flexible transmission component is determined based on the relative angular displacement between the drive and output ends. Further, combined with the torsional stiffness parameters of the flexible transmission component, the corresponding torsional energy stored in the current state is calculated.

[0038] For example, when the robotic arm stops after grasping a heavy object, the load causes a continuous downward torque at the output end of the shoulder joint, while the drive motor remains locked. At this time, the flexible transmission component undergoes elastic torsion. The control system collects the angle difference between the drive end and the output end at multiple time points after stopping, and records the torsional energy stored at each time point, forming a sequence of energy storage changes as the stopping time changes, thereby extracting the trend of torsional energy decay over time.

[0039] The energy change rate and release amplitude during the release process of the elastic stored energy inside the flexible transmission component were analyzed based on the decay trend of the torsional energy storage. For example, in the initial stage of stopping and holding, the elastic torsional energy stored inside the flexible transmission component is relatively high, the elastic recovery effect is obvious, and the torsional energy storage decreases significantly per unit time, indicating that the residual torsion is in a rapid release stage. As the stopping and holding time increases, the flexible material gradually returns to a stable state, the change in torsional energy storage per unit time gradually decreases, and the release rate decreases, indicating that the residual torsion gradually tends to stabilize.

[0040] Based on the changes in energy change rate and release amplitude, steady-state parameters are generated to characterize the residual torsional release process. Specifically, the control system determines the current release rate based on the energy change rate over a continuous time period and the degree of residual torsional release based on the cumulative release amplitude. When the rate of decrease in torsional stored energy is detected to have decreased to a stable range over multiple consecutive sampling periods, and the cumulative release amplitude tends to be fixed, the steady-state parameters are increased; when the torsional stored energy still exhibits a large rate of change, or the release amplitude continues to change, the steady-state parameters are decreased to indicate that the flexible transmission component is still in the energy release phase.

[0041] Based on the correlation between steady-state parameters and shoulder joint hold time, the residual torsional release stability before the flexible joint is activated is obtained. For example, if the steady-state parameters reach a stable threshold and the change in torsional energy remains within a small range after the robot's shoulder joint hold time reaches a set time, then it is determined that the current residual torsion has been stably released; if the steady-state parameters still do not reach the stable threshold after the hold time is increased, and the torsional energy continues to decrease, then it is determined that there is still unreleased residual torsion inside the flexible transmission component.

[0042] It is important to note that the aforementioned steady-state parameters are used to quantitatively characterize the stability of the residual torsional release process, while release stability is the state judgment result obtained based on these steady-state parameters. Specifically, the control system forms steady-state parameters based on the magnitude and trend of torsional energy storage changes between consecutive time points, and determines whether the current residual torsion has completed its main energy release based on whether the steady-state parameters continuously tend to stabilize. Therefore, the steady-state parameters reflect "to what extent it is stable," while release stability reflects "whether it has stabilized," and both are used together to determine whether the residual torsion will still affect the initial response before joint initiation.

[0043] As an example of the present invention, reference is made to... Figure 2 As shown, step S3 in this example includes: Step S31: Determine the reverse adjustment direction of the drive end according to the angle change direction corresponding to the preload offset, and generate the corresponding torsional release compensation displacement. Step S32: Based on the torsion release compensation displacement control, the drive end moves in the opposite adjustment direction, so that the elastic torsion accumulated inside the flexible transmission component is gradually released, and the relative displacement change between the drive end and the output end is obtained during the adjustment process; Step S33: Extract the stable correspondence of the flexible transmission component after the residual torsion is released based on the relative displacement change, and associate the position of the drive end and the position of the output end after the release to form the joint initial position reference; Step S34: Based on the initial joint position reference and the corresponding driving torque state, generate the initial control reference when the flexible joint executes the action command.

[0044] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0045] The control system determines the reverse adjustment direction of the drive end based on the angular change direction corresponding to the preload offset. Specifically, when the robot arm is in a horizontally extended state and remains stationary, the downward load torque generated by the end-effector causes the shoulder joint output end to deflect downward relative to the drive end, resulting in residual torsion within the flexible transmission assembly along the arm's descent direction. At this time, the control system determines, based on the angular change direction between the drive end and the output end, that the drive motor needs to perform an adjustment action opposite to this residual torsion direction, i.e., controlling the drive end to adjust in the opposite direction of the arm's lifting direction.

[0046] Based on the angular difference corresponding to the current preload offset and the elastic characteristics of the flexible transmission component, a corresponding torsional release compensation displacement is generated. For example, the control system converts the detected angular offset between the drive end and the output end into a reverse compensation angle corresponding to the drive motor, and uses this reverse compensation angle as the torsional release compensation displacement. This ensures that the drive end only generates compensating motion to counteract the elastic torsion during the release of residual torsion, without changing the current posture of the robot arm.

[0047] The control system controls the drive end to move in the reverse adjustment direction based on the generated torsional release compensation displacement, so that the elastic torsional energy accumulated inside the flexible transmission component is gradually released. Specifically, the drive motor outputs a small reverse adjustment torque, causing the motor shaft to produce a slow reverse displacement. This displacement is transmitted to the flexible transmission component through the reduction transmission component, so that the elastic torsional energy stored inside is gradually released.

[0048] During the release process, the control system continuously collects the positions of the drive end and the output end, and calculates the relative displacement change between them. For example, in the initial stage of reverse adjustment, due to the large elastic deformation inside the flexible transmission component, the movement of the drive end is mainly used to release the internal torsional energy, and the position change of the output end is small; as the residual torsion gradually decreases, the motion of the drive end is gradually transmitted to the output end, causing the relative displacement between the drive end and the output end to continuously decrease.

[0049] The control system extracts the stable correspondence of the flexible transmission component after the residual torsional release is completed based on the relative displacement changes during the release process. Specifically, it analyzes the relative displacement changes between the drive end and the output end within multiple consecutive sampling periods. When the relative displacement change gradually decreases and remains within a stable range for multiple consecutive periods, it is determined that the residual torsional release inside the flexible transmission component is complete.

[0050] Subsequently, the positions of the shoulder joint drive end and output end in the released state are acquired, and a stable correspondence between the two is established. For example, the motor shaft angle position collected by the drive motor encoder is associated with the robotic arm posture angle corresponding to the shoulder joint output shaft to form a position mapping relationship between the shoulder joint drive end and output end in the current load state, and the drive end position and output end position corresponding to this mapping relationship are used as the initial position reference of the joint.

[0051] Based on the initial joint position reference and the corresponding drive torque state, an initial control reference is generated when the flexible joint executes motion commands. Specifically, the control system records the drive end position, output end position, and drive motor output torque state of the shoulder joint when the residual torsion is released, and uses these parameters as the initial reference state for subsequent motion control. For example, when the robot receives the "raise arm to grasp" motion command again, the control system determines the initial output torque of the drive motor based on this initial control reference, so that the power output from the drive end can directly act on the effective joint movement, avoiding the problem that the initial output of the motor is used to overcome elastic deformation due to the residual torsion inside the flexible transmission component, thereby improving the response speed and motion control accuracy of the shoulder joint initiation phase.

[0052] It should be noted that the corresponding drive torque state is the holding torque output by the drive end at the moment when the residual torsion release is completed.

[0053] Preferably, step S32 includes: Based on the torsion release compensation displacement control, the drive end generates reverse rotation, so that the input displacement of the drive end is opposite to the current torsion direction of the flexible transmission component; The driving angular displacement during the reverse motion of the driving end and the corresponding response angular displacement of the output end are obtained, and the torsional release change of the flexible transmission component is extracted based on the angular displacement difference between the driving angular displacement and the response angular displacement. Based on the change process of the torsional release, the displacement response characteristics of the flexible transmission component during the transition from the elastic energy storage state to the stable transmission state are extracted. Based on the correlation between the positional changes of the driving end and the positional changes of the output end according to the displacement response characteristics, the relative displacement change between the driving end and the output end during the adjustment process is obtained.

[0054] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0055] Based on the torsional release compensation displacement generated in step S31, the shoulder joint drive end is controlled to generate a reverse rotation, so that the input displacement direction of the drive end is opposite to the current residual torsional direction of the flexible transmission component. For example, when the robot arm stops after grasping an object, if there is residual torsion in the shoulder joint flexible transmission component along the downward direction of the arm, the control system controls the drive motor to generate a small reverse rotation in the upward direction of the arm, so that the input displacement of the drive end is used to counteract the elastic torsional energy stored in the flexible transmission component, rather than directly driving the arm to move.

[0056] During the reverse motion of the drive end, the control system acquires the corresponding drive angular displacement of the drive end and the corresponding response angular displacement of the output end in real time. The drive angular displacement is acquired by an encoder mounted on the shoulder joint drive motor side, and the output response angular displacement is acquired by an angle sensor mounted on the shoulder joint output shaft side. Based on the difference between the drive angular displacement and the output angular displacement within the same adjustment time period, the angular displacement difference between the drive end and the output end is calculated, and this angular displacement difference is used as a characterizing parameter for the torsional release change of the flexible transmission component.

[0057] For example, in the initial stage of reverse adjustment, the drive motor generates a certain angular displacement. However, due to the large elastic torsion inside the flexible transmission component, the output end responds with a small angular displacement. At this time, there is a large difference between the drive angular displacement and the output angular displacement, indicating that the input displacement is mainly used to release the internal elastic stored energy. As the reverse adjustment continues, the internal torsion of the flexible transmission component gradually decreases, and the output end begins to follow the drive end. The difference between the drive angular displacement and the output angular displacement gradually decreases.

[0058] Based on the trend of the change in torsional release during the adjustment process, the displacement response characteristics of the flexible transmission component during the transition from the elastic energy storage state to the stable transmission state are extracted. Specifically, the control system analyzes the change in angular displacement difference within a continuous sampling period. When the angular displacement difference decreases rapidly, it is determined that the flexible transmission component is in the elastic energy storage release stage; when the rate of change of angular displacement difference gradually decreases and tends to a stable range, it is determined that the flexible transmission component gradually enters the stable transmission stage.

[0059] For example, during the release of residual torsion in the shoulder joint, the initial reverse displacement of the drive end is mainly used to overcome the elastic restoring force of the flexible transmission component, resulting in a response lag at the output end. As the elastic torsion is gradually released, the input displacement of the drive end is gradually converted into effective joint motion, and the response ratio at the output end gradually increases. By extracting the relationship between the drive displacement and the output displacement during this process, the displacement response characteristics of the flexible transmission component when transitioning from an energy storage state to a stable transmission state can be obtained.

[0060] Based on the aforementioned displacement response characteristics, a correspondence is established between the positional changes at the drive end and the output end, yielding the relative displacement change between the drive end and the output end during the adjustment process. Specifically, the control system establishes a dynamic correspondence based on the change in angular displacement at the drive end and the angular displacement response at the output end throughout the entire reverse adjustment process, and continuously updates the relative displacement between the two. When the relative displacement change is detected to decrease to a stable range, it is determined that the residual torsion within the flexible transmission component has been released, and this state is used as the basis for subsequently establishing the initial joint position reference.

[0061] Preferably, before the reverse rotation is generated by the displacement control drive based on torsion release compensation, the following steps are also included: Obtain the torsional direction corresponding to the preload offset of the flexible transmission component and the initial force state of the drive end, and determine the torque release direction during the reverse adjustment process based on the torsional direction and the force state; Based on the torque release direction, a slow-release driving force smaller than the target reverse adjustment torque is applied to the drive end, so that a progressive torque transmission state is formed between the drive end and the flexible transmission component. The correlation features between the displacement change at the drive end and the response change at the output end under the action of slow-release driving force are obtained, and the start-up response parameters of the flexible transmission component when it transitions from a stopped state to a releaseable state are extracted. Adjust the timing of applying the torsional release compensation displacement according to the start-up response parameters, so that the drive end performs reverse rotation in the elastic torsional release direction.

[0062] In some embodiments, taking a flexible drive mechanism for the shoulder joint of a bipedal humanoid robot as an example, the shoulder joint includes a drive motor, a reduction transmission assembly, a flexible transmission assembly, and an output joint shaft. When the robot arm completes the grasping action and enters a stopped state, due to the end-effector load, an elastic torsion along the load direction is generated inside the flexible transmission assembly of the shoulder joint, causing a preload offset between the drive end and the output end.

[0063] Before the drive end generates reverse rotation based on the displacement control of torsion release compensation, the control system first acquires the torsional direction corresponding to the preload offset of the flexible transmission component and the initial force state of the drive end. Specifically, the angular deviation between the drive end position sensor and the output end position sensor is acquired, and the current residual torsional direction of the flexible transmission component is determined based on the direction of the angular deviation. At the same time, the current output torque of the drive motor, the holding current, and the load feedback information are collected to determine the initial torque state experienced by the drive end in the stopped state.

[0064] For example, when the robot arm holds a horizontal position after grasping an object, the weight of the arm and the weight of the grasped object generate a downward rotational torque, causing the output end of the shoulder joint to shift downward relative to the drive end. At this time, the residual torsion direction inside the flexible transmission component is determined to be the downward direction. At the same time, the holding torque currently received by the drive end is determined based on the output current of the drive motor in the holding state, and the torque release direction in the subsequent reverse adjustment process is determined in combination with the above torsion direction.

[0065] Based on the determined torque release direction, a slow-release driving force, less than the target reverse adjustment torque, is applied to the drive end, creating a gradual torque transmission state between the drive end and the flexible transmission component. Specifically, the control system first applies a small-amplitude reverse driving torque to the shoulder joint drive motor, causing the drive end to rotate slowly, but without directly applying the full compensation torque. During this process, the input torque at the drive end is gradually transmitted to the flexible transmission component, causing the internal elastic torsion to begin to release.

[0066] For example, when it is determined that residual torque needs to be released in the lifting direction, the control system first controls the drive motor to output a reverse torque lower than the target adjustment value, so that the drive end slowly rotates in the lifting direction; as the internal torque of the flexible transmission component gradually decreases, the drive torque is gradually adjusted according to the release state, so that the drive end and the flexible transmission component maintain a continuous torque transmission process.

[0067] The correlation characteristics between the displacement change at the drive end and the response change at the output end under the action of the slow-release driving force are obtained, and the start-up response parameters of the flexible transmission component when it transitions from a stopped state to a releaseable state are extracted. Specifically, the control system synchronously collects the change in angular displacement at the drive end and the angular displacement response at the output end, and calculates the response difference and the rate of change of the response between the two. When the drive end generates displacement but the response at the output end is small, it is determined that the flexible transmission component is still in the elastic energy storage and release stage; when the angular displacement of the output end gradually increases and forms a stable correspondence with the change in displacement at the drive end, it is determined that the flexible transmission component has entered the releaseable state.

[0068] For example, in the initial stage of the slow-release driving force, the rotation of the drive motor is mainly used to counteract the elastic torsion inside the flexible transmission component, and the angle of the shoulder joint output shaft changes little; as the internal elastic energy is gradually released, the output end begins to follow the movement of the drive end. At this time, the control system records the displacement change of the drive end, the displacement change of the output end, and the relationship between the two to form the start-up response parameters.

[0069] The timing of the torsional release compensation displacement is adjusted according to the start-up response parameters, causing the drive end to perform reverse rotation in the elastic torsional release direction. Specifically, the control system determines the loading time and speed of the compensation displacement based on the start-up response parameters. When it is detected that the flexible transmission component is still in the elastic energy storage and release stage, the loading speed of the compensation displacement is reduced, causing the drive end to maintain a slow reverse movement; when it is detected that a stable correspondence is reached between the drive end displacement and the output end response, the loading amplitude of the compensation displacement is increased, causing the drive end to complete the subsequent torsional release action in the predetermined reverse adjustment direction.

[0070] It is important to note that the "releasable state" in the above embodiments refers to a state in which the residual torsion inside the flexible transmission component has changed from the static energy storage state of the stop-hold phase to a state in which it can be actively released through the reverse displacement of the drive end without causing a significant abrupt change in the output response. The difference between this and the "stopped state" is that in the stopped state, the drive end does not apply an active release displacement, and the flexible transmission component mainly maintains its current residual torsion; the difference between this and the "stable transmission state" is that the releasable state only indicates that the residual torsion has met the conditions for active release, and does not require that the drive end and the output end have established a stable displacement transmission relationship. Specifically, during the process of releasing the driving force, the control system simultaneously acquires the change in displacement at the drive end and the change in response at the output end. When the drive end begins to generate continuous reverse displacement, and the output end response gradually changes from being basically stationary to generating a displacement change consistent with the direction of the drive end's movement, and the response difference between the two no longer continues to increase, it is determined that the residual torsion inside the flexible transmission component has entered the active release range, thus determining that the flexible transmission component has transitioned from the stopped state to the releasable state; only then is the torsion release compensation displacement applied according to the start-up response parameters. In this way, the "releasable state" specifically corresponds to the critical stage where the residual torsion can be actively released by the drive end, avoiding the direct application of large compensation displacement when the flexible transmission component is still in a purely elastic energy storage state.

[0071] Preferably, based on the torque release direction, a slow-release driving force less than the target reverse adjustment torque is applied to the drive end, so that a gradual torque transmission state is formed between the drive end and the flexible transmission component. Specifically: Obtain the current torsional force state at the connection between the drive end and the flexible transmission component, and determine the loading direction of the mitigating drive force based on the torsional force state; The control drive end applies a gradually increasing slow-release driving force along the torque release direction, causing the drive end to displace and drive the release of the elastic energy stored inside the flexible transmission component; The difference between the displacement change at the drive end and the response change at the output end during the loading of the slow-release driving force is collected, and the degree of elastic deformation release inside the flexible transmission component is extracted based on the difference. The loading amplitude of the slow-release driving force is adjusted based on the degree of elastic deformation release, so that the input torque at the driving end is gradually transmitted to the output end.

[0072] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0073] Specifically, the current torsional stress state at the connection between the drive end and the flexible transmission component is first obtained. The control system determines the current torsional load state of the connection area between the drive end and the flexible transmission component by collecting information on the output torque of the drive motor, the holding current, and the change in the angular position of the drive end. At the same time, combined with the direction of the angular deviation between the drive end and the output end, the direction of the elastic torsional torque inside the flexible transmission component is determined, and the loading direction of the mitigating drive force is determined based on this torsional stress state.

[0074] For example, when the robot arm is in a horizontal grasping state, due to the weight of the arm and the end effector, the output end of the shoulder joint experiences a downward rotational torque, causing a downward torsional force state inside the flexible transmission component. At this time, the control system determines that the direction of the slow-release driving force is opposite to the direction of this torsional force, that is, the drive end applies a slow-release torque along the lifting direction of the arm.

[0075] The control system applies a gradually increasing, slow-release driving force to the drive end along a predetermined torque release direction, causing displacement of the drive end and releasing the elastic energy stored inside the flexible transmission component. Specifically, the drive motor first outputs an initial slow-release torque lower than the target reverse adjustment torque, causing the drive end to rotate slightly; subsequently, the driving torque is gradually increased according to the torsional release state inside the flexible transmission component, so that the input torque gradually acts on the flexible transmission component.

[0076] For example, in the initial stage of the control system, a small reverse torque is applied to the shoulder joint drive motor to make the drive end rotate slowly. At this time, the displacement generated by the drive end is mainly used to release the elastic torsion stored inside the flexible transmission component. As the elastic torsion decreases, the control system gradually increases the release driving force so that the displacement of the drive end is gradually transmitted to the output end.

[0077] During the gradual release of driving force loading, the control system synchronously collects the difference between the displacement change at the drive end and the response change at the output end. Specifically, the angular displacement at the drive end is obtained through the encoder at the drive end, and the response angular displacement corresponding to the shoulder joint output axis is obtained through the angle sensor at the output end. The difference in angular displacement between the two within the same time period is then calculated.

[0078] Based on the change process of the angular displacement difference, the degree of elastic deformation release inside the flexible transmission component is extracted. For example, in the initial loading stage of the slow-release driving force, the driving end has already produced a significant displacement, while the angular displacement change at the output end is small, indicating that the input displacement at the driving end is mainly used to release the internal elastic deformation, and the degree of elastic deformation release is low at this time; as the slow-release driving force continues to act, the angular displacement difference between the driving end and the output end gradually decreases, indicating that the elastic torsion inside the flexible transmission component gradually decreases, and the degree of elastic deformation release gradually increases.

[0079] The loading amplitude of the slow-release driving force is adjusted based on the degree of elastic deformation release, so that the input torque at the drive end is gradually transmitted to the output end. Specifically, when the degree of elastic deformation release is detected to be low, the control system maintains a small amplitude of slow-release driving force to keep the flexible transmission component in a slow energy release state; when the difference between the displacement change at the drive end and the response change at the output end is detected to decrease and the two gradually form a stable correspondence, the loading amplitude of the slow-release driving force is increased, so that more input torque is transmitted to the output end.

[0080] For example, during the release of residual torsion in the shoulder joint, the control system dynamically adjusts the release torque based on the response difference between the drive end and the output end. When the displacement of the drive end is mainly used to release elastic torsion, the torque loading speed is reduced. When the output end begins to follow the movement of the drive end, the driving torque is increased, so that the input of the drive end is gradually converted into effective movement of the shoulder joint.

[0081] It should be noted that the "degree of elastic deformation release" in the above embodiments is used to characterize the proportion of the original elastic torsion released inside the flexible transmission component after the application of the slow-release driving force, and is quantified based on the change in the angular displacement difference between the driving end and the output end. Specifically, before applying the slow-release driving force, the initial angular displacement difference between the driving end and the output end is obtained and recorded as the initial angular displacement difference; during the application of the slow-release driving force, the current angular displacement difference at the corresponding moment is obtained in real time, and the degree of elastic deformation release is determined based on the ratio of the change between the initial angular displacement difference and the current angular displacement difference. The calculation relationship can be expressed as follows: Elastic deformation release degree = (initial angular displacement difference - current angular displacement difference) / initial angular displacement difference × 100%.

[0082] For example, if there is an initial angular displacement difference of 2° between the driving end and the output end before the application of the slow-release driving force, and the current angular displacement difference decreases to 0.8° during the slow-release process, then the degree of elastic deformation release is 60%, indicating that a large part of the original elastic torsion has been released. As the current angular displacement difference further decreases, the degree of elastic deformation release increases accordingly, which serves as the basis for adjusting the loading amplitude of the slow-release driving force.

[0083] Preferably, adjusting the loading amplitude of the slow-release driving force based on the degree of elastic deformation release includes: Based on the release stage corresponding to the degree of elastic deformation release, the current slow-release driving force applied by the driving end is obtained, and the slow-release driving force is adjusted incrementally according to the preset increment; When the flexible transmission component is in the initial stage of elastic deformation release, the slow-release driving force is controlled within the range of 10% to 30% of the target reverse adjustment torque, so that the input torque at the drive end is preferentially applied to the elastic recovery process of the flexible transmission component. When the elastic deformation release reaches 70% to 90%, the slow-release driving force is increased by 0.02 N·m to 0.05 N·m each time, so that the input torque is gradually transmitted to the output end. The change in angular displacement between the drive end and the output end is obtained. When the change in angular displacement is stable within the range of 0.05° to 0.15°, the slow-release driving force is increased to 50% to 80% of the target reverse adjustment torque, thus forming a stable torque transmission state between the drive end and the output end.

[0084] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0085] Specifically, the control system first obtains the current applied slow-release driving force based on the release stage corresponding to the degree of elastic deformation release, and then adjusts the slow-release driving force incrementally according to a preset increment. The degree of elastic deformation release is determined by the difference between the displacement change at the drive end and the displacement change at the output end. When displacement is detected at the drive end but the output end response is small, it is determined that the flexible transmission component is still in the elastic energy storage release stage; when the displacement difference between the drive end and the output end gradually decreases, it is determined that the elastic energy is gradually released.

[0086] For example, when the robot's shoulder joint performs residual torsional release, the control system first obtains the current release torque value applied to the drive motor, and gradually increases the output torque of the drive motor according to the set increment, so that the input torque can gradually increase with the change of the state of the flexible transmission component, rather than directly loading the target to adjust the torque in the opposite direction.

[0087] When the flexible transmission component is in the initial stage of elastic deformation release, the control system controls the release driving force within the range of 10% to 30% of the target reverse adjustment torque, so that the input torque at the drive end preferentially acts on the elastic recovery process of the flexible transmission component. Specifically, when the shoulder joint detects that there is still a large angular displacement difference between the drive end and the output end, it indicates that the flexible transmission component still stores a lot of elastic torsional energy. At this time, the control system limits the output torque of the drive motor. For example, when the target reverse adjustment torque is 1 N·m, the initial release driving force is controlled within the range of 0.1 N·m to 0.3 N·m, so that the drive end produces a slow reverse rotation. The input torque at the drive end is first used to overcome the elastic recovery force inside the flexible transmission component.

[0088] When the elastic deformation release reaches 70%–90%, the control system increases the release driving force by an increment of 0.02 N·m–0.05 N·m, gradually transmitting the input torque to the shoulder joint output end. Specifically, the control system continuously acquires the angular displacement of the drive end and the angular displacement of the output end, and calculates the difference between the two. When the angular displacement difference continuously decreases, it indicates that the elastic deformation inside the flexible transmission component is gradually decreasing. At this point, the control system gradually increases the release driving force according to the set torque increment. For example, if the current release driving force is 0.25 N·m, an adjustment of 0.03 N·m increases the release driving force to 0.28 N·m, gradually transforming the input torque at the drive end from an elastic energy storage and release process to a drive output end motion process.

[0089] The control system acquires the change in angular displacement between the drive end and the output end, and determines whether the change in angular displacement is stable within the range of 0.05° to 0.15°. When the change in angular displacement remains within this range for multiple consecutive sampling periods, it indicates that a stable correspondence has been established between the input displacement of the drive end and the response displacement of the output end.

[0090] At this point, the control system increases the release driving force to 50%–80% of the target reverse adjustment torque, creating a stable torque transmission state between the drive end and the flexible transmission component. For example, when the target reverse adjustment torque is 1 N·m, the release driving force is increased to 0.5 N·m–0.8 N·m, ensuring that the output torque of the drive end is stably transmitted to the output end of the shoulder joint and completing the subsequent residual torsional release process.

[0091] Preferably, step S4 includes the following steps: The initial drive torque input is determined based on the initial control reference when the flexible joint executes the action command, and the drive component is controlled to drive the flexible joint to move according to the initial drive torque input. The actual angular displacement of the flexible joint output end and the corresponding change in driving torque are obtained during the execution of the motion command, and the torsional response change of the flexible transmission component as the motion is advanced is extracted. Based on the offset relationship between the torsional response change and the initial control reference, the torque adjustment amount for the corresponding action stage is generated; The subsequent drive torque input is corrected based on the torque adjustment amount, so that the output end of the flexible joint maintains the motion state corresponding to the action command.

[0092] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0093] The initial drive torque input for executing a motion command is determined based on the established initial control reference of the shoulder joint. This initial control reference includes the position of the drive end, the position of the output end, and the corresponding drive torque state after the residual torsion is released. For example, when the robot receives a "raise arm to grasp" motion command, the control system calculates the initial drive torque input required for the drive motor startup phase based on the current initial position reference of the shoulder joint and the target motion direction. It then controls the drive motor to output power according to this initial drive torque input, causing the shoulder joint drive end to move the flexible transmission component.

[0094] During the execution of motion commands, the control system acquires the actual angular displacement of the shoulder joint output end and the corresponding change in driving torque. Specifically, the actual rotation angle of the robotic arm is acquired by an angle sensor located at the output joint axis, the current change in driving torque is acquired by a torque detection unit on the drive motor side, and the torsional response changes of the flexible transmission component as the motion progresses are extracted based on the changes in angular displacement and torque corresponding to different motion stages.

[0095] For example, during the movement of the robot arm from a drooping state to a horizontally raised state, the output torque of the drive motor is mainly used to overcome the resistance generated by the robot arm's own weight and the elastic deformation of the flexible transmission component in the initial stage. At this time, the change in angular displacement of the output end is somewhat lagging behind that of the drive end. As the shoulder joint continues to move, the internal torsional state of the flexible transmission component changes, and the output end gradually follows the movement of the drive end. The control system records the relationship between the change in drive torque and output angular displacement during this process.

[0096] Based on the offset relationship between the torsional response changes extracted during the operation and the initial control reference, the torque adjustment amount for the corresponding operation stage is generated. Specifically, the control system compares the actual angular displacement and drive torque change of the output end collected in the current operation stage with the corresponding state parameters in the initial control reference. When the actual torsional response generated by the flexible transmission component deviates from the initial state, the torque adjustment amount that needs to be compensated is calculated according to the offset direction and degree.

[0097] For example, during the process of a robot grasping an object, when the increased end load causes the angular displacement of the shoulder joint output end to be less than the angular displacement corresponding to the target motion trajectory, the control system determines that the current flexible transmission component generates additional elastic torsion, causing the output end response to lag, and generates an adjustment amount to increase the driving torque; when the output end angular displacement is detected to exceed the target state, an adjustment amount to reduce the driving torque is generated.

[0098] The control system adjusts the subsequent drive torque input based on the generated torque adjustment, ensuring that the shoulder joint output maintains a motion state corresponding to the action command. Specifically, the control system dynamically adjusts the drive motor input torque according to the torque adjustment during subsequent action cycles, matching the actual output torque with the current torsional state of the flexible transmission component.

[0099] For example, if additional torsion is detected in the flexible transmission component due to load changes during the continuous lifting of the robot arm, the control system increases the driving torque input for the corresponding stage to keep the angular displacement of the output end consistent with the target motion trajectory; when the flexible transmission component returns to a stable transmission state, the control system reduces the compensation torque to maintain a stable motion relationship between the drive end and the output end.

[0100] It is important to note that in the above embodiments, the "torsional response change" is used to characterize the impact of force changes on the flexible transmission component during motion execution on the output motion response. Specifically, the control system acquires the changes in driving torque and output angular displacement at adjacent sampling times in the order of motion execution, and extracts the torsional response change of the flexible transmission component based on the output angular displacement change corresponding to a unit change in driving torque. When the driving torque changes but the output angular displacement change is small, it indicates that the added driving torque is mainly used to overcome the torsional deformation of the flexible transmission component; when the change in driving torque can cause a synchronous change in the output angular displacement, it indicates that the driving input is gradually converted into effective joint motion, thereby obtaining the torsional response change during the motion propulsion process.

[0101] Furthermore, the "offset relationship" specifically includes the angular deviation between the actual angular displacement at the output end and the corresponding target angular displacement in the initial control reference, and the torque deviation between the actual driving torque and the corresponding reference driving torque in the initial control reference. The control system calculates the above-mentioned angular deviation and torque deviation according to the current action stage, and determines the offset of the actual torsional response of the flexible transmission component relative to the initial control reference by combining the directions of change of the two. For example, when the actual output angular displacement is less than the corresponding target angular displacement and the actual driving torque is higher than the corresponding reference driving torque, it indicates that part of the driving torque is consumed by the additional torsional deformation of the flexible transmission component, and the control system generates a torque adjustment amount to increase the subsequent driving torque accordingly; conversely, it generates a corresponding torque adjustment amount to decrease the torque.

[0102] Of particular importance is acquiring the actual angular displacement of the flexible joint output end and the corresponding change in driving torque during the execution of the motion command, and extracting the torsional response changes of the flexible transmission component as the motion progresses, including: The changes in input torque at the drive end and the actual changes in angular displacement at the output end are obtained during the execution of the action command, and the correspondence between the changes in drive torque and the response of output angular displacement is established. Based on the correspondence, the change in output angular displacement corresponding to the unit change in torque is extracted, and correlation analysis is performed on the stages where the change in output angular displacement is lower than the corresponding motion response range. Based on the relationship between the unit torque change and the output angular displacement change, the torsional energy storage buffer zone inside the flexible transmission component is identified to absorb the driving torque. Based on the differences in duration, torque increment, and angular displacement response between the torsional energy storage buffer zones, elastic response migration characteristics that characterize the dynamic energy storage and release process of the flexible transmission component are generated. Based on the elastic response migration characteristic, the torsional response change of the flexible transmission component during the motion propulsion process is obtained.

[0103] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0104] Specifically, after the robot receives the "raise arm to grasp" action command, the control system acquires the change in input torque at the drive end and the change in actual angular displacement at the shoulder joint output end during the action execution process. The change in input torque at the drive end is acquired by a torque detection unit located at the output end of the drive motor, and the change in actual angular displacement at the output end is acquired by an angle sensor located at the output shaft of the shoulder joint.

[0105] For example, during the process of the robot arm moving from the initial drooping state to the target grasping posture, the control system acquires the motor output torque and shoulder joint output angle corresponding to each time node according to the set sampling period, and records the correspondence between the increase in driving torque and the change in output angular displacement in different action stages, forming a mapping relationship between the change in driving torque and the output angular displacement response.

[0106] Based on the above correspondence, the change in output angular displacement corresponding to a unit change in torque is extracted. Specifically, the control system calculates the ratio between the change in input torque at the drive end and the change in output angular displacement within adjacent sampling periods to obtain the angular displacement response corresponding to a unit change in torque, which is used to characterize the degree of transmission of driving torque by the flexible transmission component. For example, in the initial stage of shoulder joint activation, if the change in output angular displacement is small when the output torque of the drive motor gradually increases, it indicates that the input torque has not been completely converted into joint motion, but mainly acts on the elastic deformation inside the flexible transmission component; when the change in output angular displacement gradually increases, it indicates that the driving torque begins to be effectively transmitted to the output end.

[0107] A correlation analysis is performed on motion phases where the change in output angular displacement is lower than the corresponding motion response range. Specifically, the control system compares the output angular displacement response corresponding to a unit torque change with the response range under normal shoulder joint motion. When an increase in unit torque change is detected but the output angular displacement change is lower than the normal response range, the corresponding motion interval for this phase is marked, and analysis is performed in conjunction with the duration of this phase and the change in torque input. For example, in the initial phase when the robot arm begins to lift, the output torque of the drive motor has increased, but the change in the angle at the output end of the shoulder joint is still small. At this time, the control system determines that there is a situation where the driving torque is not directly converted into output motion, and this phase is designated as the torsional response analysis interval for the flexible transmission component.

[0108] Based on the relationship between the change in unit torque and the change in output angular displacement, a torsional energy storage buffer zone inside the flexible transmission component is identified to absorb the driving torque. Specifically, when the change in unit torque continuously increases while the change in output angular displacement remains at a low level, it is determined that the input energy at the drive end is mainly used for the accumulation of elastic deformation inside the flexible transmission component, and this stage corresponds to the torsional energy storage buffer zone. For example, during shoulder joint initiation, the torque applied by the drive motor is first used to overcome the residual elasticity inside the flexible transmission component and the torsional resistance caused by the load. At this time, the drive end generates angular displacement, while the output end response is weak. The control system determines this time period as the torsional energy storage buffer zone. When the output end angular displacement response recovers and increases synchronously with the driving torque, it indicates that the flexible transmission component has entered a stable transmission state from the energy storage state.

[0109] Based on the duration, torque increment, and angular displacement response difference between the torsional energy storage buffer zones, an elastic response migration characteristic quantity characterizing the dynamic energy storage and release process of the flexible transmission component is generated. Specifically, the control system acquires the duration of the torsional energy storage buffer zone, calculates the cumulative increase in driving torque during this stage, and calculates the difference between the change in angular displacement at the driving end and the change in angular displacement at the output end. These parameters are correlated to form an elastic response migration characteristic quantity describing the process of the flexible transmission component changing from an elastic energy storage state to a stable transmission state. For example, when the load on the robot arm increases, the duration of the torsional energy storage buffer zone lengthens, and the difference in angular displacement response between the driving end and the output end increases, thus increasing the corresponding elastic response migration characteristic quantity; when the load decreases or the flexible transmission component is in a stable state, the buffer zone shortens, the difference in angular displacement response decreases, and the corresponding characteristic quantity decreases.

[0110] Based on elastic response migration characteristics, the torsional response changes of the flexible transmission component during motion propulsion are obtained. Specifically, the control system determines the dynamic process of elastic energy generation, release, and stable transmission within the flexible transmission component based on the changes in elastic response migration characteristics corresponding to different motion stages, and uses this process as the torsional response state of the flexible transmission component during motion execution. For example, during a complete arm-raising and grasping motion at the shoulder joint, the control system determines the torsional response changes corresponding to the initiation stage, load change stage, and stable motion stage based on elastic response migration characteristics, thereby obtaining the real-time torsional state of the flexible transmission component during motion propulsion.

[0111] It should be noted that the "elastic response migration characteristic quantity" can be directly understood in this embodiment as the change in displacement transferred from the drive end to the output end during the slow release process, for example, it can be characterized by the ratio between the increment of the output end angular displacement and the increment of the drive end angular displacement within the same time interval; the "elastic energy retention characteristic quantity" is used to characterize the residual elastic energy that the flexible transmission component has not yet released, and can be characterized by combining the current angular displacement difference with the torsional stiffness of the flexible transmission component; the "torque-displacement energy migration spectrum" is used to represent the corresponding changes in the input torque, drive end displacement, and output end response during the slow release driving force loading process. The above characteristic quantities are all used to assist in describing the process of elastic energy storage being converted from the inside of the flexible transmission component to the output motion, and are not additional independent physical quantities.

[0112] Of particular importance is that, based on the relationship between the change in unit torque and the change in output angular displacement, the torsional energy storage buffer zone inside the flexible transmission component used to absorb driving torque is identified as follows: The correspondence between the change in driving torque and the change in output angular displacement during the execution of the action is obtained, and the amplitude of the angular displacement response corresponding to the unit torque input is normalized to generate a torque-displacement response mapping relationship. Based on the torque-displacement response mapping relationship, the energy transfer region where the angular displacement response does not grow synchronously during the driving torque input process is extracted, and the torque-displacement energy transfer spectrum corresponding to the flexible transmission component is constructed. Based on the degree of energy migration corresponding to different regions in the torque-displacement energy migration spectrum, the critical change range of the driving torque from motion output to elastic energy storage is extracted. By mapping the critical change range to the elastic deformation process of the flexible transmission component, an elastic energy stagnation characteristic quantity characterizing the torsional energy storage buffer state is generated, thereby obtaining the torsional energy storage buffer zone.

[0113] In some embodiments, the shoulder joint flexible drive mechanism of a bipedal humanoid robot is still taken as an example.

[0114] Specifically, during the robot's "arm-raising and grasping" action, the control system acquires the changes in input torque at the shoulder joint drive end and the changes in actual angular displacement at the output end. Specifically, a torque detection unit located at the output end of the drive motor acquires the drive torque values ​​at different time points, an angle sensor located at the shoulder joint output shaft acquires the actual rotation angle of the robotic arm, and the system calculates the changes in torque and output angular displacement between adjacent time points.

[0115] Based on the correspondence between the change in driving torque and the change in output angular displacement during the execution of the action, the amplitude of the angular displacement response corresponding to a unit torque input is normalized to generate a torque-displacement response mapping relationship. For example, during the shoulder joint initiation and lifting phase, when the output torque of the drive motor increases from 0.2 N·m to 0.5 N·m, the control system synchronously acquires the change in output angular displacement and calculates the amplitude of the angular displacement response corresponding to the unit torque increase. By normalizing multiple unit torque response values ​​collected at different action phases, a torque-displacement response mapping relationship is formed to describe the relationship between input torque and output displacement response.

[0116] Based on the aforementioned torque-displacement response mapping relationship, the energy migration region where the angular displacement response does not increase synchronously during the driving torque input process is extracted, and a torque-displacement energy migration spectrum corresponding to the flexible transmission component is constructed. Specifically, when a continuous increase in driving torque is detected, and the amplitude of the output angular displacement change is lower than the normal motion response range, the control system records the torque input interval and displacement response difference corresponding to this stage. Since the increase in driving torque at this time is not directly converted into joint output motion, but mainly acts on the internal elastic deformation of the flexible transmission component, this region is determined as the energy migration region.

[0117] For example, during the shoulder joint's start-up process from a stopped state, after the drive motor begins to output torque, the output shaft angle changes relatively little, indicating that the input torque is mainly used to overcome residual torsion within the flexible transmission components and to form new elastic deformation. Based on the differences between the torque input and the output angular displacement response over different time periods, the control system constructs a corresponding torque-displacement energy transfer spectrum to represent the distribution of driving torque between motion output and elastic energy storage.

[0118] Based on the degree of energy migration corresponding to different regions in the torque-displacement energy transfer spectrum, the critical change range of the driving torque's conversion from motion output to elastic energy storage is extracted. Specifically, the control system analyzes the rate of change of unit torque displacement response corresponding to each region in the torque-displacement energy transfer spectrum. When the increase in unit torque in a certain region continuously increases, while the unit output angular displacement response decreases, it indicates that the driving torque is gradually changing from joint motion output to elastic energy storage within the flexible transmission component. At this point, this region is determined as the critical change range of torque conversion. For example, during the shoulder joint initiation phase, when the driving torque increases by 0.1 N·m, the increase in output angular displacement gradually decreases, indicating that the newly added input torque is more used to drive the flexible transmission component to undergo elastic torsion; when the output angular displacement response resumes its growth, it indicates that the driving torque is converting back to joint motion output. Therefore, the control system determines the critical change range between the two based on the response change trend.

[0119] By mapping the aforementioned critical change interval to the elastic deformation process of the flexible transmission component, an elastic energy storage characteristic quantity characterizing the torsional energy storage buffer state is generated, thus obtaining the torsional energy storage buffer interval. Specifically, the control system calculates the degree of elastic energy accumulation inside the flexible transmission component based on the duration corresponding to the critical change interval, the increment of input torque, and the difference in output angular displacement response. When there is a long period of low displacement response during the input of driving torque, it indicates that the flexible transmission component continuously absorbs the input torque, and its corresponding elastic energy storage characteristic quantity is high; when the input torque and output angular displacement gradually change synchronously, it indicates that the elastic energy storage process ends and enters a stable transmission state.

[0120] For example, in the initial stage of the robot's shoulder joint lifting motion, part of the torque output by the drive motor is used to release residual torsion and generate new elastic deformation. There is a lag in the angular displacement response at the output end. This stage corresponds to the torsional energy storage buffer zone. When the flexible transmission component completes the elastic state adjustment, the driving torque can be stably transmitted to the output end. At this time, it exits the torsional energy storage buffer zone.

[0121] This invention provides a control system for a flexible joint of a humanoid robot, used to execute the control method for a flexible joint of a humanoid robot as described above. The control system for a flexible joint of a humanoid robot includes: The confirmation module is used to determine the residual torsional state of the flexible transmission component based on the angle difference between the drive end position and the output end position of the flexible joint in the stopped state. The judgment module is used to determine whether there is a preload offset that affects the initial response before the joint starts by using the residual torsional state when the flexible joint receives a motion command. The adjustment module is used to control the drive end to perform an adjustment action opposite to the target motion direction when there is preload offset, so that the flexible transmission component releases the residual torsion and establishes the initial control reference of the flexible joint. The control module is used to control the flexible joint to execute motion commands based on the initial control reference, and to correct and adjust the torque adjustment amount of the motion according to the output response during the operation of the motion command.

[0122] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0123] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A control method for flexible joints of humanoid robots, characterized in that, Includes the following steps: Step S1: Determine the residual torsional state of the flexible transmission component based on the angle difference between the driving end position and the output end position of the flexible joint in the stopped state; Step S2: In response to the flexible joint receiving an action command, determine whether there is a preload offset that affects the initial response before the joint starts by checking the residual torsional state; Step S3: When there is preload offset, the control drive end performs an adjustment action opposite to the target motion direction, so that the flexible transmission component releases the residual torsion and establishes the initial control reference of the flexible joint. Step S4: Control the flexible joint to execute motion commands based on the initial control reference, and adjust the torque adjustment amount of the motion according to the output response during the operation of the motion commands.

2. The control method for flexible joints of humanoid robots according to claim 1, characterized in that, The method for determining whether there is preload offset affecting the initial response before joint initiation by judging the residual torsional state in step S2 includes: The relative angular displacement between the drive end and the output end is obtained when the flexible joint is in a stopped state, and the torsional energy storage corresponding to the flexible transmission component is determined based on the relative angular displacement. The release stability corresponding to the residual torsional state is determined based on the relationship between the torsional energy storage and the joint stop holding time. When the residual torsional state persists in the stopped state and the torsional energy storage has not reached a stable convergence state, it is determined that there is a pre-tightening offset inside the flexible transmission component. By analyzing the direction of the angle change corresponding to the preload offset and the initial direction of the target motion, it can be determined whether the preload offset will cause a response deviation during the joint initiation phase.

3. The control method for flexible joints of humanoid robots according to claim 2, characterized in that, Based on the relationship between the torsional energy storage and the joint arrest time, the release stability corresponding to the residual torsional state is determined, including: Obtain the change in torsional energy storage of the flexible joint in a stationary state, and extract the decay trend of the torsional energy storage. Based on the attenuation trend analysis, the energy change rate and release amplitude changes during the release process of the internal elastic energy storage of the flexible transmission component are analyzed. Based on the changes in energy change rate and release amplitude, the steady-state parameters of the residual torsional release process are obtained; Based on the correlation between steady-state parameters and joint holding time, the release stability of residual torsion before the flexible joint is activated is obtained.

4. The control method for flexible joints of humanoid robots according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Determine the reverse adjustment direction of the drive end according to the angle change direction corresponding to the preload offset, and generate the corresponding torsional release compensation displacement; Step S32: Based on the torsion release compensation displacement control, the drive end moves in the opposite adjustment direction, so that the elastic torsion accumulated inside the flexible transmission component is gradually released, and the relative displacement change between the drive end and the output end is obtained during the adjustment process; Step S33: Extract the stable correspondence of the flexible transmission component after the residual torsion is released based on the relative displacement change, and associate the position of the drive end and the position of the output end after the release to form the joint initial position reference; Step S34: Based on the initial joint position reference and the corresponding driving torque state, generate the initial control reference when the flexible joint executes the action command.

5. The control method for flexible joints of humanoid robots according to claim 4, characterized in that, Step S32 includes: Based on the torsion release compensation displacement control, the drive end generates reverse rotation, so that the input displacement of the drive end is opposite to the current torsion direction of the flexible transmission component; The driving angular displacement during the reverse motion of the driving end and the corresponding response angular displacement of the output end are obtained, and the torsional release change of the flexible transmission component is extracted based on the angular displacement difference between the driving angular displacement and the response angular displacement. Based on the change process of the torsional release, the displacement response characteristics of the flexible transmission component during the transition from the elastic energy storage state to the stable transmission state are extracted. Based on the correlation between the positional changes of the driving end and the positional changes of the output end according to the displacement response characteristics, the relative displacement change between the driving end and the output end during the adjustment process is obtained.

6. The control method for flexible joints of humanoid robots according to claim 5, characterized in that, Before the reverse rotation is generated by the drive end based on the torsional release compensation displacement control, the following steps are also included: Obtain the torsional direction corresponding to the preload offset of the flexible transmission component and the initial force state of the drive end, and determine the torque release direction during the reverse adjustment process based on the torsional direction and the force state; Based on the torque release direction, a slow-release driving force smaller than the target reverse adjustment torque is applied to the drive end, so that a progressive torque transmission state is formed between the drive end and the flexible transmission component. The correlation features between the displacement change at the drive end and the response change at the output end under the action of slow-release driving force are obtained, and the start-up response parameters of the flexible transmission component when it transitions from a stopped state to a releaseable state are extracted. Adjust the timing of applying the torsional release compensation displacement according to the start-up response parameters, so that the drive end performs reverse rotation in the elastic torsional release direction.

7. The control method for flexible joints of humanoid robots according to claim 6, characterized in that, Based on the torque release direction, a slow-release driving force smaller than the target reverse adjustment torque is applied to the drive end, so that a gradual torque transmission state is formed between the drive end and the flexible transmission component. Specifically: Obtain the current torsional force state at the connection between the drive end and the flexible transmission component, and determine the loading direction of the mitigating drive force based on the torsional force state; The control drive end applies a gradually increasing slow-release driving force along the torque release direction, causing the drive end to displace and drive the release of the elastic energy stored inside the flexible transmission component; The difference between the displacement change at the drive end and the response change at the output end during the loading of the slow-release driving force is collected, and the degree of elastic deformation release inside the flexible transmission component is extracted based on the difference. The loading amplitude of the slow-release driving force is adjusted based on the degree of elastic deformation release, so that the input torque at the driving end is gradually transmitted to the output end.

8. The control method for flexible joints of humanoid robots according to claim 7, characterized in that, The loading amplitude of the slow-release driving force is adjusted based on the degree of elastic deformation release, including: Based on the release stage corresponding to the degree of elastic deformation release, the current slow-release driving force applied by the driving end is obtained, and the slow-release driving force is adjusted incrementally according to the preset increment; When the flexible transmission component is in the initial stage of elastic deformation release, the slow-release driving force is controlled within the range of 10% to 30% of the target reverse adjustment torque, so that the input torque at the drive end is preferentially applied to the elastic recovery process of the flexible transmission component. When the elastic deformation release reaches 70% to 90%, the slow-release driving force is increased by 0.02 N·m to 0.05 N·m each time, so that the input torque is gradually transmitted to the output end; The change in angular displacement between the drive end and the output end is obtained. When the change in angular displacement is stable within the range of 0.05° to 0.15°, the slow-release driving force is increased to 50% to 80% of the target reverse adjustment torque, thus forming a stable torque transmission state between the drive end and the output end.

9. The control method for flexible joints of humanoid robots according to claim 1, characterized in that, Step S4 includes the following steps: The initial drive torque input is determined based on the initial control reference when the flexible joint executes the action command, and the drive component is controlled to drive the flexible joint to move according to the initial drive torque input. The actual angular displacement of the flexible joint output end and the corresponding change in driving torque are obtained during the execution of the motion command, and the torsional response change of the flexible transmission component as the motion is advanced is extracted. Based on the offset relationship between the torsional response change and the initial control reference, the torque adjustment amount for the corresponding action stage is generated; The subsequent drive torque input is corrected based on the torque adjustment amount, so that the output end of the flexible joint maintains the motion state corresponding to the action command.

10. A control system for a flexible joint of a humanoid robot, characterized in that, For performing the control method for a flexible joint of a humanoid robot as described in claim 1, the control system for the flexible joint of the humanoid robot includes: The confirmation module is used to determine the residual torsional state of the flexible transmission component based on the angle difference between the drive end position and the output end position of the flexible joint in the stopped state. The judgment module is used to determine whether there is a preload offset that affects the initial response before the joint starts by using the residual torsion state when the flexible joint receives a motion command. The adjustment module is used to control the drive end to perform an adjustment action opposite to the target motion direction when there is preload offset, so that the flexible transmission component releases the residual torsion and establishes the initial control reference of the flexible joint. The control module is used to control the flexible joint to execute motion commands based on the initial control reference, and to correct and adjust the torque adjustment amount of the motion according to the output response during the operation of the motion command.

Images