Motor Torque Dynamic Adjustment Method and System Based on Analog Input
By obtaining the analog input signal set of servo motors, analyzing the torque adjustment parameters and generating a segmented torque adjustment strategy, the problem that traditional motor torque control methods cannot respond to external demands in real time is solved, and precise dynamic adjustment of motor torque is achieved, which improves the operating stability and adaptability of the equipment.
Patent Information
- Application Number
- CN202510369128.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-27
- Publication Date
- 2025-07-01
- Estimated Expiration
- 2045-03-27
AI Technical Summary
Traditional motor torque control methods cannot perceive changes in the external environment and the diversified torque adjustment requirements of external equipment in real time, resulting in a large deviation from the actual demand of the motor output torque, affecting the operating efficiency and stability of the equipment.
By obtaining the analog input signal set of the servo motor, analyzing the torque adjustment parameter set, combining the current state parameters of the motor, a segmented torque adjustment strategy is generated, and the torque of the servo motor is dynamically adjusted to meet the torque change rate threshold constraint.
It realizes accurate and flexible adjustment of motor torque, improves the stability and adaptability of motor operation, reduces energy consumption, and extends the service life of the equipment.
Smart Images

Figure CN119891881B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of motor control. Specifically, it relates to a method and system for dynamically adjusting the torque of a motor based on analog input. Background Art
[0002] In the current technical field of motor control, the precise control of motor torque is crucial for the stable operation of numerous industrial production processes and complex mechatronic systems. With the continuous improvement of industrial automation levels and the widespread application of various complex devices, higher requirements are put forward for the accuracy, flexibility, and adaptability of motor torque control.
[0003] Most traditional motor torque control methods adopt relatively simple and fixed modes. Some early methods only control the motor torque based on fixed parameter settings and cannot real-time sense the changes in the external environment and the diverse torque adjustment requirements of external devices. When facing the frequent and complex torque adjustment requirements of external devices, these methods often seem powerless and are difficult to make timely and accurate responses, resulting in a large deviation between the motor output torque and the actual demand, thereby affecting the operation efficiency and stability of the entire device. Summary of the Invention
[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide a method for dynamically adjusting the torque of a motor based on analog input, and the method includes:
[0005] Obtain a set of analog input signals received by the servo motor in the operating state, where the set of analog input signals includes at least one analog input signal, and the analog input signal is used to represent the torque adjustment requirement of the external device for the servo motor;
[0006] Perform signal analysis on the set of analog input signals to obtain a set of torque adjustment parameters corresponding to each analog input signal, where the set of torque adjustment parameters includes a target torque value, a torque change rate threshold, and a torque response time window;
[0007] Based on the set of torque adjustment parameters and the current operating state parameters of the servo motor, determine the current torque value and the torque dynamic response characteristics of the servo motor, where the current operating state parameters include the motor winding current sampling value and the motor rotor speed detection value;
[0008] Generate a torque dynamic adjustment strategy for the servo motor according to the set of torque adjustment parameters, the current torque value, and the torque dynamic response characteristics, where the torque dynamic adjustment strategy includes a segmented torque adjustment instruction sequence;
[0009] Execute the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy, so that the output torque of the servo motor dynamically matches the target torque value according to the preset time window, while satisfying the torque change rate threshold constraint.
[0010] On the other hand, an embodiment of the present invention further provides a motor torque dynamic adjustment system based on analog input, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.
[0011] Based on the above aspects, the embodiment of the present application obtains a set of analog input signals, and through in-depth analysis of the set of analog input signals, obtains a set of torque adjustment parameters, which not only includes the target torque value, but also introduces two key parameters, namely the torque change rate threshold and the torque response time window. In the torque determination link, based on the set of torque adjustment parameters and the current operating state parameters of the servo motor (the sampled value of the motor winding current and the detected value of the motor rotor speed), the current torque value and the torque dynamic response characteristics of the servo motor are determined, fully considering the real-time operating state of the motor, and can more accurately grasp the current working condition of the motor. In terms of strategy generation, according to the set of torque adjustment parameters, the current torque value and the torque dynamic response characteristics, a torque dynamic adjustment strategy including a segmented torque adjustment instruction sequence is generated, which can flexibly adjust the torque according to the requirements of different stages and the motor state, realizing more refined and intelligent torque dynamic adjustment. This segmented strategy can dynamically match the target torque value for different time windows and motor operating states, while strictly satisfying the torque change rate threshold constraint, avoiding the impact on the motor and equipment caused by torque mutation, and significantly improving the stability and reliability of the motor operation. Finally, in the actual execution process, execute the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy, so that the output torque of the servo motor can accurately dynamically match the target torque value within the preset time window, while strictly satisfying the torque change rate threshold constraint, not only improving the accuracy and response speed of the motor torque control, but also greatly enhancing the adaptability of the motor under different loads and working conditions, effectively improving the overall performance of the motor drive system, reducing energy consumption, and prolonging the service life of the equipment. Description of the Drawings
[0012] Figure 1 is a schematic execution flow diagram of the motor torque dynamic adjustment method based on analog input provided by the embodiment of the present invention.
[0013] Figure 2 is a schematic diagram of an exemplary hardware and software component of the motor torque dynamic adjustment system based on analog input provided by the embodiment of the present invention. Specific Embodiment
[0014] The present invention will be specifically described below with reference to the accompanying drawings of the specification. Figure 1 FIG. is a schematic flowchart of a method for dynamically adjusting the torque of a motor based on analog input provided by an embodiment of the present invention. The method for dynamically adjusting the torque of a motor based on analog input will be introduced in detail below.
[0015] Step S110: Obtain a set of analog input signals received by the servo motor in the operating state. The set of analog input signals includes at least one analog input signal, and the analog input signal is used to represent the torque adjustment requirement of an external device for the servo motor.
[0016] Taking the application scenario of a quadruped robot as an example, during the movement of the quadruped robot, the servo motors of its respective joints need to adjust the torque according to different task requirements. For example, when the quadruped robot is climbing a slope, the servo motors of the leg joints need a greater torque to support the weight of the robot and push it upward. At this time, an external control system (such as the central controller of the robot) will send an analog input signal to the servo motor. The set of these analog input signals may include multiple analog input signals, and each analog input signal represents a specific requirement for the torque adjustment of the servo motor.
[0017] Suppose a leg of the quadruped robot is performing a lifting action. To achieve a smooth lift, the servo motor at the joint needs to gradually increase the torque. During this process, the central controller will send an analog input signal to the servo motor according to the information obtained by the robot's attitude sensors (such as gyroscopes, accelerometers, etc.). These analog input signals may include information related to the lifting angle, speed, and overall center-of-gravity adjustment of the robot, jointly constituting the set of analog input signals.
[0018] For another example, when the quadruped robot needs to quickly turn, the servo motor at the turning joint needs to quickly adjust the torque to change the direction of movement. At this time, an external device (such as the motion planning module of the robot) will send corresponding analog input signals to the servo motor according to the turning angle, speed, and the real-time motion state of the robot. These analog input signals contain the torque adjustment information required for turning and, together with the servo motor signals of other joints, form the entire set of analog input signals.
[0019] Step S120: Parse the set of analog input signals to obtain a set of torque adjustment parameters corresponding to each analog input signal. The set of torque adjustment parameters includes a target torque value, a torque change rate threshold, and a torque response time window.
[0020] For the analog input signals received by the leg joint servo motors of a quadruped robot, frame processing is first performed. For example, within one cycle of the robot's movement, the continuously received analog signals are divided into multiple signal frame segments in chronological order. Each signal frame segment contains the sampled values of the analog signals within a relatively short continuous time window.
[0021] Taking the knee joint servo motor of a quadruped robot as an example, during one action cycle of lifting - stepping - lowering the leg, the received analog input signals are framed. The sampled values of the analog signals within each frame segment reflect the changing demands for knee joint torque adjustment during that time period.
[0022] Then, feature extraction is performed on each signal frame segment. For the signal amplitude feature, it may reflect the magnitude of the required torque adjustment within that frame segment. For example, if the signal amplitude is large, it may mean that during that time period, the knee joint requires a large torque change to adapt to the movement of the robot's leg. The signal frequency feature may be related to the rhythm of the robot's movement. For example, a faster movement rhythm may correspond to a higher signal frequency. The signal duty cycle feature may be related to the duty cycle of the torque adjustment. For example, in some intermittent motion controls, the duty cycle feature can reflect the proportion of the effective action time of the torque in different time periods.
[0023] These signal amplitude features, signal frequency features, and signal duty cycle features are input into a predefined parameter mapping model, which is obtained through a large amount of quadruped robot motion experiment data. These quadruped robot motion experiment data contain various analog signal samples and corresponding torque adjustment parameter labels. For example, during multiple robot actions such as climbing slopes, crossing obstacles, and turning, the mapping relationship is established between the sampled analog signals and the actual required torque adjustment parameters (such as target torque values, torque change rate thresholds, and torque response time windows) of joint servo motors such as the knee joint and hip joint.
[0024] Through the parameter mapping model, a subset of candidate torque adjustment parameters corresponding to each signal frame segment can be output. Then, the subsets of candidate torque adjustment parameters corresponding to multiple signal frame segments are aligned in time series and parameter fusion is performed. For example, for the knee joint servo motor, during the entire lifting - stepping - lowering action cycle, the subsets of candidate torque adjustment parameters for different frame segments need to be arranged in chronological order. There may be a continuous change interval of parameters between adjacent frame segments. Within this continuous change interval, weighted smoothing is performed on the target torque value and the torque change rate threshold to eliminate parameter jumps that may be caused by differences between different frame segments. For the subsets of candidate torque adjustment parameters in the discontinuous change interval, a preset interpolation algorithm is used to supplement the intermediate parameter values. Finally, the processed subsets of candidate torque adjustment parameters are merged along the time axis to generate a complete set of torque adjustment parameters, which accurately reflects the torque adjustment requirements of the knee joint servo motor during the entire action cycle.
[0025] Step S130, based on the set of torque adjustment parameters and the current operating state parameters of the servo motor, determine the current torque value and the torque dynamic response characteristics of the servo motor, where the current operating state parameters include the sampled value of the motor winding current and the detected value of the motor rotor speed.
[0026] During the operation of the hip joint servo motor of a quadruped robot, first obtain the sampled value of the motor winding current. Since there are factors such as environmental interference (such as vibrations caused by uneven ground) and electromagnetic interference of the motor itself during robot movement, the original sampled value of the motor winding current may contain noise. A sliding window mean filter is used to preliminarily denoise the original current sampled value. For example, within a short period of time when the robot is walking, the sliding window mean of the current sampled values during this period is calculated to remove some random noise. Then, wavelet transform analysis is performed on the denoised current value to decompose the current signal into high - frequency noise components and low - frequency effective signals, and the high - frequency noise is separated and removed. According to the rated current range of the servo motor, abnormal current values outside the reasonable interval are truncated and replaced. For example, if the current value exceeds the rated current range due to momentary overload or interference, it is replaced with a reasonable value within the rated current range. Then, phase compensation is performed on the processed current signal to eliminate signal distortion caused by sampling delay and make the phase of the current signal consistent with the actual situation. Finally, the compensated current signal is synchronously calibrated with the real - time position signal of the motor rotor to obtain a stable current value.
[0027] At the same time, obtain the detected value of the motor rotor speed. During the hip joint movement of a quadruped robot, the speed of the motor rotor is constantly changing. The rotational speed differential calculation is performed on the detected value of the motor rotor speed to obtain the rotor acceleration value, which reflects the acceleration or deceleration of the hip joint movement.
[0028] According to the obtained stable current value and rotor acceleration value, query the predefined torque mapping relation table, which is established based on the physical characteristics and kinematic model of the hip servo motor of the quadruped robot. For example, under different combinations of stable current values and rotor acceleration values, there correspond different hip torque values. By querying this torque mapping relation table, determine the current torque value of the servo motor.
[0029] Based on the change trend of the current torque value within a preset time interval, calculate the torque change rate and the torque fluctuation amplitude. During the walking process of the quadruped robot, the torque of the hip joint fluctuates with the movement of the leg. For example, the torque may gradually decrease when lifting the leg and gradually increase when putting down the leg. By analyzing the change of the torque value over a period of time, calculate the torque change rate and the torque fluctuation amplitude. According to the torque change rate and the torque fluctuation amplitude, determine the torque dynamic response characteristics of the servo motor, which include the maximum allowable adjustment rate and the minimum stable time threshold. For example, if the torque fluctuation amplitude is large, in order to ensure the stability of the robot's movement, the maximum allowable adjustment rate may be relatively small, and at the same time, the minimum stable time threshold may be relatively long to avoid the robot losing balance due to too fast torque adjustment.
[0030] Step S140, generate a torque dynamic adjustment strategy for the servo motor according to the torque adjustment parameter set, the current torque value, and the torque dynamic response characteristics, where the torque dynamic adjustment strategy includes a segmented torque adjustment instruction sequence.
[0031] Taking the ankle servo motor of the quadruped robot as an example, first calculate the difference between the target torque value and the current torque value. Suppose the robot is walking from a flat ground onto a slope, and the ankle needs to increase the torque to adapt to the slope of the slope. Calculate the difference between the target torque value (the torque required to adapt to the slope) and the current torque value (the torque when walking on the flat ground). Determine the torque adjustment direction according to this difference, which is the positive direction here, that is, the torque needs to be increased.
[0032] Based on the maximum allowable adjustment rate and the minimum stable time threshold in the torque dynamic response characteristics, multiple torque adjustment stages are divided. For example, the total adjustment amount is determined according to the absolute value of the difference between the target torque value and the current torque value. If the total adjustment amount is large, considering the torque dynamic response characteristics of the ankle servo motor, the theoretical minimum adjustment time is calculated. According to the ratio of the theoretical minimum adjustment time to the torque response time window, the number of stage divisions is determined. In this process, a non-uniform time allocation strategy is adopted. For example, a shorter time window is allocated to the initial stage so that the ankle can quickly approach the target value to adapt to the slope change of the ramp as soon as possible. A longer time window is allocated to the final stage to achieve precise and stable torque adjustment and ensure the stability of the robot walking on the ramp. A transition buffer is set inside each stage to absorb the torque fluctuations caused by the load mutations due to obstacles such as small stones that the robot may encounter when walking on the ramp.
[0033] According to the torque change rate threshold constraint, a corresponding torque adjustment slope is allocated to each adjustment stage, and this torque adjustment slope is used to control the acceleration of the torque change within this stage. For example, in the initial stage, since the time window is short, a larger torque adjustment slope can be set so that the torque can increase rapidly; in the intermediate stage, according to the torque change rate threshold constraint and the requirement of the robot motion smoothness, a moderate torque adjustment slope is set; in the final stage, in order to reach the target torque value precisely and stably, a smaller torque adjustment slope is set.
[0034] The stage target torque value, adjustment time window, and torque adjustment slope of each adjustment stage are combined to generate a segmented torque adjustment instruction sequence. For example, in the torque adjustment instruction sequence of the ankle servo motor, the instruction in the initial stage may include an instruction to quickly increase to an intermediate target torque value within a short time window, the intermediate stage includes an instruction to adjust to another intermediate target torque value at a certain slope within a moderate time window, and the final stage includes an instruction to slowly adjust to the target torque value suitable for the ramp within a long time window. The timing of this segmented torque adjustment instruction sequence is verified to ensure that the change in the torque adjustment slope between adjacent stages does not exceed the preset connection threshold to ensure the smoothness of the torque adjustment process.
[0035] Step S150, execute the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy, so that the output torque of the servo motor dynamically matches the target torque value according to the preset time window, while satisfying the torque change rate threshold constraint.
[0036] For the elbow joint servo motor of a quadruped robot, a first adjustment instruction is sent to the torque control module of the servo motor according to the first-stage instruction in the segmented torque adjustment instruction sequence. For example, when the robot is performing a grasping action, the first-stage target torque value may be the torque required to bend the elbow joint to an initial angle, and the corresponding adjustment time window is a relatively short time.
[0037] Within the first-stage adjustment time window, the actual torque change curve of the servo motor is monitored in real time, and the actual torque change rate is calculated. Suppose that during the bending of the elbow joint, due to the friction of the mechanical structure or the slight change of the load, the actual torque change rate may deviate from the torque adjustment slope in the first-stage instruction. When it is detected that the actual torque change rate deviates from the torque adjustment slope in the first-stage instruction, the adjustment time window or the target torque value of the subsequent stage is dynamically corrected according to the deviation amplitude.
[0038] For example, if the calculated deviation percentage between the actual torque change rate and the required rate of the instruction is less than the first threshold, only the torque adjustment slope of the subsequent stage is finely adjusted. If the deviation percentage is between the first threshold and the second threshold, the time window of the subsequent stage is extended proportionally. For example, the original planned time window for a certain intermediate stage is 5 seconds and is extended to 6 seconds due to the deviation. If the deviation percentage exceeds the second threshold, the remaining adjustment amount is recalculated and a new stage division scheme is generated. The timestamp mark of the original instruction is retained during the correction process to ensure the traceability of the corrected instruction sequence.
[0039] After completing the adjustment of the current stage, the next-stage adjustment instruction is triggered according to the corrected segmented torque adjustment instruction sequence until all stage instructions are executed. After the final-stage adjustment is completed, the output torque stability of the servo motor is continuously monitored. For example, after the robot completes the grasping action, the elbow joint needs to maintain a certain torque to stably hold the grasped object. If it is detected that the torque fluctuation exceeds the allowable range, a local adjustment instruction is regenerated and inserted into the execution queue to ensure that the elbow joint can stably maintain the grasping state and guarantee the accuracy and stability of the entire operation of the quadruped robot.
[0040] During the movement of the quadruped robot, this torque dynamic adjustment mechanism of the servo motor ensures that the robot joints can accurately and stably adjust the torque according to external demands and internal states, so as to achieve efficient and smooth movement. Whether walking on different terrains, performing various task actions or coping with external disturbances, the performance and reliability of the robot can be guaranteed.
[0041] Based on the above steps, the embodiments of the present application obtain a set of analog input signals. By deeply analyzing the set of analog input signals, a set of torque adjustment parameters is obtained, which not only includes the target torque value, but also introduces two key parameters, namely the torque change rate threshold and the torque response time window. In the torque determination stage, based on the set of torque adjustment parameters and the current operating state parameters of the servo motor (the sampled value of the motor winding current and the detected value of the motor rotor speed), the current torque value and the torque dynamic response characteristics of the servo motor are determined, fully considering the real-time operating state of the motor, and being able to more accurately grasp the current working condition of the motor. In terms of strategy generation, according to the set of torque adjustment parameters, the current torque value and the torque dynamic response characteristics, a torque dynamic adjustment strategy including a segmented torque adjustment instruction sequence is generated, which can flexibly adjust the torque according to the requirements and motor states in different stages, realizing a more refined and intelligent torque dynamic adjustment. This segmented strategy can dynamically match the target torque value for different time windows and motor operating states, while strictly meeting the constraint of the torque change rate threshold, avoiding the impact on the motor and equipment caused by torque mutation, and significantly improving the stability and reliability of the motor operation. Finally, in the actual execution process, the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy is executed, so that the output torque of the servo motor can accurately and dynamically match the target torque value within the preset time window, while strictly meeting the constraint of the torque change rate threshold, not only improving the accuracy and response speed of the motor torque control, but also greatly enhancing the adaptability of the motor under different loads and working conditions, being able to effectively improve the overall performance of the motor drive system, reduce energy consumption, and extend the service life of the equipment.
[0042] In a possible implementation manner, step S120 includes:
[0043] Step S121, performing frame processing on each analog input signal to obtain a plurality of signal frame segments, and each signal frame segment includes the sampled values of the analog signals within a continuous time window.
[0044] When the quadruped robot performs actions such as crossing an obstacle, the servo motors at the leg joints receive a set of analog input signals from an external device, and this analog input signal is intended to adjust the torque of the servo motor to achieve the precise movement of the leg joints of the quadruped robot.
[0045] First, taking the knee joint servo motor of a quadruped robot as an example, during the action of crossing an obstacle, the received analog input signal is segmented in chronological order. Each segmented signal frame segment covers the analog signal sampling values within a continuous time window, and the determination of this continuous time window is based on the key action stages during the movement of the quadruped robot, such as different stages from preparing to cross the obstacle to lifting the leg, crossing, and then putting it down. During the leg-lifting stage, the analog signal sampling values within the signal frame segment reflect the change in the demand for knee joint torque adjustment during this specific time period.
[0046] Step S122: Extract features from each signal frame segment to obtain the signal amplitude feature, signal frequency feature, and signal duty cycle feature corresponding to each signal frame segment.
[0047] For example, regarding the signal amplitude feature, when a quadruped robot crosses an obstacle, the knee joint requires different levels of torque to support the body weight and complete the crossing action. If the signal amplitude is large within a certain signal frame segment, it indicates that a larger torque adjustment is required for the knee joint during this time period. The signal frequency feature is related to the movement rhythm of the quadruped robot. During the relatively complex action of crossing an obstacle, the movement rhythm of the robot will change, and this change will be reflected in the signal frequency. For example, when the leg quickly lifts to cross the obstacle, the signal frequency may increase. The signal duty cycle feature is related to the duty cycle of torque adjustment. During the process of crossing an obstacle, the knee joint may need to intermittently increase or decrease the torque, and the duty cycle feature can reflect the effective action time ratio of the torque in different time periods.
[0048] Step S123: Input the signal amplitude feature, signal frequency feature, and signal duty cycle feature into a predefined parameter mapping model, and output a candidate torque adjustment parameter subset corresponding to each signal frame segment through the parameter mapping model.
[0049] Specifically, the parameter mapping model is established based on a large amount of experimental data of quadruped robots crossing different types of obstacles. In these experiments, various analog signal samples are collected, and at the same time, the corresponding torque adjustment parameter labels are recorded. These torque adjustment parameter labels contain information such as the target torque value, torque change rate threshold, and torque response time window. The analog signal samples of the knee joint servo motor when crossing obstacles of different heights and different shapes, as well as the corresponding torque adjustment parameters, are all used to construct this parameter mapping model. Through this parameter mapping model, a candidate torque adjustment parameter subset corresponding to each signal frame segment is output.
[0050] Step S124: Align the time series and fuse the parameters of the candidate torque adjustment parameter subsets corresponding to multiple signal segments to generate a torque adjustment parameter set corresponding to the analog input signal. Among them, the parameter mapping model is trained through the mapping relationship between multiple groups of analog signal samples and torque adjustment parameter labels.
[0051] Specifically, during the entire process of a quadruped robot crossing an obstacle, the candidate torque adjustment parameter subsets corresponding to different signal segments need to be arranged in chronological order. Taking the knee joint servo motor as an example, during the entire motion cycle of crossing an obstacle, from preparing to cross to completing the crossing, the candidate torque adjustment parameter subsets of each signal segment must be arranged in the correct time sequence. There may be a continuous change interval of parameters between adjacent signal segments, and within this continuous change interval, weighted smoothing is performed on the target torque value and the torque change rate threshold. For example, when transitioning from the leg-lifting stage to the crossing stage, in order to avoid sudden changes in torque adjustment, weighted smoothing is performed on the target torque value and the torque change rate threshold to eliminate parameter jumps that may be caused by differences between different segments. For the candidate torque adjustment parameter subsets in the discontinuous change interval, a preset interpolation algorithm is used to supplement the intermediate parameter values. For example, during the obstacle-crossing process, if a certain signal segment is missing or some parameters are inaccurate due to interference, the interpolation algorithm is used to supplement them. The processed candidate torque adjustment parameter subsets are merged along the time axis, and finally a complete torque adjustment parameter set is generated, which accurately reflects the torque adjustment requirements of the knee joint servo motor during the obstacle-crossing process of the quadruped robot.
[0052] In a possible implementation manner, step S130 includes:
[0053] Step S131: Obtain the motor winding current sampling value of the servo motor, and perform filtering processing on the motor winding current sampling value to obtain a stable current value.
[0054] When a quadruped robot walks on rough terrain, the servo motors at the leg joints are in operation. When the servo motors at the leg joints of the quadruped robot are working, the motor winding current will fluctuate due to various factors. For example, when the leg joints of the robot walk on rough terrain, due to the unevenness of the ground, the force on the joints changes continuously, which will cause changes in the motor load, thereby changing the motor winding current. The original sampled value of the motor winding current contains various interference factors, such as electromagnetic interference and noise caused by the vibration of the robot itself. In order to obtain an accurate current value for determining torque, it is necessary to filter the sampled value of the motor winding current. A sliding window mean filter is used to preliminarily denoise the original current sampled value. For example, within a short time interval, the mean value of the current sampled values within this time interval is calculated to smooth the current signal and remove some random small fluctuation noises. Then, wavelet transform analysis is performed on the denoised current value. Through wavelet transform, the current signal is decomposed into high-frequency noise components and low-frequency effective signals, and the high-frequency noise is separated and removed, thereby further improving the purity of the current signal. Then, according to the rated current range of the servo motor, the abnormal current values outside the reasonable range are truncated and replaced. For example, if the current suddenly exceeds the rated current range due to the robot being suddenly impacted by a large external force, such abnormal values may affect the accurate calculation of torque, so they are replaced with reasonable values within the rated current range. After that, phase compensation is performed on the processed current signal. Since there may be a delay during the sampling process, which will cause the phase of the current signal to not match the actual situation, the signal distortion caused by the sampling delay can be eliminated through phase compensation. Finally, the compensated current signal is synchronously calibrated with the real-time position signal of the motor rotor to obtain a stable current value.
[0055] Step S132: Obtain the detected value of the rotational speed of the motor rotor of the servo motor, and perform rotational speed differentiation calculation on the detected value of the rotational speed of the motor rotor to obtain the rotor acceleration value.
[0056] For example, when the leg joint of the robot quickly drops from the lifted state, the rotational speed of the motor rotor will increase rapidly; while when the joint bends slowly, the rotational speed of the motor rotor will slow down accordingly. Perform rotational speed differentiation calculation on the detected value of the rotational speed of the motor rotor to obtain the rotor acceleration value. This rotor acceleration value is very important for understanding the dynamic characteristics of joint movement. For example, when a quadruped robot crosses a small earthen slope, during the upward lifting process of the leg joint, the rotational speed of the motor rotor gradually increases, and the rotor acceleration value obtained through rotational speed differentiation calculation is positive, indicating that the joint is in an accelerating motion state; when the leg joint passes over the apex of the slope and starts to drop, the rotational speed of the motor rotor gradually decreases, and the rotor acceleration value is negative, indicating that the joint is in a decelerating motion state.
[0057] Step S133: Query the predefined torque mapping relation table according to the stable current value and the rotor acceleration value to determine the current torque value of the servo motor.
[0058] The torque mapping relation table is constructed based on the physical characteristics and kinematic model of the servo motor of the quadruped robot's leg joint. Under different combinations of stable current values and rotor acceleration values, there correspond different torque values. For example, when the quadruped robot walks on flat ground, the stable current value and rotor acceleration value of the leg joint are within a relatively stable range, and the corresponding torque value can maintain the normal walking of the robot; while when the robot climbs a slope, the stable current value and rotor acceleration value change. By querying the torque mapping relation table, the current torque value of the leg joint servo motor at this time can be determined, and this current torque value can meet the power requirements of the leg joint when the robot climbs the slope.
[0059] Step S134: Calculate the torque change rate and the torque fluctuation amplitude based on the change trend of the current torque value within a preset time interval.
[0060] During the walking process of the quadruped robot, the torque of the leg joint changes with the movement posture of the robot and the terrain. For example, when the quadruped robot walks on uneven ground, the leg joint needs to continuously adjust the torque to adapt to the undulations of the terrain. Analyze the change of the current torque value of the leg joint servo motor within a specific preset time interval, such as the time interval from when one foot of the robot touches a raised stone to when it completely crosses the stone. If the torque value changes frequently and significantly within this interval, then the torque change rate will be larger, and at the same time the torque fluctuation amplitude will also be larger. This may be caused by the changing force conditions of the leg joint due to the complexity of the terrain.
[0061] Step S135: Determine the torque dynamic response characteristics of the servo motor according to the torque change rate and the torque fluctuation amplitude. The torque dynamic response characteristics include the maximum allowable adjustment rate and the minimum stable time threshold.
[0062] For example, if the torque fluctuation amplitude is large during the above-mentioned stone crossing process, in order to ensure the stability and safety of the quadruped robot, the maximum allowable adjustment rate may be relatively small, because a large torque adjustment rate may cause the unstable movement of the robot joint. At the same time, the minimum stable time threshold may be relatively long to ensure that after the torque adjustment, the leg joint has enough time to stabilize, avoiding the shaking or imbalance of the robot caused by too fast torque adjustment.
[0063] In a possible implementation manner, step S140 includes:
[0064] Step S141: Calculate the difference between the target torque value and the current torque value, and determine the torque adjustment direction based on the difference.
[0065] For example, when a quadruped robot needs to walk from a flat ground onto a slope, the target torque value at the leg joints will be greater than the current torque value, and this difference is positive. Therefore, the torque adjustment direction is the direction of increasing torque.
[0066] Step S142: Based on the maximum allowable adjustment rate and the minimum stable time threshold in the torque dynamic response characteristics, divide multiple torque adjustment stages, and each stage corresponds to an adjustment time window and a stage target torque value.
[0067] Taking the quadruped robot climbing a slope as an example, determine the total adjustment amount according to the absolute value of the difference between the target torque value and the current torque value. If the total adjustment amount is large, considering the torque dynamic response characteristics of the leg joint servo motor, calculate the theoretical minimum adjustment time. Determine the number of stage divisions according to the ratio of the theoretical minimum adjustment time to the torque response time window. In this process, adopt a non-uniform time allocation strategy. For example, allocate a shorter time window for the initial stage to enable the leg joints to quickly approach the target value and adapt to the slope change of the slope as soon as possible. In the initial stage of climbing the slope, the leg joints need to quickly increase the torque to provide sufficient power, so the adjustment time window for the initial stage is shorter, and the target torque value is set to a value that allows the robot to start climbing the slope smoothly. Allocate a longer time window for the final stage to achieve precise and stable torque adjustment and ensure the stability of the robot walking on the slope. In the final stage of climbing the slope, the leg joints need to precisely reach the target torque value adapted to the slope gradient and maintain stability, so the adjustment time window for the final stage is longer. Set up a transition buffer zone within each stage to absorb the torque fluctuations caused by sudden load changes due to obstacles such as small stones that the robot may encounter when walking on the slope.
[0068] Step S143: According to the torque change rate threshold constraint, allocate a corresponding torque adjustment slope for each adjustment stage, and the torque adjustment slope is used to control the acceleration of torque change within this stage.
[0069] Specifically, the torque adjustment slope is used to control the acceleration of torque change within this stage. During the process of a quadruped robot climbing a slope, for the initial stage, since the time window is short, a larger torque adjustment slope can be set to enable the torque to increase quickly; in the middle stage, according to the torque change rate threshold constraint and the requirement of the robot's movement smoothness, set a moderate torque adjustment slope; in the final stage, to precisely and stably reach the target torque value, set a smaller torque adjustment slope. For example, in the middle stage of climbing the slope, the leg joints need to gradually increase the torque without affecting the overall smoothness of the robot, so setting a moderate torque adjustment slope can meet this requirement.
[0070] Step S144: Combine the stage target torque values, adjustment time windows, and torque adjustment slopes for each adjustment stage to generate a segmented torque adjustment instruction sequence.
[0071] For example, in the torque adjustment instruction sequence of the leg joint servo motor of a quadruped robot, the instructions in the initial stage may include an instruction to rapidly increase to an intermediate target torque value within a short time window, the intermediate stage includes an instruction to adjust to another intermediate target torque value at a certain slope within a moderate time window, and the final stage includes an instruction to slowly adjust to the target torque value adapting to the slope within a long time window.
[0072] Step S145: Perform timing verification on the segmented torque adjustment instruction sequence to ensure that the change in the torque adjustment slope between adjacent stages does not exceed a preset connection threshold.
[0073] For example, during the climbing process of a quadruped robot, the change in the torque adjustment slope from the initial stage to the intermediate stage and from the intermediate stage to the final stage needs to have a smooth transition. If the change in the torque adjustment slope between adjacent stages is too large, it may cause the movement of the leg joints to be unstable, thereby affecting the walking stability of the quadruped robot on the slope. Through timing verification, the smoothness of the entire torque adjustment process can be ensured, enabling the leg joints to accurately adjust the torque according to the predetermined strategy to adapt to the movement requirements of the quadruped robot on different terrains.
[0074] In a possible implementation manner, step S150 includes:
[0075] Step S151: Send a first adjustment instruction to the torque control module of the servo motor according to the instruction of the first stage in the segmented torque adjustment instruction sequence, where the first adjustment instruction includes the target torque value of the first stage and the corresponding adjustment time window.
[0076] When the quadruped robot is ready to climb a slope, send a first adjustment instruction to the torque control module of the servo motor according to the instruction of the first stage in the segmented torque adjustment instruction sequence. This first adjustment instruction includes the target torque value of the first stage and the corresponding adjustment time window. For example, at the beginning of the action of the quadruped robot climbing a slope, the target torque value of the first stage is pre-calculated based on factors such as the initial posture, weight of the robot, and the starting slope of the slope. This target torque value is the key to ensuring that the leg joints of the robot can start to apply sufficient power smoothly on the slope. The corresponding adjustment time window is set according to the characteristics of the power system of the quadruped robot and the requirements of the initial stage of the slope. This time window is short, aiming to enable the servo motor of the leg joints to respond quickly and make the robot adapt to the starting slope of the slope as soon as possible.
[0077] Step S152, within the first-stage adjustment time window, continuously monitor the actual torque change curve of the servo motor and calculate the actual torque change rate.
[0078] In the initial stage of the quadruped robot climbing a slope, as the servo motor starts to adjust the torque according to the first adjustment instruction, the sensor will continuously monitor the actual operating conditions of the motor. The actual torque change curve reflects the relationship between torque and time within this time window. By calculating the continuously collected torque values, the actual torque change rate can be obtained. For example, at the moment when climbing starts, since the leg joints of the robot switch from a flat ground state to a slope state, factors such as sudden changes in ground friction and minor jams in the joint mechanical structure may affect the actual torque change rate, which may not be exactly the same as the torque adjustment slope in the first-stage instruction.
[0079] Step S153, when it is detected that the actual torque change rate deviates from the torque adjustment slope in the first-stage instruction, dynamically correct the adjustment time window or the target torque value in the subsequent stage according to the deviation amplitude.
[0080] Step S154, after completing the adjustment of the current stage, trigger the next-stage adjustment instruction according to the corrected segmented torque adjustment instruction sequence until all stage instructions are executed.
[0081] During the climbing process of the quadruped robot, after the first-stage adjustment is completed, whether according to the original instruction or the corrected instruction, the next-stage adjustment instruction will be triggered according to the completion situation of this stage. For example, after the torque adjustment is completed according to the adjusted instruction in the first stage, the servo motor of the robot leg joint starts to execute the adjustment instruction in the intermediate stage, and this process will continue until all stage instructions are executed, thus ensuring that during the climbing process of the robot, the torque of the servo motor of the leg joint can be gradually adjusted according to the predetermined strategy to adapt to different slopes of the slope and the power requirements of the robot.
[0082] Step S155, after the final-stage adjustment is completed, continuously monitor the output torque stability of the servo motor. If it is detected that the torque fluctuation exceeds the allowable range, regenerate the local adjustment instruction and insert it into the execution queue.
[0083] After the quadruped robot climbs up the slope and completes the entire climbing motion, the servo motors of the leg joints enter a stable operating state. At this time, it is still necessary to continuously monitor the output torque stability of the servo motors. For example, due to some slight shaking or adjustment actions of the robot on the slope, these actions may cause torque fluctuations in the servo motors. If it is detected that the torque fluctuation exceeds the allowable range, this may affect the stable standing of the robot on the slope or subsequent actions. At this time, it is necessary to regenerate the local adjustment instructions and insert them into the execution queue. The regenerated local adjustment instructions are calculated based on the current torque fluctuation situation, the posture of the robot, and the situation of the slope. For example, if the torque fluctuation is caused by a slight shift of the center of gravity of the robot, then the local adjustment instructions may adjust the torque of the servo motor on one side of the leg joint to rebalance the center of gravity of the robot and ensure the stability of the robot on the slope. In this way, through the precise execution and dynamic correction of the torque dynamic adjustment strategy, the quadruped robot can effectively control the torque of the servo motors in various complex terrain and motion situations, ensuring the stable operation and normal work of the robot.
[0084] Among them, step S153 includes:
[0085] Step S1531, calculate the deviation percentage between the actual torque change rate and the commanded required rate.
[0086] During the climbing process of the quadruped robot, if the deviation percentage is less than the first threshold, only fine-tune the torque adjustment slope in the subsequent stage. For example, when the deviation percentage is small, it indicates that the actual situation is relatively close to the expected situation, and it may be just due to some small interference factors that cause the deviation. At this time, by fine-tuning the torque adjustment slope in the subsequent stage, the subsequent torque adjustment process can gradually return to the expected trajectory. For example, in the subsequent intermediate stage, slightly increase or decrease the torque adjustment slope to compensate for the small deviation that occurred in the first stage, ensuring the coherence and accuracy of the entire torque adjustment process.
[0087] Step S1532, when the deviation percentage is less than the first threshold, only fine-tune the torque adjustment slope in the subsequent stage.
[0088] Step S1533, when the deviation percentage is between the first threshold and the second threshold, proportionally extend the time window in the subsequent stage.
[0089] For example, during the climbing process of a quadruped robot, if in the initial stage, due to some moderate interference on the leg joints of the robot, the deviation between the actual torque change rate and the commanded rate falls within this range. At this time, to ensure the smooth progress of subsequent torque adjustment and enable the robot to climb the slope stably, the time window of the subsequent stage is extended proportionally. Suppose the original time window for the middle stage is 5 seconds, and according to the specific situation of the deviation, it is extended to 6 seconds or 7 seconds by a certain proportion. The purpose of this is to give the servo motor more time to adjust the torque to adapt to the impact brought by the deviation in the initial stage, while avoiding the instability of the robot on the slope due to too fast adjustment.
[0090] Step S1534, when the deviation percentage exceeds the second threshold, recalculate the remaining adjustment amount and generate a new stage division plan.
[0091] When a quadruped robot is climbing a slope, if the deviation in the initial stage is very large and exceeds the second threshold, this may be due to the robot encountering a large unexpected situation at the start of climbing, such as suddenly hitting a small stone on the slope or a large jam in the joints. At this time, the original torque adjustment plan may not meet the requirements of climbing, and it is necessary to recalculate the remaining adjustment amount. The recalculation of the remaining adjustment amount is based on factors such as the current actual torque value, the target torque value, and the elapsed time. Then, a new stage division plan is generated according to the recalculated result. For example, it may be necessary to increase the number of middle stages, or readjust the target torque value and time window of each stage. In this process, the new stage division plan should fully consider the stability and power requirements of the quadruped robot to ensure that the robot can continue to climb the slope smoothly.
[0092] Step S1535, during the correction process, retain the timestamp mark of the original instruction to ensure the traceability of the corrected instruction sequence.
[0093] During the entire torque adjustment process of a quadruped robot climbing a slope, whether it is fine-tuning the torque adjustment slope of the subsequent stage, extending the time window, or regenerating the stage division plan, the timestamp mark of the original instruction is retained. This timestamp mark records information such as the issuance time, execution time, and correction time of each instruction. For example, when it is necessary to analyze or troubleshoot the torque adjustment during the entire climbing process, this timestamp mark can help accurately locate the situation of each stage, understand when the deviation occurred, and what correction measures were taken. It is of great significance for optimizing the torque adjustment strategy of the quadruped robot and improving the reliability and performance of the robot.
[0094] In a possible implementation manner, step S124 includes:
[0095] Step S1241: Extract the timestamp information of each signal frame segment, and sort the candidate torque adjustment parameter subsets in the chronological order of the timestamp information.
[0096] When a quadruped robot performs complex walking actions, such as walking on rough and variable terrains (such as slopes with different gradients, potholes, and small obstacles), the analog input signals received by the servo motors of the leg joints need to be precisely processed to determine an appropriate set of torque adjustment parameters.
[0097] First, during the movement of the leg joints of the quadruped robot, the analog input signals at different times reflect different movement requirements. For example, when the robot's leg is lifted to prepare to cross a small pothole, the signal frame segment corresponding to this moment and the signal frame segment during the subsequent process of the leg crossing the pothole and being lowered have a chronological order. By extracting the timestamp information of each signal frame segment, the sequence of each signal frame segment in the entire walking action can be determined. Then, sort the candidate torque adjustment parameter subsets according to this sequence to ensure that these candidate torque adjustment parameter subsets are coherent in time.
[0098] Step S1242: Detect the overlapping regions of the candidate torque adjustment parameter subsets of adjacent signal frame segments to determine the parameter continuous change intervals.
[0099] During the walking process of the quadruped robot, there are parameter transitions between adjacent action phases. For example, when the leg transitions from the normal walking state to the action of crossing an obstacle, there is a certain overlapping region between the candidate torque adjustment parameter subsets of the signal frame segments corresponding to these two adjacent phases. By detecting this overlapping region, it can be determined that the parameters are continuously changing within this interval. This continuous change reflects the coherence of the movement of the leg joints of the quadruped robot. For example, before crossing an obstacle, the leg will gradually adjust the torque to prepare for higher load requirements, and this adjustment process is manifested as a continuous change in parameters in adjacent signal frame segments.
[0100] Step S1243: Perform weighted smoothing processing on the target torque value and the torque change rate threshold within the parameter continuous change interval to eliminate parameter jumps.
[0101] Taking the process of a quadruped robot crossing obstacles of different heights as an example, when the leg crosses from a lower obstacle to a higher obstacle, the target torque value and the torque change rate threshold need to be adjusted step by step. Without smoothing processing, parameter jumps may occur, which will lead to unstable torque adjustment of the servo motor. Through weighted smoothing processing, the target torque value and the torque change rate threshold can be smoothly transitioned within a continuous change interval. For example, in the process of crossing from a small stone (requiring less torque) to a larger stone (requiring more torque), the target torque value will gradually increase, and this increase process is smooth without sudden large changes, thus ensuring the stability and smoothness of the leg joint movement.
[0102] Step S1244, for the candidate torque adjustment parameter subset in the discontinuous change interval, use a preset interpolation algorithm to supplement the intermediate parameter values.
[0103] During the walking process of the quadruped robot, there may be some special situations that cause discontinuous change intervals between signal frame segments. For example, when the robot suddenly changes its walking direction or encounters sudden external force interference, this situation may occur. Suppose the quadruped robot is walking straight on a flat ground and suddenly turns to avoid a large obstacle. This turning action may cause a discontinuous change interval between the signal frame segments corresponding to the previous and subsequent different walking directions. At this time, use a preset interpolation algorithm to supplement the intermediate parameter values. This interpolation algorithm reasonably supplements the missing intermediate parameter values according to factors such as the parameter conditions of the previous and subsequent two signal frame segments and the time interval between them, so that the entire torque adjustment parameter set is complete on the time axis, thereby ensuring that the servo motor can perform accurate torque adjustment according to the complete parameters.
[0104] Step S1245, merge the processed candidate torque adjustment parameter subsets along the time axis to generate a complete torque adjustment parameter set.
[0105] After the quadruped robot completes the entire walking action cycle, all the processed candidate torque adjustment parameter subsets are merged along the time axis. This complete torque adjustment parameter set covers the torque adjustment requirement information of the leg joint servo motor during the entire walking process, including the target torque value, torque change rate threshold, etc. at different stages.
[0106] In a possible implementation manner, step S131 includes:
[0107] Step S1311, use a sliding window mean filter to perform preliminary denoising on the original current sampling values.
[0108] When a quadruped robot is walking, the motor winding current will be disturbed by various factors and generate noise. For example, the mechanical vibration of the robot itself, electromagnetic interference in the surrounding environment, etc. will cause fluctuations in the original current sampling value. The sliding window mean filter smooths these fluctuations by calculating the mean value of the current sampling values within a window of a specific size. Assuming the size of the sliding window is n sampling points, within a short time period of the robot's leg joint movement, the average value of the current values of these n sampling points is calculated to obtain a relatively smooth value, thereby removing some random small-amplitude noises and making the current sampling value closer to the actual current situation.
[0109] Step S1312: Perform wavelet transform analysis on the denoised current value to separate the high-frequency noise component and the low-frequency effective signal.
[0110] During the walking process of the quadruped robot, the current value after preliminary denoising may still contain some complex noise components. Wavelet transform analysis can decompose the current signal into components of different frequencies. The high-frequency noise component is usually caused by some rapidly changing interference factors, such as instantaneous electromagnetic pulses, etc.; while the low-frequency effective signal reflects the basic operating state of the motor, such as the load change related to the movement of the robot's leg joints, etc. Through wavelet transform analysis, the high-frequency noise component can be accurately separated and removed, thereby further improving the purity of the current signal.
[0111] Step S1313: According to the rated current range of the servo motor, truncate and replace the abnormal current values that exceed the reasonable range.
[0112] During the walking process of the quadruped robot, some special situations may occur that cause abnormal current values. For example, when the leg joint of the robot is suddenly subjected to a large external force impact, such as stepping into a relatively deep pothole, the motor may be overloaded, resulting in the current value exceeding the rated current range. Such abnormal current values may affect the accurate calculation and control of torque. At this time, according to the rated current range of the servo motor, the current values that exceed the reasonable range are truncated and replaced with reasonable values within the rated current range. For example, if the rated current range is [I_min, I_max], and the detected current value is greater than I_max, it is replaced with I_max to ensure that the torque value calculated based on the current value subsequently is reasonable and safe.
[0113] Step S1314: Perform phase compensation on the processed current signal to eliminate the signal distortion caused by sampling delay.
[0114] During the sampling process of the motor winding current, due to factors such as the response time of the sampling device, there may be a sampling delay. This sampling delay will cause the phase of the current signal to not match the actual situation, thereby affecting the accurate judgment of the motor operating state. For example, during the movement of the leg joints of a quadruped robot, due to the sampling delay, the phase of the current signal may lag behind the actual current change, resulting in a deviation between the calculated torque value and the actual demand. Through the phase compensation algorithm, according to factors such as the known sampling delay time, the phase of the current signal is adjusted to match the actual situation, thereby eliminating the signal distortion caused by the sampling delay.
[0115] Step S1315, synchronously calibrate the compensated current signal with the real-time position signal of the motor rotor to obtain a stable current value.
[0116] During the operation of the servo motor of the leg joints of a quadruped robot, the position of the motor rotor is closely related to the current signal. The real-time position signal of the motor rotor reflects the actual operating state of the motor. For example, when the leg joint is in different movement positions, the position of the motor rotor will change accordingly, and this change will affect the electromagnetic relationship of the motor, thereby affecting the magnitude and characteristics of the current. By synchronously calibrating the compensated current signal with the real-time position signal of the motor rotor, the current value can accurately reflect the true operating condition of the motor at the current rotor position, obtain a stable current value, and provide a reliable basis for accurately calculating the torque of the motor in the subsequent process.
[0117] In a possible implementation manner, step S142 includes:
[0118] Step S1421, determine the total adjustment amount according to the absolute value of the difference between the target torque value and the current torque value.
[0119] When a quadruped robot needs to change its walking posture, for example, from slow walking on flat ground to climbing a slope, the target torque value of the servo motor of the leg joints will change. Assume that when walking on flat ground, the current torque value of the servo motor of the leg joints is T1, and the target torque value required when climbing a slope is T2. Calculate the absolute value of the difference T2 - T1. This absolute value of the difference is the total adjustment amount, which reflects the torque adjustment amplitude required from the current state to the target state.
[0120] Step S1422, calculate the theoretical minimum adjustment time based on the ratio of the total adjustment amount to the maximum allowable adjustment rate.
[0121] During the climbing process of a quadruped robot, due to the physical characteristics of the servo motors and the overall stability requirements of the quadruped robot, there is a maximum allowable adjustment rate, which limits the speed of torque adjustment to avoid robot instability caused by too fast torque adjustment. For example, if the total adjustment amount is ΔT and the maximum allowable adjustment rate is v, then the theoretical minimum adjustment time t = ΔT / v. This theoretical minimum adjustment time is the shortest time required for torque adjustment at the maximum allowable adjustment rate under ideal conditions.
[0122] Step S1423: Determine the number of stage divisions according to the ratio of the theoretical minimum adjustment time to the torque response time window.
[0123] In the torque control strategy of a quadruped robot, there is a torque response time window, which stipulates the total time range of the entire torque adjustment process. For example, if the theoretical minimum adjustment time is t and the torque response time window is T, then the number of stage divisions n = T / t (where n takes an integer value). The number of stage divisions determined in this way can reasonably divide the entire torque adjustment process into multiple stages according to the actual torque adjustment requirements and time limits.
[0124] Step S1424: Adopt a non-uniform time allocation strategy, allocate a shorter time window for the initial stage to quickly approach the target value, and allocate a longer time window for the final stage to achieve precise stability.
[0125] During the climbing process of a quadruped robot, the initial stage requires the leg joints to quickly obtain sufficient torque to adapt to the slope of the ramp. For example, at the moment of starting to climb, allocate a shorter time window for the initial stage. Within this time window, the servo motor can adjust the torque at a relatively fast speed, so that the torque value quickly approaches an intermediate range of the target value. In the final stage, in order to ensure the precise stability of the leg joints during the climbing process, a longer time window needs to be allocated. In this stage, the torque adjustment speed will be relatively slow to accurately reach the target torque value and be able to maintain stability after reaching the target torque value, avoiding torque fluctuations from affecting the stability of the robot on the ramp.
[0126] Step S1425: Set up a transition buffer within each stage to absorb torque fluctuations caused by sudden load changes.
[0127] During the climbing process of a quadruped robot, it may encounter some small obstacles or uneven ground, which can cause sudden changes in the load on the leg joints, resulting in torque fluctuations. For example, when the robot's leg steps on a small stone, it will instantaneously increase the load on the leg joint, causing torque fluctuations. The transition buffer set within each torque adjustment stage can absorb this torque fluctuation caused by sudden load changes. The transition buffer can be implemented through some special control algorithms or hardware circuits. When torque fluctuations are detected, adjustments can be made within the buffer to ensure that the torque fluctuations do not affect the stability and accuracy of the entire torque adjustment process, ensuring that the quadruped robot can climb the slope smoothly.
[0128] In a possible implementation manner, the method further includes:
[0129] Step S210, after the torque output by the servo motor reaches the target torque value, start the torque maintenance monitoring mode.
[0130] When a quadruped robot walks on complex terrains (such as those with different slopes, potholes, obstacles, etc.) or performs specific tasks, the servo motors of the leg joints operate according to the established torque adjustment strategy.
[0131] In this embodiment, when the servo motor of the leg joint of the quadruped robot completes the torque adjustment required to switch from the flat-ground walking state to climbing and reaches the target torque value during climbing, the torque maintenance monitoring mode is then enabled. In this torque maintenance monitoring mode, it is necessary to ensure the stable output of torque in real time to guarantee the stability and normal movement of the quadruped robot during the climbing process.
[0132] Step S220, collect the vibration spectrum data and temperature rise data of the servo motor in real time, and analyze the change in the mechanical load state.
[0133] During the climbing process of a quadruped robot, the vibration spectrum data of the servo motor reflects the operating state of the internal mechanical structure of the motor and the influence of the external load. For example, as the climbing angle of the robot increases, the load on the leg joints gradually increases, which may cause changes in the vibration spectrum of the motor. At the same time, the motor generates heat during continuous torque output, and the temperature rise data can reflect the thermal state of the motor. These data are collected through temperature sensors and vibration sensors respectively. If the amplitudes of certain frequency components in the vibration spectrum suddenly increase, or the temperature rise rate accelerates, this may indicate a change in the mechanical load state. For example, it may be that the leg joints of the robot encounter a steeper slope or greater friction during the climbing process, resulting in the motor having to bear a greater load.
[0134] Step S230, when the characteristics of sudden load change are detected, immediately pause the current adjustment strategy and trigger an emergency braking instruction.
[0135] Suppose that during the uphill climbing process of a quadruped robot, a leg suddenly gets stuck on a rock hidden on the slope, which will cause an instantaneous increase in the motor load, resulting in a load mutation. When this load mutation feature is detected through the analysis of vibration spectrum data and temperature rise data, in order to avoid damage to the motor and other components of the robot, it is necessary to immediately suspend the currently executing torque adjustment strategy. At the same time, trigger an emergency braking command to stop the motor or reduce its speed. This emergency braking command can be sent to the driver of the servo motor through the control circuit, and the driver quickly adjusts the current or voltage of the motor according to the emergency braking command, thereby achieving braking.
[0136] Step S240, after braking is completed, re-obtain the analog input signal based on the latest load state and generate an adaptation adjustment strategy.
[0137] In the above example, after the servo motor of the leg joint completes braking, it is necessary to re-evaluate the current load state. For example, determine factors such as the degree to which the leg is stuck by the rock, the change in the overall center of gravity of the robot, and the remaining uphill climbing task. Then, based on this latest load state information, re-obtain the analog input signal. These analog input signals may come from the central control system of the quadruped robot, which recalculates and sends the analog input signal to the servo motor according to the robot's attitude sensors (such as gyroscopes, accelerometers, etc.) and task planning modules (such as continuing to climb uphill or trying to bypass the rock). Based on these new analog input signals, generate a torque adjustment strategy that adapts to the current load state again according to the steps described above. For example, it may be necessary to adjust parameters such as the target torque value, torque change rate threshold, and torque response time window to ensure that the leg joint can continue to work normally under the new load conditions, enabling the quadruped robot to safely complete the uphill climb or take other appropriate actions.
[0138] Step S250, record the load mutation event and the corresponding countermeasure strategy in the historical database for optimizing the training process of the subsequent parameter mapping model.
[0139] In a possible implementation manner, step S250 includes:
[0140] Step S251, extract the load mutation event records within a preset time range from the historical database. Each load mutation event record includes an event trigger timestamp, load mutation feature data, countermeasure strategy execution parameters, and adjusted operating state data.
[0141] For example, during the walking process of a quadruped robot on various terrains over a certain period in the past (such as the most recent month), whenever a load mutation event occurs, this information is recorded. The event trigger timestamp precisely records the moment when the load mutation occurs, which helps analyze the timing relationship of the events. The load mutation characteristic data contains detailed information such as the torque fluctuation amplitude, the peak value of the vibration spectrum, and the temperature rise change rate. For example, when the robot crosses a large pothole at a certain moment, the torque fluctuation amplitude may suddenly increase, the peak value of the vibration spectrum may appear at a specific frequency, and at the same time, the temperature rise change rate will also change accordingly, and all these data are recorded. The execution parameters of the coping strategy record the emergency braking instructions taken when a load mutation is detected and the various parameters in the torque adjustment strategy after readjustment (such as the target torque value, the torque change rate threshold, etc.). The operating state data after adjustment reflects the operating state of the servo motor and the quadruped robot as a whole after the coping strategy is taken, such as whether normal movement is successfully restored and whether new abnormal conditions occur.
[0142] Step S252: Classify the recorded load mutation events by event type. Based on the torque fluctuation amplitude, the peak value of the vibration spectrum, and the temperature rise change rate in the load mutation characteristic data, classify stable load mutation events, transient shock events, and continuous overload events.
[0143] Taking the operation of a quadruped robot in different scenarios as an example, if the torque fluctuation amplitude is small, the peak value of the vibration spectrum is low, and the temperature rise change rate is relatively stable, this situation may be classified as a stable load mutation event. For example, when the robot walks on a relatively flat but slightly undulating ground and occasionally encounters small bumps or depressions, the resulting load mutation belongs to this category. When the torque fluctuation amplitude is large, the peak value of the vibration spectrum suddenly increases, and the temperature rise change rate rapidly increases within a short period, it may be a transient shock event. For example, the robot's leg suddenly hits a hard obstacle, generating a large impact force instantaneously. If the torque fluctuation amplitude remains at a high level continuously, the peak value of the vibration spectrum remains at a high position for a long time, and the temperature rise change rate continues to rise, this is a continuous overload event, which may be caused by the continuous excessive load on the robot when performing a certain task (such as climbing a slope for a long time or dragging a heavy object).
[0144] Step S253: Extract a set of characteristic data associated with the parameter mapping model from the classified load mutation event records. The set of characteristic data includes the analog input signal segment before the event occurs, the set of torque adjustment parameters, the current operating state parameters, and the dynamic adjustment parameters in the coping strategy.
[0145] In the scenario of a quadruped robot, the analog input signal segment before the event reflects the signal situation sent by the central control system to the servo motor for torque adjustment before the load mutation. For example, before the robot is about to encounter an obstacle, the analog input signal may be adjusted according to the normal walking mode. The torque adjustment parameter set includes parameters such as the target torque value, the torque change rate threshold, and the torque response time window, which are set according to the normal operation requirements before the load mutation. The current operating state parameters such as the motor winding current sampling value and the motor rotor speed detection value reflect the actual operating state of the servo motor before the load mutation. The dynamic adjustment parameters in the coping strategy are the relevant parameters in the coping measures taken after detecting the load mutation, such as the braking speed during emergency braking and the adjusted torque adjustment slope, etc.
[0146] Step S254, perform time series alignment processing on the feature data set, match the analog input signal segment, the torque adjustment parameter set, and the dynamic adjustment parameters corresponding to the same load mutation event along the time axis, and generate an original training data group with time series tags.
[0147] For example, for a transient impact event where a quadruped robot's leg hits an obstacle, match the analog input signal segment within a short period before the event in chronological order with the torque adjustment parameter set, the motor winding current sampling value, the motor rotor speed detection value, and the dynamic adjustment parameters in the coping strategy (such as the start time of emergency braking, the time point of readjusting the torque, etc.) at that time, and label each data element with an accurate time tag, thus generating an original training data group. The purpose of doing this is to accurately reflect the time series relationship between the data in subsequent model training, so as to better learn the parameter mapping relationship under the condition of load mutation.
[0148] Step S255, perform signal reconstruction on the analog input signal segment in the original training data group, remove the noise interference introduced during the signal acquisition process, and standardize the signal amplitude range to generate a denoised standard signal sample.
[0149] In the actual operation of a quadruped robot, due to environmental interference (such as electromagnetic interference, mechanical vibration, etc.) and the limitations of the acquisition equipment, the analog input signal segment may contain noise. Through signal processing algorithms, such as filtering algorithms to remove high-frequency noise, and then perform standardization processing according to the known signal amplitude range. For example, if the amplitude range of the original analog input signal is [-10V to 10V], standardize it to the range of [0 to 1]. In this way, the generated denoised standard signal sample is purer, which is conducive to the parameter mapping model to accurately learn the relationship between the signal and the torque adjustment parameters.
[0150] Step S256: Conduct a correlation analysis on the torque adjustment parameter set and the dynamic adjustment parameters in the original training data set, identify the conflicting items in the parameter mapping relationship, and label the weight coefficients of the conflicting items based on the actual torque response effect in the adjusted operating state data.
[0151] During the process of a quadruped robot coping with load mutation, there may be conflicts between some parameters in the torque adjustment parameter set and the dynamic adjustment parameters in the coping strategy. For example, the adjustment of the target torque value may not match the torque change rate threshold in some cases, resulting in an unsatisfactory actual torque response effect. By analyzing the actual torque response effect of the servo motor after load mutation (such as whether smooth torque adjustment is achieved and whether abnormal operation of the motor is avoided), if it is found that a certain parameter (such as an overly high torque change rate threshold leading to too fast torque adjustment and excessive motor vibration) has a negative impact on the actual torque response effect, it is marked as a conflicting item, and the weight coefficient is labeled according to the severity of the impact. For example, if a certain conflicting item causes obvious unstable operation of the motor, a relatively high weight coefficient may be given, indicating that this parameter relationship needs to be focused on and adjusted during subsequent model training.
[0152] Step S257: Combine the denoised standard signal samples, the torque adjustment parameter set with labeled weight coefficients, and the dynamic adjustment parameters to generate an enhanced training data set, and add the enhanced training data set to the training data set of the parameter mapping model.
[0153] In the scenario of a quadruped robot, after the previous processing, the denoised standard signal samples, the torque adjustment parameter set with weight coefficient labels (such as the target torque value and its corresponding weight coefficient), and the dynamic adjustment parameters (such as the relevant parameters and their weight coefficients during emergency braking) are combined in a certain format to form an enhanced training data set. Then, this enhanced training data set is added to the training data set of the parameter mapping model to provide more targeted data for the model training, so that the model can better learn how to accurately perform parameter mapping under load mutation and improve the adaptability of the model to different load conditions.
[0154] Step S258: Clean the redundant data in the training data set of the parameter mapping model, remove the historical data sets that do not match the current servo motor model and the redundant data sets with duplicate mapping relationships, and generate an optimized training data set.
[0155] Throughout the entire operating history of the quadruped robot, there may be some data that is not relevant to the currently used servo motor model, such as data from other previously used servo motor models or invalid data due to data acquisition errors, etc. This data will interfere with the training of the model and needs to be removed from the training dataset. At the same time, redundant data groups with duplicate mapping relationships also need to be cleaned up. For example, if there are multiple data groups with exactly the same mapping relationships among signal samples, torque adjustment parameter sets, and dynamic adjustment parameters, only one of the data groups needs to be retained. After such redundant data cleaning, an optimized training dataset is obtained, which is more concise and accurate, and is conducive to improving the training efficiency and accuracy of the parameter mapping model.
[0156] Step S259, use the optimized training dataset to perform incremental training on the parameter mapping model, adjust the signal feature weight matrix and the parameter mapping layer connection coefficients in the parameter mapping model, and generate an updated parameter mapping model.
[0157] In the application scenario of the quadruped robot, use the optimized training dataset to train the parameter mapping model. During the training process, according to the sample data in the dataset, gradually adjust the signal feature weight matrix in the model. For example, if a certain signal feature (such as a certain frequency component in the analog input signal) is proven to have a strong correlation with the torque adjustment parameter in more sample data, then the weight value of this signal feature in the weight matrix will be increased. At the same time, adjust the parameter mapping layer connection coefficients to optimize the mapping relationship of the model from the input signal to the torque adjustment parameter. Through such an incremental training process, an updated parameter mapping model is generated, which can better adapt to the torque adjustment requirements of the quadruped robot under different load mutation conditions.
[0158] Step S2510, deploy the updated parameter mapping model to the control unit of the servo system, and verify the matching degree between the set of torque adjustment parameters output by the model and the actual load demand in the next adjustment cycle.
[0159] In the actual operation of the quadruped robot, deploy the updated parameter mapping model to the control unit of the leg joint servo system. During the next torque adjustment process (for example, when the robot encounters a similar load mutation situation or normal torque adjustment requirements next time), observe whether the set of torque adjustment parameters output by the model (such as the target torque value, torque change rate threshold, etc.) matches the actual load demand. For example, when the robot climbs a slope or crosses an obstacle again, according to the actual load situation (such as the slope of the slope, the size of the obstacle, etc.), check whether the torque adjustment parameters output by the model can make the servo motor output an appropriate torque to ensure the normal movement of the robot.
[0160] Step S2511: When the detected matching degree is lower than the preset threshold, trigger a model rollback instruction and re-extract the significant weight event records from the historical database for local training until the matching degree is restored within the allowable range.
[0161] If during the verification process, it is found that the matching degree between the set of torque adjustment parameters output by the model and the actual load demand is lower than the preset threshold, this indicates that there may be problems with the updated model. For example, it may be that during the incremental training process, the model overfits to some data or fails to fully learn the key parameter mapping relationships. At this time, trigger a model rollback instruction to restore the parameter mapping model to a previous version or state. Then re-extract the event records with significant weights from the historical database. These event records are the records that were considered to have an important impact on model training in the previous analysis (such as certain special load mutation events or events that have a greater impact on the actual torque response effect). Conduct local training for these significant weight event records, focusing on adjusting the model parameters related to these events, and repeatedly perform verification and training until the matching degree between the set of torque adjustment parameters output by the model and the actual load demand is restored within the allowable range, so as to ensure that the parameter mapping model always maintains high accuracy and reliability during the torque adjustment process of the quadruped robot.
[0162] Figure 2 FIG. shows a schematic diagram of exemplary hardware and software components of a motor torque dynamic adjustment system 100 based on analog input that can implement the idea of the present application provided by some embodiments of the present application. For example, the processor 120 can be used on the motor torque dynamic adjustment system 100 based on analog input and is used to execute the functions in the present application.
[0163] The motor torque dynamic adjustment system 100 based on analog input can be a general-purpose server or a special-purpose server, both of which can be used to implement the motor torque dynamic adjustment method based on analog input of the present application. Although only one server is shown in the present application, for convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.
[0164] For example, the motor torque dynamic adjustment system 100 based on analog input may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as disks, ROM, or RAM, or any combination thereof. Exemplarily, the motor torque dynamic adjustment system 100 based on analog input may further include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The motor torque dynamic adjustment system 100 based on analog input further includes an input / output (I / O) interface 150 between the computer and other input / output devices.
[0165] For ease of illustration, only one processor is described in the motor torque dynamic adjustment system 100 based on analog input. However, it should be noted that the motor torque dynamic adjustment system 100 based on analog input in the present application may further include multiple processors. Therefore, the steps executed by one processor described in the present application may also be jointly executed or separately executed by multiple processors. For example, if the processor of the motor torque dynamic adjustment system 100 based on analog input executes step A and step B, it should be understood that step A and step B may also be jointly executed by two different processors or separately executed in one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.
[0166] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned motor torque dynamic adjustment method based on analog input is implemented.
[0167] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.
Claims
1. A method for dynamic adjustment of motor torque based on analog input, characterized in that: The method comprises: Acquire a set of analog input signals received by the servo motor in a running state, wherein the set of analog input signals includes at least one analog input signal, and the analog input signal is used to characterize a torque adjustment requirement of an external device on the servo motor; Performing signal analysis on the analog input signal set to obtain a torque adjustment parameter set corresponding to each analog input signal, wherein the torque adjustment parameter set includes a target torque value, a torque change rate threshold, and a torque response time window; Determine the current torque value and torque dynamic response characteristics of the servo motor based on the torque adjustment parameter set and the current operating state parameters of the servo motor, wherein the current operating state parameters include motor winding current sampling values and motor rotor speed detection values, and the torque dynamic response characteristics include a maximum allowable adjustment rate and a minimum stabilization time threshold; Generate a torque dynamic adjustment strategy for the servo motor according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics, wherein the torque dynamic adjustment strategy includes a segmented torque adjustment instruction sequence, wherein a plurality of torque adjustment stages are divided based on a maximum allowable adjustment rate and a minimum stabilization time threshold in the torque dynamic response characteristics, and each torque adjustment stage corresponds to an adjustment time window and a stage target torque value; The segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy is executed to make the output torque of the servo motor dynamically match the target torque value according to a preset time window while satisfying the torque change rate threshold constraint.
2. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The signal analysis of the analog input signal set to obtain a torque adjustment parameter set corresponding to each analog input signal includes: Performing frame processing on each analog input signal to obtain multiple signal frame segments, each signal frame segment contains analog signal sampling values within a continuous time window; Extract features from each signal frame segment to obtain the signal amplitude feature, signal frequency feature and signal duty cycle feature corresponding to each signal frame segment; Inputting the signal amplitude feature, signal frequency feature and signal duty cycle feature into a predefined parameter mapping model, and outputting a candidate torque adjustment parameter subset corresponding to each signal frame segment through the parameter mapping model; The candidate torque adjustment parameter subsets corresponding to the multiple signal frame segments are time-aligned and parameter-fused to generate a torque adjustment parameter set corresponding to the analog input signal; wherein the parameter mapping model is obtained by training the mapping relationship between multiple groups of analog signal samples and torque adjustment parameter labels.
3. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The step of determining the current torque value and torque dynamic response characteristics of the servo motor based on the torque adjustment parameter set and the current operating state parameters of the servo motor includes: Obtaining a motor winding current sampling value of a servo motor, and filtering the motor winding current sampling value to obtain a stable current value; Obtaining a motor rotor speed detection value of the servo motor, and performing speed differential calculation on the motor rotor speed detection value to obtain a rotor acceleration value; According to the stable current value and the rotor acceleration value, a predefined torque mapping relationship table is queried to determine the current torque value of the servo motor; Calculating the torque change rate and the torque fluctuation amplitude based on the change trend of the current torque value within a preset time interval; The torque dynamic response characteristics of the servo motor are determined according to the torque change rate and the torque fluctuation amplitude.
4. The method for dynamic adjustment of motor torque based on analog input according to claim 3, characterized in that: The step of generating a torque dynamic adjustment strategy of the servo motor according to the torque adjustment parameter set, the current torque value and the torque dynamic response characteristics comprises: Calculating the difference between the target torque value and the current torque value, and determining the torque adjustment direction according to the difference; According to the torque change rate threshold constraint, a corresponding torque adjustment slope is allocated to each adjustment stage, and the torque adjustment slope is used to control the acceleration of the torque change in the stage; The stage target torque value, adjustment time window and torque adjustment slope of each adjustment stage are combined to generate a segmented torque adjustment instruction sequence; The segmented torque adjustment instruction sequence is time-checked to ensure that the torque adjustment slope change in adjacent stages does not exceed a preset connection threshold.
5. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The step of executing the segmented torque adjustment instruction sequence in the torque dynamic adjustment strategy includes: According to the first stage instruction in the segmented torque adjustment instruction sequence, a first adjustment instruction is sent to the torque control module of the servo motor, wherein the first adjustment instruction includes a first stage target torque value and a corresponding adjustment time window; In the first stage adjustment time window, the actual torque change curve of the servo motor is monitored in real time, and the actual torque change rate is calculated; When it is detected that the actual torque change rate deviates from the torque adjustment slope in the first stage command, the adjustment time window or target torque value of the subsequent stage is dynamically corrected according to the deviation amplitude; After the current stage adjustment is completed, the next stage adjustment instruction is triggered according to the revised segmented torque adjustment instruction sequence until all stage instructions are executed; After the final stage of adjustment is completed, the output torque stability of the servo motor is continuously monitored. If the torque fluctuation is detected to be beyond the allowable range, the local adjustment instruction is regenerated and inserted into the execution queue; The dynamically correcting the adjustment time window or target torque value of the subsequent stage according to the deviation amplitude includes: Calculate the deviation percentage between the actual torque change rate and the command required rate; When the deviation percentage is less than the first threshold, only the torque adjustment slope in the subsequent stage is fine-tuned; When the deviation percentage is between the first threshold and the second threshold, the time window of the subsequent stage is extended proportionally; When the deviation percentage exceeds a second threshold, recalculate the remaining adjustment amount and generate a new stage division scheme; The timestamp of the original instruction is retained during the correction process to ensure that the corrected instruction sequence is traceable.
6. The method for dynamic adjustment of motor torque based on analog input according to claim 2, characterized in that: The step of performing timing alignment and parameter fusion on the candidate torque adjustment parameter subsets corresponding to the plurality of signal frame segments to generate a torque adjustment parameter set corresponding to the analog input signal includes: Extracting timestamp information of each signal frame segment, and sorting the candidate torque adjustment parameter subsets according to the time sequence of the timestamp information; Performing overlapping region detection on candidate torque adjustment parameter subsets of adjacent signal frame segments to determine a parameter continuous change interval; In the interval of continuous parameter change, the target torque value and torque change rate threshold are weighted and smoothed to eliminate parameter jumps; For a subset of candidate torque adjustment parameters in a discontinuous variation interval, a preset interpolation algorithm is used to supplement intermediate parameter values; The processed candidate torque adjustment parameter subsets are merged according to the time axis to generate a complete torque adjustment parameter set.
7. The method for dynamic adjustment of motor torque based on analog input according to claim 3, characterized in that: The filtering process of the motor winding current sampling value to obtain a stable current value includes: A sliding window mean filter is used to perform preliminary denoising on the original current sampling values; Perform wavelet transform analysis on the denoised current value to separate high-frequency noise components from low-frequency effective signals; According to the rated current range of the servo motor, the abnormal current value that exceeds the reasonable range is truncated and replaced; Perform phase compensation on the processed current signal to eliminate signal distortion caused by sampling delay; The compensated current signal is synchronously calibrated with the real-time position signal of the motor rotor to obtain a stable current value.
8. The method for dynamic adjustment of motor torque based on analog input according to claim 4, characterized in that: The step of dividing the torque adjustment stages into multiple stages based on the maximum allowable adjustment rate and the minimum stabilization time threshold in the torque dynamic response characteristics includes: Determine the total adjustment amount according to the absolute value of the difference between the target torque value and the current torque value; Based on the ratio of the total adjustment amount to the maximum allowable adjustment rate, the theoretical minimum adjustment time is calculated; The number of stage divisions is determined according to the ratio of the theoretical minimum adjustment time to the torque response time window; A non-uniform time allocation strategy is adopted, which allocates a shorter time window to the initial stage to quickly approach the target value, and a longer time window to the final stage to achieve precise stability; A transition buffer is set inside each stage to absorb torque fluctuations caused by sudden load changes.
9. The method for dynamic adjustment of motor torque based on analog input according to claim 1, characterized in that: The method further comprises: After the servo motor output torque reaches the target torque value, the torque maintenance monitoring mode is started; Collect the vibration spectrum data and temperature rise data of the servo motor in real time and analyze the changes in mechanical load status; When a sudden load change is detected, the current adjustment strategy is immediately suspended and an emergency braking command is triggered; After braking is completed, the analog input signal is re-acquired based on the latest load status and an adaptive adjustment strategy is generated; Record load mutation events and response strategies in a historical database to optimize the training process of subsequent parameter mapping models; The process of recording the load mutation event and the response strategy in the history database for optimizing the subsequent training process of the parameter mapping model includes: Extracting load mutation event records within a preset time range from the historical database, each load mutation event record including an event triggering timestamp, load mutation feature data, response strategy execution parameters, and adjusted operating status data; Classifying the load mutation event records into event types, and dividing them into stable load mutation events, transient impact events, and continuous overload events according to the torque fluctuation amplitude, vibration spectrum peak value, and temperature rise change rate in the load mutation feature data; Extracting a feature data set associated with the parameter mapping model from the classified load mutation event records, the feature data set including an analog input signal segment before the event occurs, a torque adjustment parameter set, current operating state parameters, and dynamic adjustment parameters in a response strategy; Performing time alignment processing on the characteristic data set, matching the analog input signal segments, torque adjustment parameter sets and dynamic adjustment parameters corresponding to the same load mutation event according to the time axis, and generating an original training data set with time series labels; Reconstructing the analog input signal segments in the original training data group, removing the noise interference introduced in the signal acquisition process, and standardizing the signal amplitude range to generate a denoised standard signal sample; Performing correlation analysis on the torque adjustment parameter set and the dynamic adjustment parameter in the original training data group, identifying conflicting items in the parameter mapping relationship, and marking weight coefficients of the conflicting items based on actual torque response effects in the adjusted operating state data; The denoised standard signal sample, the torque adjustment parameter set with the weight coefficients marked, and the dynamic adjustment parameters are combined to generate an enhanced training data set, and the enhanced training data set is added to the training data set of the parameter mapping model; Clean the redundant data of the training data set of the parameter mapping model, remove the historical data sets that do not match the current servo motor model and the redundant data sets with repeated mapping relationships, and generate an optimized training data set; Performing incremental training on the parameter mapping model using the optimized training data set, adjusting the signal feature weight matrix and the parameter mapping layer connection coefficient in the parameter mapping model, and generating an updated parameter mapping model; Deploy the updated parameter mapping model to the control unit of the servo system, and verify the matching degree between the torque adjustment parameter set output by the model and the actual load demand in the next adjustment cycle; When it is detected that the matching degree is lower than the preset threshold, the model rollback instruction is triggered and the significant weight event records in the historical database are re-extracted for local training until the matching degree is restored to the allowable range.
10. A motor torque dynamic adjustment system based on analog input, characterized in that: The motor torque dynamic adjustment system based on analog input includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the motor torque dynamic adjustment method based on analog input as described in any one of claims 1 to 9.
Citation Information
Patent Citations
Gait track planning method and device, readable storage medium and robot
CN113110484A
Crawling state planning method and system for quadruped robot capable of crossing large obstacles
CN116449711A