A robot motion control method and device, electronic equipment and storage medium

Abstract

The application provides a robot motion control method and device, electronic equipment and storage medium, which are used for improving the stability and consistency of linear walking control. The method comprises the following steps: receiving a control instruction and time information thereof, and obtaining actual motion states of left and right driving units; estimating a time delay based on the time information and generating a communication quality parameter; constructing an inter-wheel synchronization error and performing a prediction compensation to obtain a prediction error; inputting the actual motion states, the time delay and the communication quality parameter into a self-supervised residual prediction model to obtain a slip risk indicator and a prediction uncertainty; determining a control mode based on the prediction error, the communication quality parameter, the slip risk indicator and the prediction uncertainty; generating a synchronization control quantity based on the prediction error and the control mode; determining a synchronization control weight based on the control mode and a control reliability parameter; weighting the synchronization control quantity according to the synchronization control weight and distributing the synchronization control quantity as target control instructions of the left and right driving units, and outputting the target control instructions to a driving control unit for execution.

Description

Technical Field

[0001] This application relates to the field of robot control technology, and in particular to a robot motion control method, device, electronic device and storage medium. Background Technology

[0002] Mobile robots are widely used in scenarios such as warehouse handling, park inspection, and indoor delivery. As application scenarios expand from structured environments to semi-structured environments, the requirements for the stability and consistency of the robot's movement process are gradually increasing. Basic movement capabilities such as straight-line walking and low-curvature walking have become important foundations for ensuring task efficiency and operational safety.

[0003] In existing technologies, mobile robots typically receive motion control commands from external sources and combine them with feedback information from the drive unit for control. Since control commands and feedback information often need to be transmitted via a communication link in actual deployment, changes in the communication link's state can cause instability in the arrival of control information, making it difficult to maintain a consistent timing relationship in the control process, thus affecting the motion control effect under certain operating conditions. Summary of the Invention

[0004] This application provides a robot motion control method, device, electronic device, and storage medium, which can improve the stability and consistency of linear walking control.

[0005] The first aspect of this application provides a robot motion control method, including:

[0006] Receive control commands and their timing information, and obtain the actual motion state of the left and right drive units; Based on the time information, the delay is estimated and communication quality parameters are generated; Based on the actual motion state, the inter-wheel synchronization error is constructed and predicted and compensated to obtain the prediction error; The actual motion state, the time delay, and the communication quality parameters are input into the trained self-supervised residual prediction model to obtain the slip risk indicator and prediction uncertainty. The control mode is determined based on the prediction error, the communication quality parameters, the slip risk indicator, and the prediction uncertainty. A synchronous control quantity is generated based on the prediction error and the control mode; The synchronization control weight of the synchronization control quantity is determined based on the control mode and at least one control reliability parameter, wherein the control reliability parameter is at least one of the communication quality parameter, the slip risk indicator and / or the prediction uncertainty. The synchronization control quantity is weighted and adjusted according to the synchronization control weight, and the weighted synchronization control quantity is allocated as the target control command of the left and right drive units under preset constraints. The target control command is output to the drive control unit for execution.

[0007] Optionally, the step of inputting the actual motion state, the time delay, and the communication quality parameters into the trained self-supervised residual prediction model to obtain the slip risk indicator and prediction uncertainty includes: The actual motion state is constructed as the model input feature and input into the self-supervised residual prediction model, wherein the model input feature includes at least one of the actual speed and its change of the left and right drive units, drive current or torque command; An observable residual is constructed based on the actual motion state, and the observable residual is used as the self-supervised learning target of the self-supervised residual prediction model; wherein the observable residual includes wheel synchronization error, or includes a composite residual obtained by fusing wheel synchronization error and yaw residual, and the yaw residual is a residual constructed from the kinematic relationship between the body heading change measurement and the actual speed of the left and right drive units; Based on the self-supervised residual prediction model, the observable residual or its changing trend for the next control period is predicted to obtain the prediction result, wherein the prediction result includes the predicted value of the observable residual and / or the predicted value of the changing trend and / or parameters characterizing the prediction distribution. Based on the prediction results, a slip risk indicator and prediction uncertainty are generated.

[0008] Optionally, generating the slip risk indicator and prediction uncertainty based on the prediction results includes: Calculate the prediction uncertainty based on the prediction results; The slip risk indicator is generated by jointly mapping the prediction results and the prediction uncertainty, wherein the joint mapping is at least related to the magnitude and / or trend of the predicted observable residual values ​​in the prediction results.

[0009] Optionally, estimating the delay and generating communication quality parameters based on the time information includes: The sending end timestamp, receiving end timestamp, and message sequence number corresponding to the control command are obtained as the time information. Based on the time information, the clock deviation and clock drift between the master controller and the slave controller are estimated to obtain the clock deviation parameter and the clock drift parameter; The timestamp of the transmitting end is corrected based on the clock deviation parameter and the clock drift parameter to obtain the corrected timestamp; Calculate the delay sample sequence based on the receiving end timestamp and the correction timestamp; Determine the time delay uncertainty based on the time delay sample sequence; Initial communication quality parameters are generated based on the time delay sample sequence and the time delay uncertainty.

[0010] Optionally, generating initial communication quality parameters based on the time delay sample sequence and the time delay uncertainty includes: Robust statistics are performed on the time delay sample sequence within a preset statistical window to obtain the time delay dispersion index; The packet loss rate is determined based on the message sequence number; The information freshness index is determined based on the timestamp of the receiving end, wherein the information freshness index is the difference between the current control time and the timestamp corresponding to the latest successfully received control instruction; Initial communication quality parameters are generated based on the delay dispersion index, the packet loss rate index, the information freshness index, and the delay uncertainty. The initial communication quality parameters are subjected to exponential smoothing to obtain smoothed communication quality parameters; Hysteresis determination is performed on the smooth communication quality parameters to obtain stable communication quality parameters.

[0011] Optionally, generating the synchronization control quantity based on the prediction error and the control mode includes: Determine the set of synchronous control structure parameters corresponding to the control mode based on the control mode; The change in error is calculated based on the prediction error; The prediction error and the error change are input into the control law corresponding to the synchronous control structure parameter set to obtain the synchronous control quantity.

[0012] Optionally, after allocating the weighted synchronous control quantity as target control commands for the left and right drive units under preset constraints, the method further includes: The update strategy for the target control command is determined based on the communication quality parameters, wherein the update strategy includes a periodic update strategy or an event-triggered update strategy; The step of outputting the target control command to the drive control unit for execution includes: The target control command is output to the drive control unit for execution according to the update strategy.

[0013] Optionally, determining the synchronization control weight of the synchronization control quantity based on the control mode and at least one control reliability parameter includes: The weight benchmark value is determined based on the control mode; The weight correction amount is determined based on at least one control reliability parameter; The synchronization control weight is determined based on the weight baseline value and the weight adjustment amount.

[0014] A second aspect of this application provides a robot motion control device for implementing the robot motion control method of the first aspect and any possible embodiment of the first aspect, the robot motion control device comprising: The receiving unit is used to receive control commands and their timing information, and to obtain the actual motion state of the left and right drive units; The first generation unit is used to estimate the delay and generate communication quality parameters based on the time information. A construction unit is used to construct the inter-wheel synchronization error based on the actual motion state and perform prediction compensation to obtain the prediction error; The input unit is used to input the actual motion state, the time delay and the communication quality parameters into the trained self-supervised residual prediction model to obtain the slip risk indication and prediction uncertainty. The first determining unit is used to determine the control mode based on the prediction error, the communication quality parameters, the slip risk indication, and the prediction uncertainty. The second generation unit is used to generate a synchronous control quantity based on the prediction error and the control mode; The second determining unit is configured to determine the synchronization control weight of the synchronization control quantity based on the control mode and at least one control reliability parameter, wherein the control reliability parameter is at least one of the communication quality parameter, the slip risk indication, and / or the prediction uncertainty. The allocation unit is used to adjust the synchronization control quantity according to the synchronization control weight, and allocate the weighted synchronization control quantity as the target control command of the left and right drive units under preset constraints. The output unit is used to output the target control command to the drive control unit for execution.

[0015] A third aspect of this application provides an electronic device, comprising: Processor, memory, input / output units, and bus; The processor is connected to the memory, the input / output unit, and the bus; The memory stores a program, and the processor calls the program to execute the method of the first aspect and any possible implementation of the first aspect.

[0016] The fourth aspect of this application provides a computer-readable storage medium storing a program that, when executed on a computer, causes the computer to perform the methods of the first aspect and any possible implementation thereof.

[0017] As can be seen from the above technical solutions, this application has the following advantages: In this embodiment, the control mode is determined based on prediction error, communication quality parameters, slip risk indication, and prediction uncertainty. Synchronization control quantity and synchronization control weight are generated within the control mode, allowing the correction strategy and correction intensity to be dynamically adjusted according to the communication link status and slip risk. This improves the stability and consistency of straight-line walking control under complex conditions, reducing the risks of deviation, jitter, and over-correction. Furthermore, the delay is estimated based on time information, and communication quality parameters are generated. Combined with prediction compensation for inter-wheel synchronization error, the prediction error is obtained, making the correction basis closer to the current control moment and reducing correction lag caused by delay. The actual motion state, delay, and communication quality parameters are input into a trained self-supervised residual prediction model to obtain the slip risk indication and prediction uncertainty, quantifying the changes in ground adhesion and the model's reliability. The synchronization control quantity is weighted according to the synchronization control weight and allocated as target control commands to the left and right drive units under preset constraints, ensuring left-right coordination within the execution capability range and improving execution safety. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating one embodiment of the robot motion control method in this application; Figure 2 This is a flowchart illustrating a sub-implementation of the robot motion control method in this application; Figure 3 This is a flowchart illustrating another sub-implementation of the robot motion control method in this application; Figure 4 This is a flowchart illustrating another sub-implementation of the robot motion control method in this application; Figure 5 This is a flowchart illustrating another sub-implementation of the robot motion control method in this application; Figure 6 This is a flowchart illustrating another sub-implementation of the robot motion control method in this application; Figure 7 This is a flowchart illustrating another sub-implementation of the robot motion control method in this application; Figure 8 This is a schematic diagram of one embodiment of the robot motion control device in this application; Figure 9 This is a schematic diagram of the structure of one embodiment of the electronic device in this application. Detailed Implementation

[0019] This application provides a robot motion control method, device, electronic device, and storage medium for use in... The method of this application can be applied to servers, terminals, or other devices with logic processing capabilities; therefore, this application does not limit its application. For ease of description, the following description uses a control device as the executing entity.

[0020] The embodiments of this application will now be described with reference to the accompanying drawings.

[0021] Please see Figure 1 , Figure 1 One embodiment of the robot motion control method provided in this application includes: 101. Receive control commands and their timing information, and obtain the actual motion state of the left and right drive units; Robot motion control methods are designed for mobile robots with two drive units, typically including warehouse handling robots, park inspection robots, and indoor delivery robots. These robots perform high-frequency movements such as straight-line walking and low-curvature turning. Control commands are usually issued from a host scheduling terminal, remote control terminal, or autonomous planning module, and reach the robot body via a communication link. The communication link is prone to delays, jitter, packet loss, and out-of-order delivery under conditions of congestion, interference, or multi-node forwarding, leading to asynchronous outputs from the left and right drives, manifesting as deviation, jitter, and over-correction. Simultaneously, differences in ground adhesion, load variations, and tire wear can introduce slippage and changes in controllability. This method incorporates time information, communication quality, slippage risk, and uncertainty into the control decision, making straight-line motion more stable and consistent.

[0022] The control device receives control commands that describe the desired motion, such as desired linear velocity, desired angular velocity, or equivalent left and right wheel reference targets. Simultaneously, the control device receives time information bound to the control commands, which characterizes the moment the control commands are generated or transmitted. The control device acquires the actual motion state of the left and right drive units, which includes at least the actual speeds of the left and right drive units; to enhance the characterization of load changes and actuator states, the actual motion state may also include speed changes, drive current, or drive torque feedback. Within the same control cycle, the control device aligns and records the control commands, time information, and actual motion state, providing a consistent data foundation for subsequent time delay estimation, error construction, and model inference.

[0023] 102. Estimate latency and generate communication quality parameters based on time information; The control device estimates the transmission delay of control commands based on time information. The delay can be determined by the difference between the current control moment and the time information, forming a delay sample sequence within a continuous control cycle. To reduce errors caused by master-slave clock inconsistency, the control device can estimate clock deviation and clock drift within a sliding window, and correct the time information before calculating the delay. The control device generates communication quality parameters, which characterize the link's real-time performance and reliability. These parameters can be composed of indicators such as mean delay, delay dispersion, arrival interval dispersion, timeout count, packet loss count, and information freshness, and can be fused into a single scalar or vector form. To suppress decision jitter caused by link state fluctuations, the control device can perform exponential smoothing or threshold hysteresis processing on the communication quality parameters.

[0024] 103. Construct the inter-wheel synchronization error based on the actual motion state and perform prediction compensation to obtain the prediction error; The control device constructs the inter-wheel synchronization error based on the actual motion state. This error can be obtained from the actual speed difference between the left and right drive units, or it can be combined with speed changes to form a more sensitive error representation. The control device performs predictive compensation for the inter-wheel synchronization error to compensate for error lag caused by time delay: when time delay exists, the measured synchronization error reflects the deviation at an earlier moment. Predictive compensation can extrapolate the synchronization error using the error change trend and time delay, thereby obtaining a predicted error closer to the current control moment. To avoid noise amplification, the control device can filter, limit, or employ a robust differential strategy to assess the error change trend.

[0025] 104. Input the actual motion state, time delay and communication quality parameters into the trained self-supervised residual prediction model to obtain the slip risk indication and prediction uncertainty; The control unit constructs model input features and performs model inference. These input features include at least the actual motion state, latency, and communication quality parameters. The control unit inputs these features into a trained self-supervised residual prediction model and outputs a slip risk indicator and prediction uncertainty. The slip risk indicator characterizes the probability of wheel-to-ground adhesion degradation, slippage, or decreased controllability; the prediction uncertainty characterizes the reliability of the model output and can be expressed as prediction distribution dispersion, prediction interval width, or an equivalent uncertainty measure. During model training, observable residuals can be constructed using runtime data as learning objectives to complete self-supervised training; during model deployment, the control unit infers and outputs the risk and uncertainty online.

[0026] 105. Determine the control mode based on prediction error, communication quality parameters, slip risk indication, and prediction uncertainty; The control device determines the control mode by comprehensively considering prediction error, communication quality parameters, slip risk indication, and prediction uncertainty. The control mode is used to select different sets of control strategies and parameter sets. Control mode determination can employ threshold segmentation, rule-based logic, or equivalent decision logic: a conservative mode is entered when communication quality deteriorates or prediction uncertainty increases; a steady-state priority mode is entered when slip risk increases; and a degradation mode is entered when communication quality severely deteriorates. After the control mode is output, it provides a unified scheduling basis for the generation of synchronization control quantities and the determination of weights.

[0027] 106. Generate synchronous control quantities based on prediction errors and control modes; The control device selects the synchronous control structure parameter set according to the control mode and generates a synchronous control quantity based on the prediction error. The synchronous control quantity is used to suppress the tendency of deviation caused by the asynchrony of the left and right drives. The control device can first calculate the error change from the prediction error, and then substitute it into the control law under the corresponding parameter set of the control mode to obtain the synchronous control quantity; the control law can adopt proportional, integral, derivative form or equivalent control structure, and the parameter set is switched synchronously when the control mode is switched to take into account both response speed and stability.

[0028] 107. Determine the synchronization control weight of the synchronization control quantity based on the control mode and at least one control reliability parameter, wherein the control reliability parameter is at least one of the following: communication quality parameter, slip risk indication and / or prediction uncertainty; The control device determines synchronization control weights to adjust the influence of the synchronization control quantity on the final output. Synchronization control weights can be determined using a baseline value and a correction factor: the control mode provides the baseline weight value, and the control reliability parameter provides the weight correction factor or mapping relationship. When communication quality deteriorates, slippage risk increases, or prediction uncertainty increases, the synchronization control weights are set to more conservative values ​​to reduce the risk of over-correction; when communication quality is good and the risk is low, the synchronization control weights can be set to more aggressive values ​​to improve synchronization accuracy. The control device can introduce rate-of-change limits or smooth updates to the weights to reduce output jitter caused by sudden weight changes.

[0029] 108. Adjust the synchronization control quantity according to the synchronization control weight, and allocate the weighted synchronization control quantity as the target control command of the left and right drive units under preset constraints. The control device weights and adjusts the synchronization control quantity according to the synchronization control weights to form the correction intensity for wheel asynchrony. Under preset constraints, the control device allocates the weighted synchronization control quantity as target control commands for the left and right drive units. The target control commands can be obtained by combining the reference target corresponding to the control command with the correction deviation component. The preset constraints are used to limit the amplitude and intensity of the target control commands. The preset constraints can include speed limits, acceleration limits, torque or current limits, rate of change limits, etc. The allocation process satisfies the preset constraints and maintains the coordination between the left and right drive commands, avoiding output exceeding the execution capacity or introducing sudden changes.

[0030] 109. Output the target control command to the drive control unit for execution.

[0031] The control device sends the target control command to the drive control unit, which completes closed-loop execution based on the target control command; the drive control unit continuously feeds back the actual motion state, and the control device forms closed-loop control accordingly.

[0032] In this embodiment, the control device determines the control mode based on prediction error, communication quality parameters, slip risk indication, and prediction uncertainty. Within this control mode, it generates synchronous control quantities and synchronous control weights, allowing the correction strategy and intensity to dynamically adjust according to the communication link status and slip risk. This improves the stability and consistency of straight-line walking control under complex conditions, reducing the risks of deviation, jitter, and over-correction. Furthermore, it estimates the time delay based on time information and generates communication quality parameters. Combined with prediction compensation for inter-wheel synchronization errors, it obtains the prediction error, making the correction basis closer to the current control moment and reducing correction lag caused by time delay. The actual motion state, time delay, and communication quality parameters are input into a trained self-supervised residual prediction model to obtain slip risk indication and prediction uncertainty, quantifying changes in ground adhesion and model reliability. The synchronous control quantity is weighted according to the synchronous control weights and allocated as target control commands to the left and right drive units under preset constraints, ensuring left-right coordination within the execution capability range and improving execution safety.

[0033] Please see Figure 2 In some embodiments of this application, step 104 in the above embodiments, which inputs the actual motion state, latency, and communication quality parameters into the trained self-supervised residual prediction model to obtain the slip risk indicator and prediction uncertainty, may include the following steps: 201. Construct the actual motion state as the model input feature and input it into the self-supervised residual prediction model, wherein the model input feature includes at least one of the actual speed and its change of the left and right drive units, drive current or torque command; The control device performs feature processing on the acquired actual motion state, mapping the actual speeds of the left and right drive units to feature vectors within the same sampling period, and obtaining the speed change based on the speed difference or filtering difference between adjacent control periods. Simultaneously, it obtains at least one of the drive current or torque commands from the drive control link to reflect changes in the intensity of the drive load and wheel-to-ground interaction. Subsequently, the above features are combined with time delay and communication quality parameters to form model input features, which are then input into a self-supervised residual prediction model for inference under a unified time alignment reference.

[0034] 202. Construct observable residuals based on actual motion states, and use the observable residuals as the self-supervised learning target of the self-supervised residual prediction model; wherein the observable residuals include wheel synchronization error, or include composite residuals obtained by fusing wheel synchronization error and yaw residuals, and the yaw residuals are residuals constructed from the kinematic relationship between the measurement of body heading change and the actual speed of the left and right drive units; The inter-wheel synchronization error is constructed using the actual speeds of the left and right drive units. This error characterizes the degree of synchronization between the left and right drives under straight-line walking conditions. To enhance the observability of yaw and deviation trends, a measurement of the aircraft's heading change is introduced. This measurement, combined with the kinematic relationship of the actual speeds of the left and right drive units, constructs a yaw residual. This yaw residual characterizes the degree of deviation between the actual heading change and kinematic consistency. The inter-wheel synchronization error is used as the observable residual, or a composite residual is formed by fusing the inter-wheel synchronization error and the yaw residual according to a preset fusion rule. This allows the observable residual to simultaneously reflect both wheel asynchrony and heading consistency deviation. During model training or online self-supervised update, the observable residual serves as the learning objective, providing the model with a supervisory signal that requires no additional manual annotation.

[0035] 203. Based on the self-supervised residual prediction model, predict the observable residuals or their changing trends for the next control period to obtain prediction results, wherein the prediction results include the predicted values ​​of the observable residuals and / or the predicted values ​​of the changing trends and / or parameters characterizing the prediction distribution. After obtaining the model input features, forward inference is performed to predict the observable residuals for the next control cycle, yielding residual prediction values. Simultaneously, the trend of the observable residuals is predicted to characterize dynamic features such as residual growth, decay, or direction reversal. To support uncertainty calculation, the prediction results can also include parameters characterizing the prediction distribution, such as a set of parameters or equivalent distribution descriptors to describe the degree of prediction dispersion, providing both point estimates and distribution information.

[0036] 204. Calculate the prediction uncertainty based on the prediction results; Uncertainty-related quantities are extracted from the prediction results, and the prediction uncertainty is calculated. Prediction uncertainty can be obtained directly from parameters characterizing the prediction distribution, or from the prediction interval width or equivalent dispersion. When using randomized inference, prediction uncertainty can also be calculated from the output dispersion obtained from multiple forward calculations of the same input. Prediction uncertainty is used to characterize the reliability of the prediction results and subsequently serves as a control reliability parameter in control mode determination and synchronous control weight determination.

[0037] 205. Generate a slip risk indicator by jointly mapping the prediction results and prediction uncertainty, wherein the joint mapping is at least related to the magnitude and / or trend of the predicted observable residual values ​​in the prediction results.

[0038] By jointly mapping the prediction results with the prediction uncertainty, a slip risk indicator is obtained. The joint mapping is at least related to the magnitude of the predicted observable residual values, used to characterize the upward trend of slip risk as the residual level deviation increases; the joint mapping can also be related to the trend of observable residual changes, used to characterize the upward trend of risk when the residuals increase rapidly or change direction abruptly. The joint mapping also introduces prediction uncertainty, allowing the slip risk indicator to be output in a more conservative manner when the confidence level is low, avoiding the formation of a radical control decision basis under uncertain conditions.

[0039] In this embodiment, the actual motion state is constructed as the model input feature and input into the self-supervised residual prediction model. Observable residuals are constructed based on the actual motion state and used as the self-supervised learning objective. The model predicts the observable residuals or their changing trends for the next control cycle to obtain the prediction result. The prediction uncertainty is calculated based on the prediction result, and a slip risk indicator is generated by jointly mapping the prediction result and the prediction uncertainty. This allows slip risk and reliability to be driven by the observable residuals and their dynamic trends, and continuously obtains discrimination criteria related to wheel-ground conditions in a self-supervised manner. This provides a more stable and interpretable risk and uncertainty input for subsequent control mode determination and synchronous control weight scheduling, improving the reliability of control decisions under complex ground conditions. Furthermore, yaw residuals are introduced and fused with wheel-to-wheel synchronization errors to form composite residuals, enabling risk discrimination to simultaneously cover both wheel-to-wheel asynchrony and heading consistency deviations, improving the ability to identify coupled yaw and slip conditions.

[0040] Please see Figure 3 In some embodiments of this application, step 102 in the above embodiments, which estimates the delay based on time information and generates communication quality parameters, may include the following steps: 301. Obtain the sender's timestamp, receiver's timestamp, and message sequence number corresponding to the control command as time information; When receiving control commands, the sender's timestamp, receiver's timestamp, and message sequence number are synchronously parsed and recorded. The sender's timestamp reflects the time when the command was generated or sent at the sender, the receiver's timestamp reflects the time when the command arrived at the receiver, and the message sequence number reflects the order of the commands in the transmission sequence. Using the message sequence number, out-of-order arrivals, duplicate messages, and missing messages can be identified, providing a foundation for subsequent delay sample construction and link reliability assessment.

[0041] 302. Estimate the clock skew and clock drift between the master controller and slave controller based on time information, and obtain the clock skew parameter and clock drift parameter; A mapping relationship between the transmitting and receiving clocks is constructed using the timing information of multiple consecutive control commands, and clock skew and clock drift are estimated within a sliding window. In engineering terms, clock skew corresponds to a fixed difference between the clocks at both ends, while clock drift corresponds to the difference in their running rates. The estimation process can employ least-squares fitting, recursive estimation, or robust fitting to ensure that the clock skew and clock drift parameters are updated over time and adapt to changes in clock characteristics caused by factors such as temperature drift and load variations.

[0042] 303. Correct the sending end timestamp based on the clock skew parameter and clock drift parameter to obtain the corrected timestamp; Based on the clock skew and clock drift parameters, the transmitter's timestamp is transformed to the receiver's time base to obtain the corrected timestamp. Since the corrected timestamp and the receiver's timestamp share the same time base, this reduces the deviation in delay calculation caused by clock inconsistencies at both ends, making subsequent delay estimates closer to the actual link delay.

[0043] 304. Calculate the delay sample sequence based on the receiver timestamp and the correction timestamp; A single delay sample is obtained by subtracting the correction timestamp from the receiver's timestamp, and these samples are accumulated according to the control cycle to form a delay sample sequence. Combined with the message sequence number, duplicate messages can be removed, out-of-order messages can be processed, and missing messages can be registered for loss, thus preventing abnormal messages from polluting the delay statistics.

[0044] 305. Determining time delay uncertainty based on time delay sample sequences; Delay uncertainty is obtained by statistically analyzing the delay sample sequence within a preset statistical window. Delay uncertainty can be characterized by the degree of delay dispersion, such as standard deviation, quantile interval width, or robust dispersion index; it can also be characterized by measures such as confidence interval width to depict the reliability range of the delay estimate under the current link state. Through delay uncertainty, factors such as link jitter and sudden congestion are explicitly quantified and transmitted to subsequent control decision-making stages.

[0045] 306. Generate initial communication quality parameters based on time delay sample sequences and time delay uncertainty.

[0046] Initial communication quality parameters are generated based on the delay sample sequence and delay uncertainty to characterize the real-time performance and stability of the link. In implementation, the delay mean, delay dispersion, and delay uncertainty can be fused to form scalar or vector forms of the initial communication quality parameters; alternatively, packet loss correlation quantities obtained from message sequence number statistics can be incorporated into the fusion relationship, allowing the initial communication quality parameters to simultaneously cover both delay level and delay fluctuation characteristics, providing input for subsequent control mode determination and control reliability assessment.

[0047] In this embodiment, time information is constructed using the sender's timestamp, receiver's timestamp, and message sequence number. Clock skew and clock drift are further estimated, and sender timestamp correction is performed. Then, the receiver's timestamp and the corrected timestamp form a delay sample sequence, and the delay uncertainty is determined. This allows the link delay to be stably and traceably quantified under a unified time base, thereby improving the accuracy and robustness of communication delay assessment and reducing the interference of clock inconsistencies and out-of-order packet loss on control decisions. Furthermore, initial communication quality parameters are generated based on the delay sample sequence and delay uncertainty, enabling a comprehensive characterization of link real-time performance and volatility, providing a more reliable link status basis for subsequent control mode scheduling and correction intensity adjustment.

[0048] Please see Figure 4 In some embodiments of this application, step 306 in the above embodiments, which generates initial communication quality parameters based on the time delay sample sequence and time delay uncertainty, may include the following steps: 401. Perform robust statistics on the time delay sample sequence within the preset statistical window to obtain the time delay dispersion index; Within a preset statistical window, robust statistics are performed on the latency sample sequences to obtain a latency dispersion index, which is used to characterize the degree of latency fluctuation. Robust statistics can use dispersion measures that are insensitive to outliers, such as median absolute deviation, truncated standard deviation, or quantile difference, so that extreme latency samples caused by sudden congestion do not dominate the dispersion results, thus reflecting the link jitter level more stably.

[0049] 402. Determine the packet loss rate based on the message sequence number; The continuity of message sequence numbers is used to determine message loss, and the packet loss rate is obtained by statistically analyzing the ratio of missing messages to the total number of messages. Out-of-order and duplicate messages can be identified through sequence number relationships and are removed or merged during the statistical analysis to avoid misjudging the packet loss rate. The packet loss rate is used to characterize link reliability; an increased packet loss rate usually indicates a decrease in the effective refresh capability of control commands.

[0050] 403. Determine the information freshness index based on the receiver timestamp, where the information freshness index is the difference between the current control time and the timestamp corresponding to the latest successfully received control command; The information freshness index reflects the recent validity of control commands. During a continuous control cycle, the timestamp corresponding to the most recently successfully received control command is recorded, and the information freshness index is obtained by subtracting this timestamp from the current control time. When the link experiences prolonged periods without updates or continuous packet loss, the information freshness index increases, indicating that the command information used by the control end is gradually becoming outdated, which can easily lead to delayed control responses or deviations from external expectations.

[0051] 404. Generate initial communication quality parameters based on the delay dispersion index, packet loss rate index, information freshness index, and delay uncertainty; Initial communication quality parameters are obtained by fusing latency dispersion, packet loss rate, information freshness, and latency uncertainty. The fusion method can employ weighted summation, segmented mapping, or lookup table mapping, allowing the initial communication quality parameters to comprehensively reflect link fluctuations, link reliability, and command update effectiveness. Introducing latency uncertainty enables the initial communication quality parameters to reflect changes in the reliability range of latency assessment, making subsequent control decisions more conservative when uncertainty increases.

[0052] 405. Perform exponential smoothing on the initial communication quality parameters to obtain smoothed communication quality parameters; Exponential smoothing is applied to the initial communication quality parameters to make them insensitive to short-term spikes and gradually adapt as the link status continues to change. Exponential smoothing can be achieved through a recursive approach, ensuring that the smoothed communication quality parameters balance real-time performance and stability, and reducing the risk of frequent handovers triggered by momentary jitter.

[0053] 406. Perform hysteresis determination on the smooth communication quality parameters to obtain stable communication quality parameters.

[0054] To further suppress state jitter caused by frequent fluctuations, hysteresis judgment is performed on the smoothed communication quality parameters to obtain stable communication quality parameters. Hysteresis judgment can set entry and exit thresholds to prevent the communication quality parameters from frequently exceeding the limits when fluctuating around the thresholds, thereby reducing the probability of jitter in subsequent control mode determination and strategy switching, and improving the continuity of control link decisions.

[0055] In this embodiment, the control device performs robust statistics on the delay sample sequence within a preset statistical window to obtain a delay dispersion index, determines a packet loss rate index based on the message sequence number, and determines an information freshness index based on the receiver timestamp. It then generates initial communication quality parameters based on the delay dispersion index, packet loss rate index, information freshness index, and delay uncertainty. This allows for a unified quantification and characterization of link fluctuations, reliability, and instruction update effectiveness, thereby improving the stability and availability of communication quality assessment. Furthermore, the control device performs exponential smoothing on the initial communication quality parameters and further performs hysteresis judgment to obtain stable communication quality parameters. This reduces the risk of parameter jitter and frequent strategy switching caused by transient fluctuations in communication status, providing a more continuous and reliable link status basis for subsequent control mode scheduling and correction intensity adjustment.

[0056] Please see Figure 5 In some embodiments of this application, step 106 in the above embodiments, which generates a synchronization control quantity based on the prediction error and the control mode, may include the following steps: 501. Determine the set of synchronous control structure parameters corresponding to the control mode based on the control mode; Once the control mode is determined, the parameter scheduling process begins. The control device selects a set of synchronous control structure parameters based on the control mode, ensuring that the control law parameters match the control objective under different operating conditions. The synchronous control structure parameter set may include one or more of the proportional gain, integral gain, and derivative gain, as well as error filter coefficients, integral limiting parameters, and output limiting parameters. This is used to balance correction accuracy and dynamic stability when communication status changes, slip risk changes, or prediction uncertainty changes.

[0057] 502. Calculate the change in error based on the prediction error; After obtaining the prediction error for the current control cycle, the prediction errors for adjacent control cycles are estimated using differential or equivalent methods to obtain the error change. The error change is used to characterize the dynamic rate of change of the prediction error, enabling the control law to respond more promptly to situations such as rapid error increases or rapid reversals. To reduce the amplification effect of measurement noise on the differential results, filtering, limiting, or robust differential strategies can be introduced into the calculation of the error change.

[0058] 503. Input the prediction error and the error change into the control law corresponding to the synchronous control structure parameter set to obtain the synchronous control quantity.

[0059] Given a fixed set of parameters for the synchronous control structure, the prediction error and error variation are substituted into the corresponding control law to output the synchronous control quantity. This synchronous control quantity is used to correct wheel asynchrony. The control law can be a proportional-derivative, proportional-integral, or proportional-integral-derivative structure, or an equivalent discrete control form. The parameter set of the synchronous control structure determines the gain and adjustment speed of the control law under the current control mode, ensuring that the synchronous correction output remains stable and controllable under different operating conditions.

[0060] In this embodiment, the control device schedules the set of synchronous control structure parameters according to the control mode, and extracts the error change from the prediction error and substitutes it into the control law to obtain the synchronous control quantity. This allows the parameters of synchronous correction to be adaptively adjusted as the working conditions change. At the same time, the error change is used to enhance the response capability to the dynamic changes of error, thereby improving the timeliness and stability of the wheel-to-wheel synchronous control and reducing the risk of deviation and oscillation.

[0061] Please see Figure 6 In some embodiments of this application, after step 108 in the above embodiments allocates the weighted synchronous control quantity as the target control command for the left and right drive units under preset constraints, the robot motion control method may further include the following steps: 601. Determine the update strategy for target control commands based on communication quality parameters, wherein the update strategy includes a periodic update strategy or an event-triggered update strategy; Upon receiving the target control command, an update strategy is introduced to manage the issuance rhythm of the target control command, ensuring that the issuance behavior matches the communication link status. When communication quality parameters are at a relatively high level, the update strategy can be determined as a periodic update strategy, that is, continuously outputting the target control command according to a preset control cycle to maintain a high refresh rate and high tracking performance. When communication quality parameters deteriorate, the update strategy can be determined as an event-triggered update strategy to reduce the risk of invalid refreshes and command jitter under link jitter and packet loss conditions. The event-triggered update strategy can determine whether to issue an update based on trigger conditions, such as the change in the target control command exceeding a threshold, communication quality parameters crossing a threshold, or reaching the maximum hold interval; through trigger conditions, the update behavior is associated with the control effectiveness, and the expiration of commands due to prolonged inactivity is avoided.

[0062] In this embodiment, step 109, outputting the target control command to the drive control unit for execution, may specifically include: 602. Output the target control command to the drive control unit for execution according to the update strategy.

[0063] Under the periodic update strategy, target control commands are output according to a preset period. Under the event-triggered update strategy, target control commands are output when the trigger condition is met, and the target control commands output at the previous moment are retained when the trigger condition is not met. The output target control commands are executed in a closed loop by the drive control unit and continuously generate actual motion state feedback to support error construction, mode determination, and synchronous control calculation in subsequent control cycles.

[0064] In this embodiment, the control device determines the update strategy of the target control command based on the communication quality parameters, and outputs the target control command to the drive control unit for execution according to the update strategy. This allows the refresh rate of the target control command to be dynamically matched with the real-time performance and reliability of the link. When the communication quality is good, a high refresh rate is maintained to improve the tracking performance. When the communication quality deteriorates, event-triggered updates are used to reduce the risk of invalid refreshes and command jitter, thereby improving the stability and effectiveness of the target control command issuance and reducing the risk of motion control jitter and deviation under link fluctuation conditions.

[0065] Please see Figure 7 In some embodiments of this application, step 107 in the above embodiments, which determines the synchronization control weight of the synchronization control quantity based on the control mode and at least one control reliability parameter, may include the following steps: 701. Determine the weight benchmark value based on the control mode; After the control mode is determined, the weight baseline value setting process begins. The weight baseline value is used to specify the basic strength of the synchronous correction effect under the current operating conditions. The control device can preset different weight baseline values ​​for different control modes, so that the weight baseline value is larger in the mode with stable communication and low risk to enhance synchronous correction; and the weight baseline value is smaller in the mode corresponding to increased slip risk, increased uncertainty, or link degradation to suppress jitter or oscillation caused by aggressive correction.

[0066] 702. Determine the weight correction amount based on at least one control reliability parameter; Based on the baseline weighted values, control reliability parameters are introduced to determine the weight correction amount, which reflects the real-time reliability of the link and operating conditions. Control reliability parameters include at least one or more of the following: communication quality parameters, slip risk indicators, and prediction uncertainty. The weight correction amount can be obtained by mapping the control reliability parameters, for example, using piecewise thresholds, linear mapping, or lookup table mapping, so that the weight correction amount is negative when communication quality deteriorates, slip risk increases, or prediction uncertainty increases; or the weight correction amount can tend to zero or produce a positive correction when risk decreases or communication improves, thereby achieving dynamic convergence of the weight to a more suitable range with increasing reliability. To avoid weight jitter, the calculation of the weight correction amount can also introduce amplitude limits or rate of change constraints.

[0067] 703. Determine the synchronous control weight based on the weight benchmark value and the weight adjustment amount.

[0068] The synchronization control weight is obtained by combining the baseline weight value and the weight correction amount. The synchronization control weight is used to subsequently adjust the synchronization control amount. The combination method can be summation and limiting, segmented combination, or equivalent fusion method, so that the synchronization control weight is kept within a preset range and meets the monotonicity and conservatism requirement, that is, the weight does not increase when the reliability decreases and does not decrease when the reliability increases, thereby improving the predictability and stability of the correction intensity adjustment.

[0069] In this embodiment, the control device determines the weight reference value according to the control mode and the weight correction amount based on the control reliability parameters. Then, the weight reference value and the weight correction amount determine the synchronization control weight, so that the synchronization correction strength is constrained by both the operating mode and the link / slip reliability. When the risk increases or the link deteriorates, it automatically tends to be conservative, and when the risk decreases or the link improves, it restores the correction capability, thereby improving the stability and consistency of the linear travel synchronization control and reducing the control oscillation risk caused by excessive correction and weight jitter.

[0070] In the foregoing embodiments, the specific implementation process of the method provided in this application has been described in detail. It should be understood that, to implement the above method, this application also provides corresponding devices and computer-readable storage media, the technical concepts of which are consistent with the foregoing method embodiments, and both are used to implement all or part of the steps in the foregoing method. The technical solutions of the devices and computer-readable storage media involved in this application will be further described below in conjunction with specific embodiments.

[0071] Please see Figure 8 , Figure 8 One embodiment of the robot motion control device provided in this application is used to achieve, as Figures 1 to 7 The robot motion control method in any possible implementation of the illustrated embodiment includes: The receiving unit 801 is used to receive control commands and their timing information, and to obtain the actual motion state of the left and right drive units; The first generation unit 802 is used to estimate the delay based on time information and generate communication quality parameters; The construction unit 803 is used to construct the inter-wheel synchronization error based on the actual motion state and perform prediction compensation to obtain the prediction error; Input unit 804 is used to input the actual motion state, time delay and communication quality parameters into the trained self-supervised residual prediction model to obtain the slip risk indication and prediction uncertainty; The first determining unit 805 is used to determine the control mode based on the prediction error, communication quality parameters, slip risk indication and prediction uncertainty. The second generation unit 806 is used to generate synchronous control quantities based on prediction errors and control modes; The second determining unit 807 is used to determine the synchronization control weight of the synchronization control quantity based on the control mode and at least one control reliability parameter, wherein the control reliability parameter is at least one of the communication quality parameter, slip risk indication and / or prediction uncertainty. The allocation unit 808 is used to adjust the synchronization control quantity according to the synchronization control weight, and allocate the weighted synchronization control quantity as the target control command of the left and right drive units under preset constraints. Output unit 809 is used to output target control commands to drive control unit for execution.

[0072] In this implementation, the functions of each unit are as described above. Figures 1 to 7 The steps in the illustrated embodiments are the same and will not be repeated here.

[0073] Please see Figure 9 , Figure 9 One embodiment of the electronic device provided in this application includes: Processor 901, memory 902, input / output unit 903, and bus 904; The processor 901 is connected to the memory 902, the input / output unit 903, and the bus 904; The memory 902 stores a program, which the processor 901 calls to execute. Figures 1 to 7 The steps in the illustrated embodiment.

[0074] In this embodiment, the function of processor 901 is the same as described above. Figures 1 to 7 The steps in the illustrated embodiments are the same and will not be repeated here.

[0075] This application also provides a computer-readable storage medium on which a program is stored. When the program is executed on a computer, it causes the computer to perform the aforementioned actions. Figures 1 to 7 The method in any possible implementation.

[0076] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0080] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A robot motion control method, characterized in that, include: Receive control commands and their timing information, and obtain the actual motion state of the left and right drive units; Based on the time information, the delay is estimated and communication quality parameters are generated; Based on the actual motion state, the inter-wheel synchronization error is constructed and predicted and compensated to obtain the prediction error; The actual motion state, the time delay, and the communication quality parameters are input into the trained self-supervised residual prediction model to obtain the slip risk indicator and prediction uncertainty. The control mode is determined based on the prediction error, the communication quality parameters, the slip risk indicator, and the prediction uncertainty. A synchronous control quantity is generated based on the prediction error and the control mode; The synchronization control weight of the synchronization control quantity is determined based on the control mode and at least one control reliability parameter, wherein the control reliability parameter is at least one of the communication quality parameter, the slip risk indicator and / or the prediction uncertainty. The synchronization control quantity is weighted and adjusted according to the synchronization control weight, and the weighted synchronization control quantity is allocated as the target control command of the left and right drive units under preset constraints. The target control command is output to the drive control unit for execution.

2. The robot motion control method according to claim 1, characterized in that, The step of inputting the actual motion state, the time delay, and the communication quality parameters into a trained self-supervised residual prediction model to obtain the slip risk indicator and prediction uncertainty includes: The actual motion state is constructed as the model input feature and input into the self-supervised residual prediction model, wherein the model input feature includes at least one of the actual speed and its change of the left and right drive units, drive current or torque command; An observable residual is constructed based on the actual motion state, and the observable residual is used as the self-supervised learning target of the self-supervised residual prediction model; wherein the observable residual includes wheel synchronization error, or includes a composite residual obtained by fusing wheel synchronization error and yaw residual, and the yaw residual is a residual constructed from the kinematic relationship between the body heading change measurement and the actual speed of the left and right drive units; Based on the self-supervised residual prediction model, the observable residual or its changing trend for the next control period is predicted to obtain the prediction result, wherein the prediction result includes the predicted value of the observable residual and / or the predicted value of the changing trend and / or parameters characterizing the prediction distribution. Based on the prediction results, a slip risk indicator and prediction uncertainty are generated.

3. The robot motion control method according to claim 2, characterized in that, The generation of the slip risk indicator and prediction uncertainty based on the prediction results includes: Calculate the prediction uncertainty based on the prediction results; The slip risk indicator is generated by jointly mapping the prediction results and the prediction uncertainty, wherein the joint mapping is at least related to the magnitude and / or trend of the predicted observable residual values ​​in the prediction results.

4. The robot motion control method according to claim 1, characterized in that, The step of estimating the delay and generating communication quality parameters based on the time information includes: The sending end timestamp, receiving end timestamp, and message sequence number corresponding to the control command are obtained as the time information. Based on the time information, the clock deviation and clock drift between the master controller and the slave controller are estimated to obtain the clock deviation parameter and the clock drift parameter; The timestamp of the transmitting end is corrected based on the clock deviation parameter and the clock drift parameter to obtain the corrected timestamp; Calculate the delay sample sequence based on the receiving end timestamp and the correction timestamp; Determine the time delay uncertainty based on the time delay sample sequence; Initial communication quality parameters are generated based on the time delay sample sequence and the time delay uncertainty.

5. The robot motion control method according to claim 4, characterized in that, The process of generating initial communication quality parameters based on the time delay sample sequence and the time delay uncertainty includes: Robust statistics are performed on the time delay sample sequence within a preset statistical window to obtain the time delay dispersion index; The packet loss rate is determined based on the message sequence number; The information freshness index is determined based on the timestamp of the receiving end, wherein the information freshness index is the difference between the current control time and the timestamp corresponding to the latest successfully received control instruction; Initial communication quality parameters are generated based on the delay dispersion index, the packet loss rate index, the information freshness index, and the delay uncertainty. The initial communication quality parameters are subjected to exponential smoothing to obtain smoothed communication quality parameters; Hysteresis determination is performed on the smooth communication quality parameters to obtain stable communication quality parameters.

6. The robot motion control method according to claim 1, characterized in that, The generation of synchronous control quantities based on the prediction error and the control mode includes: Determine the set of synchronous control structure parameters corresponding to the control mode based on the control mode; The change in error is calculated based on the prediction error; The prediction error and the error change are input into the control law corresponding to the synchronous control structure parameter set to obtain the synchronous control quantity.

7. The robot motion control method according to any one of claims 1-6, characterized in that, After allocating the weighted synchronous control quantity as target control commands for the left and right drive units under preset constraints, the method further includes: The update strategy for the target control command is determined based on the communication quality parameters, wherein the update strategy includes a periodic update strategy or an event-triggered update strategy; The step of outputting the target control command to the drive control unit for execution includes: The target control command is output to the drive control unit for execution according to the update strategy.

8. The robot motion control method according to any one of claims 1-6, characterized in that, The process of determining the synchronization control weight of the synchronization control quantity based on the control mode and at least one control reliability parameter includes: The weight benchmark value is determined based on the control mode; The weight correction amount is determined based on at least one control reliability parameter; The synchronization control weight is determined based on the weight baseline value and the weight adjustment amount.

9. A robot motion control device, characterized in that, For implementing the robot motion control method as described in any one of claims 1 to 8, the robot motion control device comprises: Receive control commands and their timing information, and obtain the actual motion state of the left and right drive units; Based on the time information, the delay is estimated and communication quality parameters are generated; Based on the actual motion state, the inter-wheel synchronization error is constructed and predicted and compensated to obtain the prediction error; The actual motion state, the time delay, and the communication quality parameters are input into the trained self-supervised residual prediction model to obtain the slip risk indicator and prediction uncertainty. The control mode is determined based on the prediction error, the communication quality parameters, the slip risk indicator, and the prediction uncertainty. A synchronous control quantity is generated based on the prediction error and the control mode; The synchronization control weight of the synchronization control quantity is determined based on the control mode and at least one control reliability parameter, wherein the control reliability parameter is at least one of the communication quality parameter, the slip risk indicator and / or the prediction uncertainty. The synchronization control quantity is weighted and adjusted according to the synchronization control weight, and the weighted synchronization control quantity is allocated as the target control command of the left and right drive units under preset constraints. The target control command is output to the drive control unit for execution.

10. An electronic device, characterized in that, include: Processor, memory, input / output units, and bus; The processor is connected to the memory, the input / output unit, and the bus; The memory stores a program, and the processor calls the program to execute the method as described in any one of claims 1 to 8.

Images