Prospective preprocessing method and system in six-axis mechanical motion automatic control

Abstract

The invention discloses a prospective preprocessing method and system in six-axis mechanical motion automatic control, particularly relates to the field of six-axis mechanical motion automatic control, and is used for solving the problem of insufficient track prediction and control precision in a load sudden change scene. The inertia tensor is updated in real time in combination with dual-channel Kalman filtering, dynamic inertia data are accessed, and the sensitivity of trajectory prediction to load change is improved; double-domain feature extraction of time-domain thermal stress density and frequency-domain energy dispersion is adopted, acceleration pulse layout is dynamically optimized by using a complementary integral mechanism, and the overload risk of a driver is reduced; meanwhile, the instruction dispatcher avoids cache overflow through advanced rhythm regulation and control, and continuity of instruction streams is guaranteed; the inertia model is continuously refreshed through closed-loop feedback, and synchronous evolution of inertia change and trajectory prediction is achieved; the overload shutdown risk caused by inertia lag of six-axis mechanical motion is reduced, and the adaptive capacity of the prospective preprocessing method in a dynamic complex operation environment is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of six-axis mechanical motion automation control. More specifically, the present invention relates to a prospective preprocessing method and system in six-axis mechanical motion automation control. Background Art

[0002] In complex collaborative tasks, six-axis mechanical motion control relies on the coordinated operation of high-frequency closed-loop response and prospective preprocessing mechanisms. The currently widely used prospective preprocessing method predicts the target trajectory and dynamic parameters in the next several control cycles in advance, and generates a continuous instruction stream based on the modeled mechanical behavior to meet the requirements of trajectory accuracy, response speed, and multi-axis collaborative stability.

[0003] However, in scenarios with sudden load changes such as grasping, the inertia characteristics of the end load change drastically, and such changes often fail to complete the mapping update within the control cycle. Due to the lagging update of the inertia parameters, the preprocessing module continues to perform predictions using the expired model, which is extremely likely to cause servo drive current shocks in a short time, resulting in drive overload shutdown, electrical damage, and even motion interruption. This problem of lagging load transient mapping is precisely the key technical bottleneck in the prospective preprocessing link of six-axis mechanical automation control.

[0004] To solve the above problems, a technical solution is provided. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a prospective preprocessing method and system in six-axis mechanical motion automation control. By setting load impact description information, combining dual-channel Kalman filtering to update the inertia tensor in real time, accessing dynamic inertia data, and enhancing the sensitivity of trajectory prediction to load changes; adopting dual-domain feature extraction of time-domain thermal stress density and frequency-domain energy dispersion, and using a complementary integration mechanism to dynamically optimize the acceleration pulse layout to reduce the risk of drive overload; at the same time, the instruction distributor avoids buffer overflow through advanced beat control to ensure the continuity of the instruction stream; the closed-loop feedback continuously refreshes the inertia model to achieve synchronous evolution of inertia changes and trajectory prediction; reducing the risk of overload shutdown of six-axis mechanical motion caused by inertia lag, and enhancing the adaptability of the prospective preprocessing method in a dynamic and complex operating environment to solve the problems raised in the above background art.

[0006] To achieve the above object, the present invention provides the following technical solutions: A prospective preprocessing method in six-axis mechanical motion automation control, comprising the steps of: S1: When the jaw pressure exceeds the set threshold, generate load impact description information and write it with a time stamp into the inertia estimation channel; S2: The inertia estimation channel adopts dual-channel Kalman filtering, fuses the servo current data and the encoder angular velocity data, and outputs the updated inertia tensor data packet within half a control cycle; S3: The prediction cache replaces the original inertia model parameters with the inertia tensor data packet, and re-solves the acceleration vector and jerk vector of the next time slot according to the sliding window method to form a trajectory segment data set; S4: The power mapper maps the trajectory segment data set to the drive thermal power surface, calculates the thermal stress density in the time domain and the energy dispersion in the frequency domain, and generates an energy shock decision quantity using the dual-domain complementary integration algorithm; if the energy shock decision quantity is higher than the preset threshold, the acceleration pulses in the trajectory are split into multiple micro-pulse segments, and their time order is rearranged; S5: The instruction distributor adjusts the cache writing rhythm according to the instruction stack prediction function, and pushes the rearranged instruction stream to the drive end according to the preset lead depth to prevent cache overflow; S6: The feedback rectifier collects the real-time torque error and pose error, and inputs them into the Bayesian integration module to refresh the prior model of the inertia tensor, providing a correction basis for generating the load shock description information in the next cycle.

[0007] In a preferred embodiment, step S1 includes the following contents: The gripper pressure is monitored in real time through a pressure sensor or a torque sensor installed on the gripper, and the acquisition frequency is consistent with the control cycle; when the gripper pressure exceeds the pressure threshold determined according to the design parameters, load capacity and historical operation data of the six-axis robotic arm, load shock description information is generated, and the load shock description information is stored in a structured data format and includes the gripper pressure value and the gripper status identifier; at the same time, the moment when the load mutation occurs is recorded as a timestamp, and the timestamp is aligned with the time reference of the control cycle; after binding the load shock description information and the timestamp, it is written into the inertia estimation channel.

[0008] In a preferred embodiment, step S2 includes the following contents: Receive the load shock description information and timestamp transmitted by the inertia estimation channel; collect the servo current data and encoder angular velocity data within the current control cycle; use a dual-channel Kalman filter, where the current channel estimates the inertia tensor according to the servo current data and the mapping relationship between current and torque, and the angular velocity channel assists in correcting the inertia tensor estimation according to the encoder angular velocity data and kinematic relationship. The initial state of the dual-channel Kalman filter is provided by the load shock description information. The dual-channel Kalman filter calculates the prior inertia tensor estimation based on the inertia tensor estimation value and process noise at the previous moment in the prediction stage, and corrects the prior inertia tensor estimation using the servo current data and encoder angular velocity data in the measurement update stage. The correction amplitude is determined by the dynamically adjusted Kalman gain matrix.

[0009] In a preferred embodiment, step S2 includes the following: Fuse the estimation results of the current channel and the angular velocity channel, and use the confidence-weighted fusion method to generate the fused inertia tensor estimation; impose physical constraints on the fused inertia tensor estimation to ensure that each component of the inertia tensor is non-negative and conforms to the dynamic range of the robotic arm; encapsulate the fused inertia tensor estimation into an inertia tensor data packet, including each component of the inertia tensor and the timestamp, and transmit and output it within half of the control period.

[0010] In a preferred embodiment, step S3 includes the following: Replace the original inertia model parameter set in the prediction cache module with the updated inertia tensor, and calibrate the timing of the inertia parameter update according to the timestamp; configure the window width and step size using the sliding window method, divide the future time into multiple time slots, and the next time slot is the first prediction point after the current time; construct the dynamic equation based on the updated inertia model parameter set, and use the historical data and the current state within the sliding window to calculate the acceleration vector of the next time slot using the Newton-Euler method. Specifically, subtract the Coriolis force, centrifugal force terms, and gravity term from the torque vector at the current moment, and then process it through the inverse matrix of the inertia tensor to obtain the acceleration vector; calculate the jerk vector by taking the time derivative of the acceleration vector; arrange the predicted acceleration vector and jerk vector in chronological order as a trajectory segment data set.

[0011] In a preferred embodiment, step S4 includes the following: Convert the acceleration vector and jerk vector in the trajectory segment data set into the thermal power values of the driver through a power mapper to form a thermal power surface.

[0012] In a preferred embodiment, step S4 further includes the following: Use the locally weighted average method to calculate the time-domain thermal stress density. Select a fixed time window at each time point, use the Gaussian kernel function to perform weighted averaging on the thermal power values within the time window, and normalize the result to the range of 0 to 1; Perform a fast Fourier transform on the thermal power surface to obtain the spectrum, and use the spectral entropy method to calculate the frequency-domain energy dispersion and normalize it to the range of 0 to 1; Multiply the time-domain thermal stress density by the frequency-domain energy concentration through the dual-domain complementary integration algorithm and integrate with respect to time to generate an energy shock judgment quantity; if the energy shock judgment quantity exceeds the preset energy shock threshold, identify the peak points in the acceleration vector, split them into multiple micro-pulse segments, and rearrange the time order.

[0013] In a preferred embodiment, step S5 includes the following: Convert the rearranged trajectory data into an instruction stream executable by the driving end. The instruction stream is arranged in chronological order and each instruction corresponds to a control cycle. Calculate the instruction accumulation prediction function using the moving average method, average the instruction data volume in the recent several control cycles and normalize it to the range of 0 to 1. Dynamically adjust the cache writing beat according to the instruction accumulation prediction function and the driving end cache capacity. Specifically, the cache writing beat is equal to the basic writing rate multiplied by a ratio of 1 minus the value of the instruction accumulation prediction function to the driving end cache capacity. Write the instruction stream into the driving end cache according to the adjusted cache writing beat, and ensure that the driving end cache stores instructions for a preset leading depth of control cycles. Real-time monitor the occupancy rate of the driving end cache. If the occupancy rate of the driving end cache exceeds the preset threshold, pause writing and issue an alarm.

[0014] In a preferred embodiment, step S6 includes the following contents: Collect real-time torque error and real-time pose error from the driving end. The real-time torque error is the deviation between the actual torque and the instruction torque in the instruction stream, and the real-time pose error is the deviation between the actual pose and the instruction pose in the instruction stream. Apply low-pass filtering to the real-time torque error and real-time pose error to generate smooth torque error signals and pose error signals. Convert the smooth torque error signals and pose error signals into inertia tensor correction amounts through a feedback rectifier. Calculate the inertia tensor correction amounts using a dynamic gain adjustment strategy, including a proportional term of the torque error, a change rate term of the torque error, and a gain term of the pose error. Use a Bayesian integration module to fuse the inertia tensor prior model and the inertia tensor correction amounts to generate an inertia tensor posterior model. Specifically, add the inertia tensor prior model and the inertia tensor correction amounts and multiply by the conditional probability based on the pose error signal. Assign the inertia tensor posterior model to the inertia tensor prior model of the next moment and transfer it to the inertia estimation channel to provide a correction basis for generating load impact description information in the next cycle.

[0015] A forward-looking preprocessing system in six-axis mechanical motion automation control includes: a load detection module, an inertia update module, a trajectory generation module, an energy optimization module, an instruction push module, and a feedback correction module. Load detection module: When the jaw pressure exceeds the set threshold, generate load impact description information and append a timestamp to write it into the inertia estimation channel. Inertia update module: The inertia estimation channel uses dual-channel Kalman filtering to fuse servo current data and encoder angular velocity data, and outputs an updated inertia tensor data packet within half a control cycle. Trajectory Generation Module: The prediction cache replaces the original inertia model parameters with inertia tensor data packets, and re-solves the acceleration vector and jerk vector of the next time slot according to the sliding window method to form a trajectory segment data set; Energy Optimization Module: The power mapper maps the trajectory segment data set to the drive thermal power surface, calculates the thermal stress density in the time domain and the energy dispersion in the frequency domain, and uses the dual-domain complementary integral algorithm to generate an energy shock decision quantity; if the energy shock decision quantity is higher than the preset threshold, the acceleration pulses in the trajectory are split into multiple micro-pulse segments and their time order is rearranged; Instruction Push Module: The instruction distributor adjusts the cache writing rhythm according to the instruction accumulation prediction function, and pushes the rearranged instruction stream to the drive end according to the preset lead depth to prevent cache overflow; Feedback Correction Module: The feedback rectifier collects the real-time torque error and pose error, and inputs them into the Bayesian integration module to refresh the prior model of the inertia tensor, providing a correction basis for generating load shock description information in the next cycle.

[0016] Technical effects and advantages of the prospective preprocessing method and system in the six-axis mechanical motion automation control of the present invention: By setting load shock description information in the sudden inertia change scenario, combining dual-channel Kalman filtering to update the inertia tensor data packet in real time, and synchronously accessing dynamic inertia information during the trajectory segment generation process, the present invention significantly improves the sensitivity of the trajectory prediction stage to load changes. At the same time, the invention introduces a dual-domain feature extraction method based on the thermal stress density in the time domain and the energy dispersion in the frequency domain, and uses the dual-domain complementary integral mechanism to generate an energy shock decision quantity as the basis for micro-pulse rearrangement decision, dynamically optimizing the acceleration pulse layout. Further, through the lead rhythm control of the instruction distributor, the problem of controller cache overflow is effectively avoided, ensuring the continuity of the instruction stream. Finally, combined with the closed-loop feedback of the real-time torque error and pose error, the prior inertia model is continuously refreshed to achieve the synchronous evolution of inertia change and trajectory prediction. The overall solution improves the trajectory continuity, current output stability and energy dissipation efficiency of the six-axis mechanical motion under load mutation conditions, greatly reduces the risk of overload shutdown caused by inertia lag, and enhances the adaptability of the prospective preprocessing method in a dynamically complex operating environment. Description of the Drawings

[0017] Figure 1 It is a flow chart of the prospective preprocessing method in the six-axis mechanical motion automation control of the present invention.

[0018] Figure 2 It is a structural diagram of the prospective preprocessing system in the six-axis mechanical motion automation control of the present invention. Detailed Embodiments

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] Embodiment 1: Figure 1 A prospective preprocessing method in the six-axis mechanical motion automation control of the present invention is given, including: S1: When the jaw pressure exceeds the set threshold, generate load impact description information and write it with a time stamp appended into the inertia estimation channel.

[0021] S2: The inertia estimation channel uses dual-channel Kalman filtering to fuse servo current data and encoder angular velocity data, and outputs an updated inertia tensor data packet within half a control cycle.

[0022] S3: The prediction cache replaces the original inertia model parameters with the inertia tensor data packet, and re-solves the acceleration vector and jerk vector of the next time slot according to the sliding window method to form a trajectory segment data set.

[0023] S4: The power mapper maps the trajectory segment data set to the drive thermal power surface, calculates the time-domain thermal stress density and the frequency-domain energy dispersion, and generates an energy impact decision quantity using the dual-domain complementary integration algorithm; if the energy impact decision quantity is higher than the preset threshold, split the acceleration pulse in the trajectory into multiple micro-pulse segments and rearrange their time order.

[0024] S5: The instruction distributor adjusts the cache writing rhythm according to the instruction accumulation prediction function, and pushes the rearranged instruction stream to the drive end according to the preset lead depth to prevent cache overflow.

[0025] S6: The feedback rectifier collects the real-time torque error and pose error, and inputs them into the Bayesian integration module to refresh the prior model of the inertia tensor, providing a correction basis for generating load impact description information in the next cycle.

[0026] In the field of six-axis mechanical motion automation control, the forward-looking preprocessing technology is the core means to achieve high precision, high speed, and multi-axis collaborative stability. This technology predicts the motion trajectories and dynamic parameters within a number of future control cycles and generates a continuous control instruction stream based on the mechanical model to meet the requirements of complex collaborative tasks. For example, when an industrial robot performs tasks such as grasping, handling, or assembling, the six-axis robotic arm needs to respond quickly and maintain the continuity of the trajectory. However, in scenarios with sudden load changes such as grasping an object, the inertia characteristics of the end load will change drastically. Since traditional forward-looking preprocessing methods rely on static or slowly updated inertia models and fail to reflect sudden load changes in a timely manner, the predicted control instructions do not match the actual mechanical state. This mismatch may cause servo drive current surges, drive overload shutdowns, and even electrical damage or motion interruptions. Therefore, to address the problem of inertia parameter lag in scenarios with sudden load changes, the present invention proposes an improved forward-looking preprocessing method that enhances the adaptability and stability of the system by continuously monitoring load changes and quickly updating control parameters. Step S1 of the present invention, as the starting point of the entire solution, is responsible for capturing the initial signal of sudden load changes and providing accurate basic data for subsequent steps.

[0027] Step S1 includes the following: S1-1. Real-time monitoring of gripper pressure: The gripper is a key component at the end of the six-axis robotic arm for performing grasping tasks, and its pressure change directly reflects the dynamic characteristics of the end load. To achieve real-time monitoring of the force on the gripper, pressure sensors or torque sensors installed on the gripper are used to continuously collect the pressure data of the gripper. The collection process is carried out within each control cycle. The pressure sensor converts the current force state of the gripper into an electrical signal, samples and digitally processes this electrical signal, and generates the pressure value at the current moment. The collection frequency is consistent with the duration of the control cycle.

[0028] S1-2. Setting and comparing the pressure threshold: A pressure threshold is determined in advance based on the design parameters, load capacity, and historical operation data of the six-axis robotic arm. This pressure threshold is used to distinguish the normal pressure range of the gripper during normal grasping from the abnormal pressure range during sudden load changes. Within each control cycle, the current gripper pressure value collected by the pressure sensor is compared with the preset pressure threshold. The specific comparison process is as follows: read the current gripper pressure value and compare it numerically with the pressure threshold. If the current gripper pressure value is less than or equal to the pressure threshold, it is determined that the load is within the normal range and subsequent processing is not triggered; if the current gripper pressure value is greater than the pressure threshold, it is determined that a sudden load change has occurred and subsequent steps are initiated. The purpose of setting the pressure threshold is to distinguish normal operations from abnormal load mutations and avoid unnecessary reactions of the system to normal pressure fluctuations.

[0029] S1-3. Generation of load impact description information: When it is determined that the current jaw pressure value exceeds the pressure threshold, a load impact description information is generated. The current jaw pressure value is extracted as a quantitative index of the load mutation intensity, and at the same time, the status identifier of the jaw is recorded, including the opening and closing angle of the jaw or the direction of the grasping force, to describe the end posture when the load mutation occurs. Subsequently, the current jaw pressure value and the jaw status identifier are integrated into a structured data unit to ensure that the data content is complete and the format is unified, facilitating subsequent parsing and use.

[0030] S1-4. Timestamp attachment: After generating the load impact description information, the exact moment when the load mutation occurs is recorded to generate a timestamp. Subsequently, the generated timestamp is bound to the load impact description information to form a complete data packet, ensuring that the two are associated and used in subsequent processing.

[0031] S1-5. Writing to the inertia estimation channel: The data packet containing the load impact description information and the timestamp is written to a dedicated inertia estimation channel. Through the data transmission interface, the data packet is sent to the inertia estimation channel according to the predefined protocol format.

[0032] In the case of load mutation scenarios, such as when grasping an object, the inertia characteristics of the end load change rapidly. Traditional methods are difficult to adapt quickly, resulting in a mismatch between the control instruction and the actual state, leading to current impact and driver overload. However, relying solely on jaw pressure monitoring cannot fully reflect the impact of load changes on the overall dynamic characteristics of the manipulator. Therefore, it is necessary to fuse multi-source data to accurately update the inertia tensor. The present invention proposes step S2, which fuses servo current data and encoder angular velocity data through dual-channel Kalman filtering to quickly generate an updated inertia tensor data packet within half a control cycle, providing an accurate basis for subsequent trajectory prediction.

[0033] Step S2 includes the following: S2-1. Data input and preparation: Receive the load impact description information and the timestamp from the inertia estimation channel.

[0034] The load impact description information is the initial signal of the load mutation generated when the jaw pressure exceeds a pre - set threshold, and the timestamp records the exact moment when the load mutation occurs. At the same time, the servo current data and encoder angular velocity data within the current control cycle are collected. The servo current data represents the real - time current values of the drive motors of each axis, and indirectly reflects the actual torque output by the motor through the corresponding relationship between current and torque; the encoder angular velocity data represents the real - time angular velocity information of each axis, which is used to describe the motion state of the robotic arm. The servo current data and encoder angular velocity data are subjected to time synchronization processing to ensure that the time points of these data are exactly aligned with the timestamp of the load impact description information. The synchronized data is organized into an observation input for subsequent two - channel Kalman filtering processing to estimate the inertia tensor of the robotic arm.

[0035] S2 - 2. Two - channel Kalman filter configuration: A two - channel Kalman filter is used to estimate the inertia tensor of the robotic arm.

[0036] The two - channel Kalman filter includes a current channel and an angular velocity channel, which process the servo current data and encoder angular velocity data respectively. The state vector of the filter is defined as the estimated value of the inertia tensor, and the inertia tensor contains the inertia components of each axis of the robotic arm. The current channel analyzes the torque change caused by the load mutation according to the servo current data using the mapping relationship between current and torque, so as to derive the estimation of the inertia tensor. The angular velocity channel analyzes the actual motion state of the robotic arm according to the encoder angular velocity data using the kinematic relationship, and assists in correcting the estimation of the inertia tensor. The initial state of the filter is provided by the load impact description information to ensure that the filter can start quickly and enter the estimation process when the load mutation occurs.

[0037] The design of the two - channel Kalman filter allows the system to use both dynamic information and kinematic information for estimation at the same time. The current channel directly reflects the torque change of the motor output through the servo current data, which is suitable for capturing the dynamic influence caused by the load mutation; the angular velocity channel provides the actual motion state of the robotic arm through the encoder angular velocity data, which helps to correct the deviation in the estimation. The initial state is set by the load impact description information to ensure that the filter can quickly adapt to the new dynamic characteristics when the load mutates.

[0038] S2 - 3. Filter state update: The state update process of the two - channel Kalman filter is divided into a prediction stage and a measurement update stage.

[0039] In the prediction stage, based on the estimated value of the inertia tensor at the previous moment, combined with the influence of process noise, the prior estimated value of the inertia tensor at the current moment is calculated. The process noise represents the uncertainty of the inertia tensor changing over time and is used to describe the natural fluctuations in the dynamic environment.

[0040] During the measurement update phase, the prior inertia tensor estimate is corrected using servo current data and encoder angular velocity data respectively. Specifically, in the current channel, the inertia tensor estimate of the current channel is adjusted by comparing the actual servo current data with the current value predicted based on the prior estimate; in the angular velocity channel, the inertia tensor estimate of the angular velocity channel is adjusted by comparing the actual encoder angular velocity data with the angular velocity value predicted based on the prior estimate. The correction amplitude of each channel is determined by the Kalman gain matrix, which is dynamically adjusted according to the noise characteristics of the servo current data and encoder angular velocity data to balance the accuracy and stability of the estimate.

[0041] For example, the filter state update is performed as follows: State prediction: Based on the posterior state at the previous moment and the process noise, the prior state is calculated: ; represents the inertia increment predicted based on the motion model, which is derived from the load mutation trend.

[0042] Measurement update: Use and to correct : Current channel update: ; is the Kalman gain matrix for the current channel.

[0043] is the current observation matrix, which is constructed based on the torque-current relationship.

[0044] Angular velocity channel update: ; is the Kalman gain matrix for the angular velocity channel.

[0045] is the angular velocity observation matrix, which is constructed based on the kinematic relationship.

[0046] represents the real-time current value of the drive motor for each axis, reflecting the actual torque output by the motor.

[0047] represents the angular velocity of each axis, which is measured by the encoder.

[0048] The combination of the prediction phase and the measurement update phase enables the dual-channel Kalman filter to quickly converge to the new inertia tensor value when the load suddenly changes. The dynamically adjusted Kalman gain matrix enhances the filter's adaptability to changes in data quality, ensuring reliable inertia tensor estimation under various operating conditions.

[0049] S2-4. Fusion and Optimization: Fuse the inertia tensor estimation of the current channel and the inertia tensor estimation of the angular velocity channel to generate the final inertia tensor estimation. The fusion process is as follows: First, calculate a fusion coefficient based on the confidence of the servo current data and the confidence of the encoder angular velocity data. The value range of the fusion coefficient is between 0 and 1. The confidence is determined by the noise level and consistency of the data. Data with lower noise and higher consistency has higher confidence.

[0050] Next, use the fusion coefficient to perform a weighted average of the inertia tensor estimation of the current channel and the inertia tensor estimation of the angular velocity channel to calculate the fused inertia tensor estimation. After the fusion is completed, the system imposes physical constraints on the inertia tensor estimation to ensure that its components are non-negative and within the dynamic range of the robotic arm, so as to avoid generating unreasonable estimation results.

[0051] For example, the fusion method can be as follows: Data Fusion: Fuse the current channel estimation and the angular velocity channel estimation to obtain the fused inertia tensor estimation : ; is the fusion coefficient, which is determined according to the confidence of and , and its value range is [0,1].

[0052] Optimization Constraints: Impose physical constraints on to ensure that the components of the inertia tensor are non-negative and within the dynamic range of the robotic arm. The fused inertia tensor estimation is more comprehensive and accurate. The application of physical constraints avoids control errors caused by unreasonable estimation results and ensures the safe operation of the robotic arm in scenarios of sudden load changes.

[0053] S2-5. Output Inertia Tensor Data Packet: The fused inertia tensor estimate is encapsulated into an inertia tensor data packet, which contains the components of the inertia tensor and a timestamp consistent with the load impact description information in step S1. The inertia tensor data packet is generated within half a control cycle and transmitted to the prediction cache module. The entire processing process is subject to strict time constraints to ensure the timely generation and transmission of the inertia tensor data packet to meet the real-time processing requirements of subsequent steps.

[0054] In step S1 and S2, load impact description information is generated by monitoring the gripper pressure, and the servo current data and encoder angular velocity data are fused using dual-channel Kalman filtering to quickly output an updated inertia tensor data packet. However, simply updating the inertia tensor is not sufficient to fully address load mutations. The updated inertia parameters need to be applied to trajectory prediction to ensure the accuracy of control commands. Step S3 is designed for this purpose. It uses the inertia tensor data packet provided by step S2 to update the prediction model and recalculate the motion parameters for the next time slot to generate a trajectory segment data set.

[0055] Step S3 includes the following: S3-1. Inertia model parameter update: The inertia tensor data packet contains two parts: one is the updated inertia tensor obtained by fusing the servo current data and encoder angular velocity data using dual-channel Kalman filtering, and the other is the timestamp recording the data generation time. The updated inertia tensor reflects the latest dynamic characteristics of the robotic arm after load mutation, and the timestamp marks the generation time of the data. There is a prediction cache module in the system responsible for maintaining a set of inertia model parameter sets, which are used to support subsequent trajectory prediction calculations.

[0056] Replace the original inertia model parameter set in the prediction cache module with the updated inertia tensor extracted from the inertia tensor data packet to ensure that the prediction model can reflect the current load state of the robotic arm. At the same time, calibrate the timing of the inertia parameter update according to the timestamp to make the update operation consistent with the time reference of the current control cycle and avoid errors in prediction results caused by time deviation. The specific calibration process is as follows: Compare the timestamp with the start time of the current control cycle. If there is a deviation, adjust the effective time of the updated inertia tensor to align it with the control cycle.

[0057] S3-2. Sliding window method configuration: Processing technical logic: The sliding window method is used for trajectory prediction. The sliding window method gradually updates the prediction data by moving a window with a fixed width on the time axis.

[0058] The width of the window is defined as the prediction time domain range, indicating the time span of the future motion state that can be predicted.

[0059] The step size of the window is defined as the frequency of prediction update, representing the time interval for each window movement, and is set to be less than or equal to the duration of the control period.

[0060] The future time is divided into multiple time slots, each corresponding to a prediction point. The next time slot specifically refers to the first prediction point after the current time, that is, the current time plus the duration of a control period.

[0061] Within each control period, the window starts from the current time, covers a segment of historical data forward, and extends backward to future time points within the prediction time domain. The prediction result is generated by analyzing the data within the window.

[0062] The sliding window method can continuously update the prediction data by gradually moving the window, thus adapting to the dynamic changes of the robotic arm. The settings of the window width and step size balance the prediction depth and update frequency, allowing the system to anticipate future motion states in advance and respond promptly to current load changes. The division of time slots decomposes the prediction data into discrete time points, facilitating subsequent trajectory generation and data processing.

[0063] S3-3. Solving the acceleration and jerk vectors: Based on the updated set of inertia model parameters, the dynamic equations of the six-axis robotic arm are constructed.

[0064] The dynamic equations describe the relationships among the angular displacement, angular velocity, angular acceleration, and angular jerk of each axis of the robotic arm. Using the historical data within the sliding window (such as the angular velocity and angular displacement at the previous moment) and the current motion state, the acceleration vector and jerk vector for the next time slot are predicted. The specific calculation process is as follows: First, the system uses the Newton-Euler method to calculate the acceleration of each axis. During the calculation, the torque vector at the current moment is used as the input, subtracting the effects of the Coriolis force, centrifugal force terms, and gravity terms, and then processing the remaining part through the inverse matrix of the inertia tensor to obtain the acceleration vector.

[0065] The torque vector is provided by the servo drive. The Coriolis force and centrifugal force terms are calculated through the square or product of the angular velocity, and the gravity term is determined according to the current pose of the robotic arm.

[0066] Next, the jerk vector is calculated by taking the difference between the acceleration vector for the next time slot and the acceleration vector at the current moment, and then dividing by the duration of the control period to obtain the value of the jerk.

[0067] The Coriolis force is a virtual force observed in a rotating reference frame due to the rotational motion of the reference frame. Its magnitude is related to the mass of the object, the velocity, and the angular velocity of the reference frame.

[0068] S3-4. Generating the trajectory segment data set: Organize the predicted acceleration vector and jerk vector into a trajectory segment dataset.

[0069] The trajectory segment dataset contains prediction data for multiple time slots. The data for each time slot is recorded in the form of a triple, which includes a time point, an acceleration vector, and a jerk vector. Arrange these triples in chronological order to form a continuous trajectory segment. The generated trajectory segment dataset is passed to the power mapper in step S4 for energy shock analysis. Extract the data for each time slot from the prediction results of the sliding window, record them one by one in chronological order, and encapsulate them into a dataset.

[0070] Steps S1 to S3 sequentially provide a dynamic basis for adapting to sudden load changes by monitoring the gripper pressure, updating the inertia tensor, and generating a trajectory segment dataset. However, trajectory prediction alone is not sufficient to fully address the risk of driver overload. Especially when sudden load changes cause a sharp change in the end inertia characteristics, the mismatch between the control instruction and the actual state may cause current shock and excessive thermal load. Step S4 is designed to address this problem by evaluating the impact of the trajectory on the thermal power of the driver and optimizing the energy distribution to ensure the safe operation of the driver.

[0071] Step S4 includes the following: S4-1. Trajectory segment dataset mapping: Receive the trajectory segment dataset from step S3. The trajectory segment dataset contains prediction data for multiple time slots, and each time slot records a time point, an acceleration vector, and a jerk vector.

[0072] The power mapper converts the acceleration vector and jerk vector in the trajectory segment dataset into the thermal power value of the driver. The conversion process is based on the thermal model of the driver, considering the impact of acceleration and jerk on the power consumption of the driver. The thermal power value of the driver changes over time, forming a thermal power surface, which reflects the change in the thermal load of the driver during the execution of the trajectory. The specific operation is as follows: For each time slot, input the acceleration vector and jerk vector of this time slot according to the predefined thermal power function, and calculate the corresponding thermal power value. The thermal power function is determined by the physical characteristics and heat dissipation characteristics of the driver, and is usually proportional to the square of the driver current.

[0073] The generation of the thermal power surface enables the system to quantify the thermal impact of the trajectory execution on the driver, facilitating the identification of potential risks of excessive thermal load.

[0074] S4-2. Calculation of thermal stress density in the time domain: Calculate the thermal stress density in the time domain to describe the local concentration degree of the thermal power on the time axis. The calculation method adopts the local weighted average method: for each time point, select a fixed time window, and the window contains the thermal power values within a certain time range before and after this time point. Perform a weighted average on the thermal power values within the window, and the weighting method uses the Gaussian kernel function. The Gaussian kernel function makes the thermal power values closer to the current time point have a greater weight, and the farther away, the smaller the weight. After the calculation is completed, normalize the thermal stress density of all time points to the range of 0 to 1 so as to unify the dimension with the subsequent frequency domain indicators.

[0075] The thermal stress density in the time domain can reflect the local concentration degree of the thermal power in time. The high-density area indicates that the driver bears a large thermal load in a short time and is prone to overload risks. Through the local weighted average method, the thermal power data can be smoothed, the noise influence can be reduced, and at the same time, the local characteristics can be retained.

[0076] The normalized thermal stress density provides a standardized indicator for the system, which is convenient for comparison and fusion with the frequency domain indicators and helps to comprehensively evaluate the thermal influence of the trajectory execution on the driver.

[0077] S4-3. Calculation of the energy dispersion degree in the frequency domain: Perform a frequency domain analysis on the thermal power surface, perform a fast Fourier transform on the data of the thermal power changing with time to obtain the spectrum of the thermal power. Based on the spectrum, calculate the energy dispersion degree in the frequency domain, and adopt the spectral entropy method: first calculate the normalized power spectral density, that is, normalize the power values of the spectrum so that their sum is 1; then calculate the entropy value of the normalized power spectral density. The larger the entropy value, the more dispersed the energy is in the frequency domain. Finally, normalize the energy dispersion degree in the frequency domain to the range of 0 to 1 to ensure consistency with the time domain indicators.

[0078] The energy dispersion degree in the frequency domain reflects the dispersion degree of the thermal power in the frequency domain. The concentration of energy at a specific frequency may cause the driver to resonate or overload at that frequency. Through the spectral entropy method, the uniformity of the energy distribution can be quantified, providing a basis for evaluating the stability of the trajectory execution.

[0079] The normalized energy dispersion degree in the frequency domain and the thermal stress density in the time domain are complementary to each other, jointly constituting a comprehensive indicator for evaluating the influence of the trajectory execution on the driver, improving the comprehensiveness and accuracy of the evaluation.

[0080] S4-4. Dual-domain complementary integration algorithm: Adopt the dual-domain complementary integration algorithm, combine the thermal stress density in the time domain and the energy dispersion degree in the frequency domain to generate an energy impact judgment quantity. The specific operation is as follows: Multiply the time-domain thermal stress density by the frequency-domain energy concentration (i.e., 1 minus the frequency-domain energy dispersion), and then integrate over the entire trajectory segment time to obtain the energy impact verdict. The energy impact verdict synthesizes the characteristics of the time domain and the frequency domain, reflecting the overall energy impact of the trajectory execution on the actuator.

[0081] The dual-domain complementary integration algorithm can comprehensively evaluate the impact of trajectory execution on the thermal load and energy distribution of the actuator by fusing the indicators of the time domain and the frequency domain. The time-domain thermal stress density focuses on the local thermal load concentration, and the frequency-domain energy concentration focuses on the degree of energy concentration in the frequency domain. The two are complementary and jointly determine the severity of the energy impact.

[0082] The energy impact verdict provides a comprehensive quantitative index for the system, facilitating the judgment of whether the trajectory execution will cause excessive energy impact on the actuator.

[0083] S4-5. Energy Impact Verdict and Pulse Rearrangement: Preset an energy impact threshold, which is determined according to the heat capacity and heat dissipation ability of the actuator. If the calculated energy impact verdict exceeds the energy impact threshold, it is considered that the energy impact of the trajectory execution on the actuator is too large, and the acceleration pulses in the trajectory need to be adjusted. The specific adjustment method is as follows: Identify the peak points in the acceleration vector and regard them as acceleration pulses; split the pulse into multiple micro-pulse segments, and the amplitude of each micro-pulse is the original pulse amplitude divided by the number of splits; then, rearrange the time sequence of the micro-pulses and adopt a uniform distribution strategy to make the energy smoothly distributed in time. After the adjustment, generate a rearranged trajectory segment data set, including the adjusted acceleration vector and jerk vector.

[0084] When the energy impact verdict exceeds the threshold, it indicates that the trajectory execution may cause the actuator to be overloaded. By splitting and rearranging the acceleration pulses, the concentrated energy impact can be dispersed over a longer time range, reducing the impact of a single impact on the thermal load of the actuator.

[0085] Step S4 receives the trajectory segment data set in step S3, generates a thermal power surface, calculates the time-domain thermal stress density and the frequency-domain energy dispersion, uses the dual-domain complementary integration algorithm to generate the energy impact verdict, and splits and rearranges the acceleration pulses when the verdict exceeds the threshold, successfully optimizing the energy impact of the trajectory execution on the actuator. This process ensures the safe operation of the six-axis robotic arm in the scenario of sudden load changes and provides reliable trajectory data for the instruction distribution in step S5.

[0086] Steps S1 to S4 sequentially lay a dynamic foundation for adapting to load mutations by monitoring gripper pressure, updating the inertia tensor, generating a trajectory segment dataset, and optimizing energy shock. However, the optimized trajectory data needs to be pushed to the drive end in an efficient and safe manner to prevent instruction cache overflow and ensure the real-time and continuity of control instructions. Step S5 is designed for this requirement. By adjusting the cache write rhythm and pushing the instruction stream through an instruction distributor, it provides a stable instruction basis for the feedback rectification in subsequent step S6.

[0087] Step S5 includes the following: S5-1. Instruction stream reception and parsing: Receive the rearranged trajectory data, which contains the optimized acceleration vector and jerk vector, and each vector corresponds to a specific time point. Convert the trajectory data into an instruction stream that the drive end can execute. The form of the instruction stream can be a position instruction, a speed instruction, or a torque instruction, and the specific form is determined by the requirements of the drive end. The instruction stream is arranged in chronological order, and each instruction corresponds to a control cycle, and the duration of the control cycle is a preset fixed value. The conversion process ensures that the instruction stream corresponds one-to-one with the time points of the trajectory data, so that the drive end can execute the motion trajectory of the robotic arm as expected.

[0088] S5-2. Instruction accumulation prediction function: Define an instruction accumulation prediction function to evaluate the occupancy of the instruction stream in the drive end cache. The instruction accumulation prediction function uses the moving average method to calculate the instruction data volume: Select a fixed moving window that contains the instruction data volumes of the recent several control cycles. Calculate the average of the instruction data volumes within the moving window to obtain the instruction accumulation prediction value at the current moment. The instruction data volume represents the data size of the instruction stream within each control cycle, and the unit is bytes. The size of the moving window is determined by the number of control cycles. The larger the window, the smoother the prediction result, but the slower the response speed to changes in the data volume. Normalize the instruction accumulation prediction value to the range of 0 to 1 for comparison with the drive end cache capacity.

[0089] The capacity of the drive end cache is limited, and the data volume of the instruction stream may fluctuate over time, resulting in unstable cache occupancy. The instruction accumulation prediction function smooths the fluctuations of the instruction data volume through the moving average method, anticipates the occupancy trend of the cache in advance, and avoids the situation of cache overflow caused by a sudden increase in the data volume.

[0090] The normalized instruction accumulation prediction value provides a standardized evaluation index, which is convenient for the system to monitor the cache status in real time and adjust the write rate of the instruction stream in a timely manner, thus ensuring the stable operation of the cache.

[0091] S5-3. Cache write rhythm adjustment: Dynamically adjust the cache write beat based on the output result of the instruction accumulation prediction function and the drive - end cache capacity. The cache write beat controls the rate at which the instruction stream is written into the drive - end cache. The adjustment strategy is as follows: Set a base write rate, and then dynamically reduce the write rate according to the ratio of the instruction accumulation prediction value to the cache capacity. The specific operation is: the cache write beat is equal to the base write rate multiplied by (1 minus the ratio of the instruction accumulation prediction value to the cache capacity). When the instruction accumulation prediction value approaches the cache capacity, the cache write beat decreases to slow down the writing speed of the instruction stream; when the instruction accumulation prediction value is low, the cache write beat remains unchanged or increases appropriately to improve the writing efficiency.

[0092] The dynamic adjustment of the cache write beat can control the writing speed of the instruction stream in real - time according to the data volume of the instruction stream and the actual occupancy of the cache. This adaptive adjustment mechanism can transfer the instruction stream as fast as possible on the premise of ensuring that the cache does not overflow, thus improving the efficiency and security of instruction transfer.

[0093] S5 - 4. Instruction Stream Push and Look - Ahead Depth: Preset a look - ahead depth, which represents the number of control cycles that the instruction stream leads the drive - end execution. The instruction dispatcher writes the instruction stream into the drive - end cache according to the adjusted cache write beat, and ensures that the cache always stores instructions for the look - ahead depth number of control cycles. The setting of the look - ahead depth is used to cope with delays or fluctuations during the instruction transfer process, ensuring that there are always enough instructions available for the drive - end during execution. Real - time monitor the occupancy rate of the cache, where the cache occupancy rate is defined as the ratio of the instruction accumulation prediction value to the cache capacity. If the cache occupancy rate exceeds the preset threshold, suspend the writing of the instruction stream and issue an alarm to prevent cache overflow.

[0094] The setting of the look - ahead depth enables the system to reserve a certain buffer space during the instruction transfer process to cope with unforeseen delays or fluctuations, ensuring the continuity and real - time nature of the instruction stream. The real - time monitoring and alarm mechanism for the cache occupancy rate further improves the security and stability of the system.

[0095] Step S5 receives the rearranged trajectory data generated in step S4, parses it into an instruction stream, dynamically adjusts the cache write beat using the instruction accumulation prediction function, and pushes the instruction stream to the drive - end according to the preset look - ahead depth, effectively preventing cache overflow. This process ensures the real - time nature and stability of the instruction stream in scenarios of sudden load changes, providing a reliable instruction basis for the feedback rectification in step S6.

[0096] The instruction distributor is a key component in a six-axis mechanical motion automation control system. Its main responsibility is to push the optimized motion instruction stream to the drive end according to the preset beat and lead depth. It receives the trajectory data from the trajectory rearrangement processing module and parses it into an instruction stream that the drive end can directly execute. To ensure the real-time and stability of instruction transmission, the instruction distributor dynamically adjusts the cache writing beat through an instruction accumulation prediction function and monitors the occupancy of the drive end cache to prevent cache overflow. This mechanism ensures that the robotic arm can continuously and smoothly execute control instructions in complex scenarios such as sudden load changes, thus improving the reliability and motion accuracy of the system.

[0097] Steps S1 to S5 have successively completed the generation of load impact description information, the update of the inertia tensor, the generation of trajectory segments, the optimization of energy impact, and the push of the instruction stream, providing a dynamic control basis for the robotic arm to adapt to sudden load changes. However, the persistence and unpredictability of load changes require the control process to have an adaptive ability to maintain the accuracy of the inertia tensor model and provide a reliable basis for subsequent cycles. Step S6 addresses this need by forming a closed-loop control mechanism through real-time error acquisition and model refresh, laying the foundation for generating corrected load impact description information for the next cycle of Step S1.

[0098] Step S6 includes the following: S6-1. Real-time error acquisition: Collect real-time torque error and real-time pose error from the drive end.

[0099] Real-time torque error refers to the deviation between the actual torque executed by the drive end and the commanded torque in the instruction stream pushed in Step S5; Real-time pose error refers to the deviation between the actual pose executed by the drive end and the commanded pose in the instruction stream pushed in Step S5.

[0100] Perform low-pass filtering on the collected real-time torque error and real-time pose error. By retaining low-frequency signals and filtering out high-frequency noise, a smooth error signal is generated. The smooth error signal is used as the input data for subsequent processing to ensure the stability and accuracy of the calculation results.

[0101] S6-2. Feedback rectifier: The feedback rectifier converts the smooth real-time torque error and real-time pose error into inertia tensor correction amounts. The conversion process adopts a dynamic gain adjustment strategy, and the specific operation is as follows: First, calculate the proportional term of the torque error, which represents the direct impact of the torque deviation on the correction amount; Then, calculate the rate-of-change term of the torque error, which represents the contribution of the change in torque error over time to the correction amount; Next, calculate the gain term of the pose error, which represents the corrective effect of the pose deviation on the correction amount; Finally, add the proportional term of the torque error, the rate-of-change term of the torque error, and the gain term of the pose error to obtain the inertia tensor correction amount.

[0102] The inertia tensor correction amount represents the adjustment amplitude of the inertia tensor at the current moment. According to the dynamic response characteristics of the robotic arm and the frequency of load mutations, the proportional gain, rate-of-change gain, and pose error gain are pre-calibrated to ensure a balance between fast response and stability.

[0103] By comprehensively analyzing the real-time torque error and real-time pose error, the feedback rectifier can comprehensively evaluate the execution deviation of the robotic arm and convert it into an adjustment amount for the inertia tensor model. The dynamic gain adjustment strategy enables the system to adaptively adjust the correction amount according to the actual operating state of the robotic arm, thereby improving the accuracy and robustness of the correction.

[0104] The feedback rectifier is a core module in the six-axis robotic motion automation control system. Its function is to convert the real-time collected torque error and pose error into the correction amount of the inertia tensor to optimize the prior model of the inertia tensor. Using the dynamic gain adjustment strategy, by comprehensively analyzing the smoothed torque error and pose error, the adjustment amplitude of the inertia tensor is calculated. To achieve a balance between fast response and stability, the feedback rectifier calibrates the proportional gain, differential gain, and pose error gain to ensure the accuracy and applicability of the correction amount. Finally, the feedback rectifier provides accurate correction information for the subsequent Bayesian integration module, supporting the dynamic adaptation of the inertia tensor model, thereby improving the control accuracy and robustness of the system.

[0105] S6-3. Bayesian Integration Module: The Bayesian integration module uses the Bayesian inference method to fuse the prior model of the inertia tensor with the inertia tensor correction amount to generate a posterior model. The specific operation is as follows: Add the prior model and the inertia tensor correction amount, and multiply by a conditional probability value to obtain the posterior model. The conditional probability value represents the likelihood of the inertia tensor correction amount given the real-time pose error. The calculation of the likelihood uses an exponential decay model. The specific process is as follows: Calculate the difference between the real-time pose error and the reference value, and then quantify the degree to which this difference deviates from the reference value through an exponential function. The reference value is calibrated based on the historical execution data of the robotic arm, and the constant of the exponential decay is determined by the accuracy requirements of the robotic arm. The exponential decay model emphasizes the exponential impact of the real-time pose error deviating from the reference value, avoiding the simple linear assumption of the error impact in traditional methods.

[0106] For example, the calculation and processing method can be as follows: Function: Using the Bayesian inference method, the prior model of the inertia tensor is fused with the correction quantity to generate a posterior model .

[0107] Bayesian update: The probability integral update formula is adopted: ; where: is the prior model at the current moment, and the initial value is the inertia tensor data packet output in step S2 .

[0108] is the conditional probability, indicating the likelihood of the correction quantity under the given pose error .

[0109] Likelihood calculation: The exponential decay model is used to calculate the likelihood: ; where: is the reference value of the pose error, calibrated according to historical execution data.

[0110] is the decay constant, which controls the sensitivity of the likelihood and is determined by the accuracy requirements of the robotic arm.

[0111] This formula avoids traditional variance calculations and emphasizes the exponential impact of errors deviating from the reference value.

[0112] The Bayesian inference method can provide a more accurate inertia tensor estimate in an uncertain environment by fusing the prior model of the inertia tensor and the real-time calculated inertia tensor correction quantity. The exponential decay model highlights the impact of real-time pose errors in likelihood calculation, enhances the sensitivity of the system to deviations, and thus improves the accuracy of the posterior model.

[0113] S6-4. Prior model refresh: Assign the posterior model to the prior model at the next moment to complete the update of the inertia tensor. Use the posterior model calculated at the current moment as the prior model for the next moment for the processing of the next cycle. The refreshed prior model is passed to the inertia estimation channel in step S1 as the correction basis for generating the load impact description information. The refresh of the prior model realizes the dynamic adaptability of the inertia tensor model, enabling the system to continuously optimize the model parameters in the load mutation scenario and maintain the accuracy of the model. The refreshed prior model provides the latest dynamic information for step S1, ensuring that the generation of the load impact description information is based on the current actual state of the robotic arm. This enables the robotic arm to operate smoothly in the load mutation scenario, improving the stability and reliability of the system.

[0114] The processing technical logic of step S6 forms a closed-loop control mechanism with steps S5 and S1. Step S5 pushes the instruction stream to the drive end through the instruction distributor to provide the control basis for the execution of the robotic arm. Step S6 receives the real-time torque error and real-time pose error after the drive end executes, uses the feedback rectifier and Bayesian integration module to refresh the prior model of the inertia tensor, and passes the updated prior model to step S1 to provide the correction basis for the generation of the load impact description information in the next cycle. As the core link of feedback and model refresh, step S6 ensures the dynamic adaptability of the inertia tensor model and improves the control accuracy and stability of the six-axis robotic arm in the load mutation scenario.

[0115] Embodiment 2: Figure 2 A prospective preprocessing system in the six-axis mechanical motion automation control of the present invention is given, including: a load detection module, an inertia update module, a trajectory generation module, an energy optimization module, an instruction push module, and a feedback correction module; Load detection module: When the jaw pressure exceeds the set threshold, generate the load impact description information and write it with a timestamp into the inertia estimation channel.

[0116] Inertia update module: The inertia estimation channel uses dual-channel Kalman filtering to fuse the servo current data and the encoder angular velocity data, and outputs the updated inertia tensor data packet within half a control cycle.

[0117] Trajectory generation module: The prediction cache replaces the original inertia model parameters with the inertia tensor data packet, and re-solves the acceleration vector and jerk vector of the next time slot according to the sliding window method to form a trajectory segment data set.

[0118] Energy optimization module: The power mapper maps the trajectory segment data set to the drive thermal power surface, calculates the time-domain thermal stress density and the frequency-domain energy dispersion, and uses the dual-domain complementary integration algorithm to generate the energy impact decision quantity; if the energy impact decision quantity is higher than the preset threshold, split the acceleration pulses in the trajectory into multiple micro-pulse segments and rearrange their time order.

[0119] Instruction Push Module: The instruction dispatcher adjusts the cache writing rhythm according to the instruction accumulation prediction function, and pushes the rearranged instruction stream to the driving end according to a preset lead depth to prevent cache overflow.

[0120] Feedback Correction Module: The feedback rectifier collects real-time torque error and pose error, and inputs them into the Bayesian integration module to refresh the prior model of the inertia tensor, providing a correction basis for generating load impact description information in the next cycle.

[0121] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula that is closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0122] It should be noted that the system of the present invention can be deployed on the device itself to achieve embedded applications, or can also run on a PC or other terminals with a user interface, so as to meet various hardware environments and usage requirements.

[0123] Only some exemplary embodiments of the present invention have been described above by way of illustration. Undoubtedly, for those of ordinary skill in the art, without departing from the spirit and scope of the present invention, the described embodiments can be modified in various different ways. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

[0124] It should be noted that in this article, if there are relational terms such as first and second, they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or device including the element.

[0125] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A prospective preprocessing method in six-axis mechanical motion automation control, characterized in that, Including the steps: S1: When the gripper pressure exceeds the set threshold, generate load impact description information, append a timestamp and write it into the inertia estimation channel; S2: The inertia estimation channel uses dual-channel Kalman filtering to fuse servo current data and encoder angular velocity data, and outputs an updated inertia tensor data packet within half a control cycle; S3: The prediction cache replaces the original inertia model parameters with the inertia tensor data packet, and re-solves the acceleration vector and jerk vector of the next time slot according to the sliding window method to form a trajectory segment data set; S4: The power mapper maps the trajectory segment data set to the drive thermal power surface, calculates the thermal stress density in the time domain and the energy dispersion in the frequency domain, and uses the dual-domain complementary integral algorithm to generate an energy impact decision quantity; if the energy impact decision quantity is higher than the preset threshold, split the acceleration pulses in the trajectory into multiple micro-pulse segments and rearrange their time order; S5: The instruction distributor adjusts the cache write rhythm according to the instruction accumulation prediction function, and pushes the rearranged instruction stream to the drive end according to the preset lead depth to prevent cache overflow; S6: The feedback rectifier collects the real-time torque error and pose error, and inputs them into the Bayesian integration module to refresh the prior model of the inertia tensor and provide a correction basis for generating load impact description information in the next cycle.

2. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 1, characterized in that Step S1 includes the following: Real-time monitor the gripper pressure through a pressure sensor or torque sensor installed on the gripper, and the acquisition frequency is consistent with the control cycle; when the gripper pressure exceeds the pressure threshold determined according to the design parameters, load capacity and historical operation data of the six-axis robot arm, generate load impact description information, which is stored in a structured data format and includes the gripper pressure value and the gripper status identifier; at the same time, record the moment when the load mutation occurs as the timestamp, and align the timestamp with the time reference of the control cycle; bind the load impact description information with the timestamp and write it into the inertia estimation channel.

3. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 2, characterized in that, Step S2 includes the following: Receive the load impact description information and timestamp transmitted by the inertia estimation channel; collect the servo current data and encoder angular velocity data within the current control cycle; use a dual-channel Kalman filter, where the current channel estimates the inertia tensor based on the servo current data and the mapping relationship between current and torque, and the angular velocity channel assists in correcting the inertia tensor estimation based on the encoder angular velocity data and the kinematic relationship. The initial state of the dual-channel Kalman filter is provided by the load impact description information. The dual-channel Kalman filter calculates the prior inertia tensor estimate based on the inertia tensor estimate value and process noise at the previous moment during the prediction stage, and corrects the prior inertia tensor estimate using the servo current data and encoder angular velocity data during the measurement update stage. The correction amplitude is determined by the dynamically adjusted Kalman gain matrix.

4. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 3, characterized in that, Step S2 includes the following: Fuse the estimation results of the current channel and the angular velocity channel, and use the confidence weighted fusion method to generate the fused inertia tensor estimate; impose physical constraints on the fused inertia tensor estimate to ensure that each component of the inertia tensor is non-negative and within the dynamic range of the robot arm; Encapsulate the fused inertia tensor estimate into an inertia tensor data packet, which contains the components of the inertia tensor and the timestamp, and transmit the output within half of the control period.

5. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 4, characterized in that Step S3 includes the following: Replace the original inertia model parameter set in the prediction cache module with the updated inertia tensor, and calibrate the timing of the inertia parameter update according to the timestamp; configure the window width and step size using the sliding window method, divide the future time into multiple time slots, and the next time slot is the first prediction point after the current time; construct the dynamic equation based on the updated inertia model parameter set, and use the historical data and the current state within the sliding window to calculate the acceleration vector of the next time slot using the Newton-Euler method. Specifically, subtract the Coriolis force, centrifugal force terms, and gravitational force from the torque vector at the current moment, and then process it through the inverse matrix of the inertia tensor to obtain the acceleration vector; calculate the jerk vector by taking the time derivative of the acceleration vector; arrange the predicted acceleration vector and jerk vector in chronological order as a trajectory segment data set.

6. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 5, characterized in that Step S4 includes the following: Convert the acceleration vector and jerk vector in the trajectory segment data set into the thermal power values of the driver through a power mapper to form a thermal power surface.

7. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 6, characterized in that, Step S4 also includes the following: Calculate the time-domain thermal stress density using the locally weighted average method. Select a fixed time window at each time point, use the Gaussian kernel function to perform weighted averaging on the thermal power values within the time window, and normalize the result to the range of 0 to 1; Perform a fast Fourier transform on the thermal power surface to obtain the frequency spectrum, and calculate the frequency-domain energy dispersion using the spectral entropy method and normalize it to the range of 0 to 1; Multiply the time-domain thermal stress density by the frequency-domain energy concentration through the dual-domain complementary integration algorithm and integrate over time to generate an energy shock decision quantity; if the energy shock decision quantity exceeds the preset energy shock threshold, identify the peak points in the acceleration vector, split them into multiple micro-pulse segments, and rearrange the time order.

8. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 7, characterized in that, Step S5 includes the following: Convert the rearranged trajectory data into an instruction stream executable by the drive end. The instruction stream is arranged in chronological order and each instruction corresponds to a control period; calculate the instruction accumulation prediction function using the moving average method, average the instruction data volume in the recent several control periods and normalize it to the range of 0 to 1; dynamically adjust the cache write rhythm according to the instruction accumulation prediction function and the drive end cache capacity. Specifically, the cache write rhythm is equal to the basic write rate multiplied by a ratio of 1 minus the value of the instruction accumulation prediction function to the drive end cache capacity; write the instruction stream into the drive end cache according to the adjusted cache write rhythm, and ensure that the drive end cache stores instructions for a preset advanced depth of control periods; monitor the drive end cache occupancy rate in real time, and if the drive end cache occupancy rate exceeds the preset threshold, pause writing and issue an alarm.

9. The prospective preprocessing method in the six-axis mechanical motion automation control according to claim 8, characterized in that, Step S6 includes the following: Collect the real-time torque error and real-time pose error from the drive end. The real-time torque error is the deviation between the actual torque and the command torque in the instruction stream, and the real-time pose error is the deviation between the actual pose and the command pose in the instruction stream; Apply low-pass filtering to the real-time torque error and real-time pose error to generate smooth torque error signals and pose error signals; Convert the smooth torque error signals and pose error signals into inertia tensor correction quantities through a feedback rectifier, and calculate the inertia tensor correction quantities using a dynamic gain adjustment strategy, including a proportional term of the torque error, a rate-of-change term of the torque error, and a gain term of the pose error; Use a Bayesian integration module to fuse the inertia tensor prior model with the inertia tensor correction quantity to generate an inertia tensor posterior model. Specifically, add the inertia tensor prior model to the inertia tensor correction quantity and multiply by the conditional probability based on the pose error signal; assign the inertia tensor posterior model to the inertia tensor prior model at the next moment and transmit it to the inertia estimation channel to provide a correction basis for generating load impact description information in the next cycle.

10. A prospective preprocessing system in six-axis mechanical motion automation control is used to implement the prospective preprocessing method in six-axis mechanical motion automation control according to any one of claims 1-9, characterized in that, Include: Load detection module, inertia update module, trajectory generation module, energy optimization module, instruction push module, and feedback correction module; Load detection module: When the jaw pressure exceeds the set threshold, generate load impact description information and write it to the inertia estimation channel with a timestamp attached; Inertia update module: The inertia estimation channel uses dual-channel Kalman filtering to fuse servo current data and encoder angular velocity data, and outputs an updated inertia tensor data packet within half a control cycle; Trajectory generation module: The prediction cache replaces the original inertia model parameters with the inertia tensor data packet, and re-solves the acceleration vector and jerk vector for the next time slot according to the sliding window method to form a trajectory segment data set; Energy optimization module: The power mapper maps the trajectory segment data set to the drive thermal power surface, calculates the time-domain thermal stress density and the frequency-domain energy dispersion, and uses a dual-domain complementary integration algorithm to generate an energy impact decision quantity; if the energy impact decision quantity is higher than the preset threshold, split the acceleration pulses in the trajectory into multiple micro-pulse segments and rearrange their time order; Instruction push module: The instruction distributor adjusts the cache writing rhythm according to the instruction accumulation prediction function, and pushes the rearranged instruction stream to the drive end according to the preset lead depth to prevent cache overflow; Feedback correction module: The feedback rectifier collects the real-time torque error and pose error and inputs them into the Bayesian integration module to refresh the prior model of the inertia tensor and provide a correction basis for generating load impact description information in the next cycle.

Images