Robot joint drive control method and system

Abstract

The application is suitable for the technical field of robot control, and provides a robot joint driving control method and system. The motion state and thermal state of each joint are collected. The motion state and thermal state of each joint in a prediction time domain are respectively predicted according to the collection results. The thermal health factor of each joint is determined according to the predicted thermal state. The thermal health factor is negatively correlated with the predicted thermal state, and the torque constraint boundary of each joint is corrected according to the thermal health factor. The deviation between the predicted motion state and the obtained reference motion state is minimized as an optimization target. The optimal torque sequence in a control time domain is solved in the corrected torque constraint boundary, and each joint is driven according to the first term of the optimal torque sequence. The torque constraint boundary of the joint can be adjusted before the joint temperature reaches a fixed temperature threshold, the risk of joint overheating is reduced, and the mutation amplitude of the joint output torque is reduced.

Description

Technical Field

[0001] This application relates to the field of robot control technology, specifically to a robot joint drive control method and system. Background Technology

[0002] Robot joints typically use high-power-density permanent magnet synchronous motors as actuators. During continuous walking, running, jumping, and carrying, the joint motors generate heat while outputting torque. As a result, the temperature of the joint windings and the joint assembly rises, and the torque output accuracy of the joint decreases with increasing temperature.

[0003] Currently, the torque distribution for each joint in this field is typically based on kinematic and dynamic models. Joint thermal protection employs a passive approach, meaning that the joint's output torque is limited only after the joint temperature reaches a pre-set fixed temperature threshold. In this method, the torque limitation intervention occurs later than the joint temperature rise, resulting in significant fluctuations in joint temperature before and after the limitation intervention. Consequently, the joint's output torque drops from its rated value to the limited value at the moment of limitation intervention, affecting the continuity of robot motion.

[0004] Maintaining joint motion tracking performance while reducing the risk of joint overheating and minimizing sudden changes in joint output torque are problems that need to be solved in this field. Summary of the Invention

[0005] In view of this, this application provides a robot joint drive control method and system, which can adjust the torque constraint boundary of the joint before the joint temperature reaches a fixed temperature threshold, which helps to reduce the risk of joint overheating and reduce the sudden change amplitude of joint output torque.

[0006] The first aspect of this application provides a robot joint drive control method, comprising: Collect the motion and thermal states of each joint; Based on the collected results, predict the motion and thermal states of each joint in the prediction time domain; The thermal health factor of each joint is determined based on the predicted thermal state. The thermal health factor is negatively correlated with the predicted thermal state, and the torque constraint boundary of each joint is corrected according to the thermal health factor. With the goal of minimizing the deviation between the predicted motion state and the obtained reference motion state, the optimal torque sequence in the control time domain is solved within the corrected torque constraint boundary, and each joint is driven according to the first term of the optimal torque sequence.

[0007] Optionally, the acquisition of the motion and thermal states of each joint includes: Collect the joint angles and angular velocities of each joint to determine their motion state. The winding temperature and joint assembly temperature of each joint are collected as the thermal state of each joint.

[0008] Optionally, determining the thermal health factors of each joint based on the predicted thermal state includes: The predicted winding temperature and the joint assembly temperature are weighted and summed to obtain the comprehensive thermal state index of each joint at each prediction step in the prediction time domain. The calculation formula is as follows: in, This indicates the comprehensive thermal state index. This indicates the predicted winding temperature. This indicates the predicted temperature of the joint assembly. Indicates the regulating factor; The maximum value of the comprehensive thermal state index of each joint in each prediction step within the prediction time domain is taken as the maximum value of the comprehensive thermal state index of each joint. The thermal health factor of each joint is determined based on the maximum value of the comprehensive thermal state index of each joint. The calculation formula is as follows: in, Indicates the joint number. Indicates the first Thermal health factors of each joint Indicates the first The maximum value of the comprehensive thermal state index of each joint. Indicates the first The safe operating temperature of each joint. Indicates the first The maximum operating temperature of each joint. This represents the minimum constraint value for thermal health factors, and Greater than zero.

[0009] Optionally, the torque constraint boundary includes the maximum permissible output torque and the maximum permissible output torque variation; The step of correcting the torque constraint boundaries of each joint according to the thermal health factor includes: The maximum permissible output torque of each joint is the product of the thermal health factor of each joint and the rated torque of the motor of each joint. The maximum permissible output torque variation of each joint is calculated by multiplying the thermal health factor of each joint by the rated torque variation of the motor of each joint within a unit control cycle.

[0010] Optionally, the prediction of the thermal state of each joint in the prediction time domain includes: The optimal torque sequence obtained from the previous control cycle is used as the torque input for each prediction step in the prediction time domain. Based on the torque input and the predicted joint angular velocity, the motor heat loss of each joint in each prediction step is determined. The motor heat loss includes winding copper loss proportional to the square of the torque input and friction loss proportional to the joint angular velocity. Using the collected winding temperature and joint assembly temperature as initial values, the thermal state of each joint in the prediction time domain is gradually derived based on the thermal network model, the motor heat loss, and the obtained joint ambient temperature.

[0011] Optionally, each joint is driven by a permanent magnet synchronous motor, and the calculation formula for determining the motor heat loss of each joint in each prediction step is as follows: in, Indicates the first Heat loss of the motor in each joint, Indicates the first The resistance of the motor windings at each joint, Indicates the first Motor torque coefficient of each joint Indicates the first Torque input of each joint, Indicates the first The frictional torque of each joint, Indicates the first The joint angular velocity of each joint.

[0012] Optionally, the thermal network model is a two-node thermal network model, wherein the two nodes of the two-node thermal network model are the winding node and the joint assembly node, respectively; The heat loss of the motor is injected into the winding node, the winding node transfers heat to the joint assembly node through the winding thermal resistance, and the joint assembly node dissipates heat to the environment through the thermal resistance from the joint assembly to the environment.

[0013] Optionally, the optimization objective of minimizing the deviation between the predicted motion state and the acquired reference motion state includes: The optimization objective is to minimize the weighted sum of the motion state tracking term and the control torque smoothing term, whereby the motion state tracking term represents the deviation and the control torque smoothing term represents the torque change between adjacent control cycles in the control time domain.

[0014] Optionally, predicting the motion state of each joint in the prediction time domain includes: The sum of the load torque and friction torque of each joint is used as the external disturbance of each joint, and the external disturbance is kept constant at the value of the current control cycle within the prediction time domain. Using the acquired joint angles and joint angular velocities as initial values, and based on the joint dynamics model, the joint angles and joint angular velocities of each joint in the prediction time domain are gradually derived from the torque sequence to be optimized in the control time domain and the external disturbance.

[0015] A second aspect of this application provides a robot joint drive control system, comprising: The status acquisition module is used to collect the motion and thermal states of each joint. The state prediction module is used to predict the motion state and thermal state of each joint in the prediction time domain based on the acquisition results. The torque boundary correction module is used to determine the thermal health factor of each joint based on the predicted thermal state, and to correct the torque constraint boundary of each joint according to the thermal health factor. The thermal health factor is negatively correlated with the predicted thermal state. The torque solving module is used to solve for the optimal torque sequence in the control time domain within the corrected torque constraint boundary, with the optimization objective of minimizing the deviation between the predicted motion state and the acquired reference motion state, and to drive each joint according to the first term of the optimal torque sequence.

[0016] The first aspect of this application involves collecting the motion and thermal states of each joint and predicting their respective motion and thermal states in the prediction time domain based on the collected data. The temperature trend of each joint over a future period can be obtained in the current control cycle. The thermal health factor is determined by the predicted thermal state and is negatively correlated with it. The torque constraint boundary of each joint tightens as the predicted thermal state increases. The torque limiting action begins before the joint temperature actually reaches the temperature threshold, thus slowing down the temperature rise and reducing the fluctuation range of the joint temperature before and after the limiting intervention. Since the torque constraint boundary changes continuously with the predicted thermal state, the torque limiting process is gradual, and the abrupt change in the joint output torque at the moment of limiting intervention is reduced. The optimal torque sequence is obtained by solving within the corrected torque constraint boundary with the goal of minimizing the deviation between the predicted motion state and the obtained reference motion state. Each joint outputs torque in a way that minimizes this deviation while satisfying the above torque constraints. In this way, the motion tracking performance and thermal safety performance of the joints are coordinated in the same solution, and the motion continuity of the robot during continuous operation is improved.

[0017] It is understandable that the beneficial effects of the second aspect mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the robot joint drive control method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the overall framework of the robot joint drive control method provided in the embodiments of this application; Figure 3 This is the first embodiment provided in this application. A schematic diagram of the curves showing the change of thermal health factors of each joint with the maximum value of the comprehensive thermal state index. Figure 4 This is a schematic diagram of the multi-joint structure of the humanoid robot provided in the embodiment of this application in a handling scenario. Detailed Implementation

[0020] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0021] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0022] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0023] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0025] The robot joint drive control method provided in this application embodiment can be applied to devices with multiple motor-driven joints, such as humanoid robots, multi-joint robotic arms, or multi-axis servo drive units, for controlling the drive torque of each joint. Specifically, it can be executed by the device's joint controller when running a computer program with corresponding functions.

[0026] During the output torque process of the motor in the joint, the heat generated comes from two parts: one part is the copper loss in the windings caused by the current in the motor windings and its resistance, which increases with the increase of the motor's output torque; the other part is the friction loss generated during the rotation of the joint transmission pair, which increases with the increase of the joint's angular velocity. Heat is generated at the motor windings, transferred to the joint assembly via the heat conduction path between the windings and the joint assembly, and then dissipated to the environment from the outer surface of the joint assembly. The heat capacity of the windings is less than that of the joint assembly; therefore, under the same heat injection, the winding temperature rises faster than the joint assembly temperature. The winding temperature is the first temperature inside the joint to approach its upper temperature limit.

[0027] When using a passive thermal protection method with a fixed temperature threshold, the temperature measured by the temperature sensor is the result of heat accumulation inside the joint, and the torque limiting intervention occurs later than the heat generation. At the moment the limit is triggered, the maximum permissible output torque of the joint drops from the rated value to the limited value, and the joint's output torque undergoes a step change, increasing the tracking error at that moment. When the limit is released, the maximum permissible output torque of the joint jumps back to the rated value, and the joint temperature repeatedly oscillates around the threshold, resulting in significant temperature fluctuations. Simultaneously, the load distribution across multiple joints is uneven, with the heavier-loaded joint triggering the limit first, thus restricting the overall motion continuity of the robot.

[0028] The above process demonstrates that there is a time difference between the intervention of thermal protection and the heat accumulation process of the joint, and the heat accumulation process can be calculated from the joint's torque and angular velocity using a mechanistic model. In the embodiments of this application, the controller not only calculates the motion state of each joint over a future period within each control cycle, but also calculates the winding temperature and joint assembly temperature of each joint over the same period. It then provides a continuously valued thermal health factor based on the calculated worst-case thermal state, using this factor to proportionally tighten the torque constraint boundary of the joint. The tightened boundary is then used as a constraint condition in the torque calculation process. Therefore, the torque limiting action begins before the joint temperature reaches the threshold, and the limiting magnitude gradually increases with the rise of the predicted temperature, preventing the torque from dropping in a step.

[0029] The sequence of events in the above process is as follows: The thermal health factor is calculated before solving for the torque of the current control cycle, while the prediction of the thermal state uses the torques of future prediction steps, and the torque of the current control cycle has not yet been solved. In this embodiment, the future torque used for thermal state prediction is taken from the optimal torque sequence already solved in the previous control cycle. The load change of the robot joint is relatively slow compared to the control cycle, and the difference between the optimal torque sequences of two adjacent control cycles is small. Using the result of the previous control cycle as the torque input for the thermal state prediction of the current control cycle allows for the estimation of the future thermal state without introducing an algebraic loop.

[0030] like Figure 1 As shown, the robot joint drive control method provided in this application embodiment includes the following steps S101 to S104 executed by the joint controller: Step S101: Collect the motion and thermal states of each joint.

[0031] In applications, motion state characterizes the joint's motion during the current control cycle, while thermal state characterizes the joint's temperature during the current control cycle. Assume the robot has... Each joint has a number of joints, and the motion and thermal states of each joint are arranged according to their joint numbers. The subsequent predictions, thermal health factor calculations, and torque boundary corrections are all performed element-wise on a joint-by-joint basis, i.e., the first... The thermal health factor of the joint is only determined by the first joint. The predicted thermal state of the first joint is determined, and it only applies to the first joint. The torque constraint boundaries of each joint are independent of each other.

[0032] Step S102: Based on the acquisition results, predict the motion state and thermal state of each joint in the prediction time domain.

[0033] In application, first define the control cycle. That is, the time interval between two adjacent model predictive control calculations; This represents the discrete time step, i.e., the number of the current control cycle. Indicates the prediction step size; Indicates the prediction time domain length, in Indicates the length of the control time domain. Not greater than The controller predicts the future based on the current state. The motion and thermal states within each control cycle, and the future... The control output torque is optimized within each control cycle. The prediction of motion state and the prediction of thermal state are performed separately. The former is recursively calculated step by step using the collected motion state as the initial value, and the latter is recursively calculated step by step using the collected thermal state as the initial value. Both give the value of each prediction step in the same prediction time domain.

[0034] Step S103: Determine the thermal health factor of each joint based on the predicted thermal state. The thermal health factor is negatively correlated with the predicted thermal state, and the torque constraint boundary of each joint is corrected according to the thermal health factor.

[0035] In application, the thermal health factor is a dimensionless scalar whose value decreases as the predicted thermal state increases. It is used to characterize the remaining thermal load-bearing capacity of a joint in the prediction time domain. The thermal health factors of each joint are arranged according to the joint number. 3D vector. The torque constraint boundary is the range of torque that the joint is allowed to output within the current control cycle. The correction is to tighten this range proportionally using the thermal health factor. The smaller the thermal health factor, the greater the tightening.

[0036] Step S104: With the goal of minimizing the deviation between the predicted motion state and the obtained reference motion state, solve for the optimal torque sequence in the control time domain within the corrected torque constraint boundary, and drive each joint according to the first term of the optimal torque sequence.

[0037] In application, the reference motion state is provided by the higher-level gait planning or trajectory planning stages, representing the joint angles and angular velocities the robot expects to achieve in each control cycle. The controller solves a constrained optimization problem once in each control cycle, obtaining... The optimal torque sequence in the time domain is controlled at each moment, and this sequence is expressed as: in, This represents the optimal torque sequence. This represents the optimal control torque; the sequence has a total of The first item in the sequence is taken as the driving torque output to the motor driver of each joint in this control cycle, and the remaining items are not executed. After entering the next control cycle, the motion state and thermal state are collected again and the solution is re-solved. This cycle is repeated to form a rolling time domain optimization.

[0038] like Figure 2 As shown, the above steps are organized within the controller into four layers: prediction layer, decision layer, control layer, and execution layer. After the current state parameters are input, the prediction layer simultaneously runs the motion prediction model and the thermal state prediction model, respectively providing future... The state of movement and the future The thermal state of the joint is determined in step S102. The decision layer receives the thermal state output by the prediction layer and determines the thermal health factors of each joint. The thermal health factors of each joint are arranged into a thermal health factor vector according to the joint number. The decision layer then generates a dynamic joint output torque capability boundary based on this, corresponding to step S103. The control layer solves the torque sequence with the joint output torque capability boundary as a constraint and outputs the optimal control torque, corresponding to step S104. The execution layer receives the optimal control torque output by the control layer, and the joint executes the torque. At the same time, the state of the joint is collected and fed back to the prediction layer, forming a closed loop. Figure 2 The decision-making and control layers also illustrate the calculation of comprehensive thermal state indices, the assessment of thermal health factors, the generation of joint output torque capability boundaries, and the specific forms of objective functions and constraints. These are part of the optional embodiments described below and will be elaborated upon later. Furthermore, Figure 2 Enter the motion state variable marked in the current state parameter box, which is the same variable as the motion state variable defined above; Figure 2 The thermal health factor assessment box also shows the range of thermal health factor values, that is, the thermal health factor is not less than the minimum constraint value of thermal health factor and not greater than one, which is consistent with the results of the above segmented values. Figure 2 The torque capability boundary shown by the maximum output torque constraint and the maximum output torque change constraint of the joint is updated cycle by cycle according to the control cycle, that is, the thermal health factor calculated in the current cycle is recalculated once in each control cycle.

[0039] Since the temperature trend of the joints over a future period is calculated in the current control cycle, the tightening of the torque constraint boundary occurs before the joint temperature reaches the upper limit of its temperature resistance. The rise in joint temperature is slowed down in advance, and the fluctuation range of joint temperature before and after protection intervention is reduced accordingly. The thermal health factor is continuously evaluated according to the predicted thermal state, and the torque constraint boundary changes continuously accordingly. The limitation process of joint output torque is gradual, and there is no longer a step drop from the rated value to the limit value. The increase in the tracking error of the joint at the time of limitation intervention is reduced. The torque sequence is obtained by solving within the tightened boundary with the goal of minimizing the tracking deviation. The joint outputs torque in a way that minimizes the deviation while satisfying the thermal constraints. The motion tracking performance and thermal safety performance are determined by the same solution, without the need to switch control modes between the two. The thermal health factors of each joint are independent of each other. The joint with a heavier load reduces its maximum allowable output torque first, while the joint with a lighter load continues to output according to the original boundary, thereby improving the motion continuity of the whole machine.

[0040] In one embodiment, step S101 includes: collecting the joint angle and joint angular velocity of each joint as the motion state of each joint; and collecting the winding temperature and joint assembly temperature of each joint as the thermal state of each joint.

[0041] In application, joint angles are measured by joint position sensors, joint angular velocities are obtained differentially from the output of the position sensors, winding temperatures are measured by temperature sensors embedded in the motor windings, and joint assembly temperatures are measured by temperature sensors arranged on the joint housing. Joint angles and joint angular velocities together constitute the motion state; the former is used to calculate the positional deviation from the reference motion state, while the latter is used to calculate both speed deviation and frictional losses. Winding temperatures and joint assembly temperatures together constitute the thermal state; the former reflects the temperature of the part of the motor that first approaches its upper temperature limit, while the latter reflects the boundary temperature at which the joint dissipates heat to the environment. Since the rates of change of these two temperatures differ, collecting only one is insufficient to characterize the joint's thermal state. Because the joint's thermal state is characterized by both temperatures, the calculation of thermal health factors can simultaneously take into account the rapid temperature rise of the windings and the slow heat accumulation of the joint assembly, resulting in a more complete characterization of the thermal state compared to collecting only a single temperature.

[0042] In one embodiment, the process of determining the thermal health factors of each joint in step S103 includes: weighted summing of the predicted winding temperature and joint assembly temperature to obtain the comprehensive thermal state index of each joint in each prediction step within the prediction time domain, calculated using the following formula: in, This indicates the comprehensive thermal state index. This indicates the predicted winding temperature. This represents the predicted joint assembly temperature. This represents the adjustment factor, whose values ​​satisfy: The adjustment factor is used to allocate the weights of winding temperature and joint assembly temperature in the overall thermal state index. The weight of winding temperature is the adjustment factor, and the weight of joint assembly temperature is the difference between one and the adjustment factor; the sum of the two is one. The closer the adjustment factor value is to one, the more the overall thermal state index favors winding temperature. For example, a value of 0.7 is more sensitive to the heat storage of the joint assembly, while a value of 0.85 is more sensitive to the instantaneous temperature rise of the winding.

[0043] The maximum value of the comprehensive thermal state index of each joint in each prediction step within the prediction time domain is taken as the maximum value of the comprehensive thermal state index of each joint. The calculation formula is as follows: in, This represents the maximum value of the comprehensive thermal state index. This indicates the length of the prediction time domain. The maximum value is taken instead of the minimum or average value because the temperature trend of the joint within the prediction time domain is not necessarily monotonous. Taking the maximum value means using the most unfavorable thermal state within the prediction time domain as the basis for defense.

[0044] like Figure 3 As shown, the first Thermal health factors of each joint Following the The maximum value of the comprehensive thermal state index of each joint The change is divided into three segments, and the calculation formula is as follows: in, Indicates the first The safe operating temperature of each joint. Indicates the first The maximum operating temperature of each joint. This represents the minimum constraint value for thermal health factors, and The value should be greater than zero, and a value no less than 0.3 is recommended. When the maximum value of the comprehensive thermal state index is not higher than the safe operating temperature, the thermal health factor is set to one, and the joint outputs torque according to its rated capacity. When the maximum value of the comprehensive thermal state index is not lower than the extreme operating temperature, the thermal health factor takes the minimum constraint value. When the maximum value of the comprehensive thermal state index is between the safe operating temperature and the extreme operating temperature, the thermal health factor decreases linearly from one as the maximum value of the comprehensive thermal state index increases to the minimum constraint value. The minimum constraint value of the thermal health factor is used to reserve some torque output capacity for the joint when the joint temperature exceeds the extreme operating temperature. If this value is zero, the maximum allowable output torque will also be zero, and the optimization problem will have no feasible solution under the torque amplitude constraint, and the joint will be unable to output the torque required to maintain the body posture.

[0045] for Each joint is listed in order of its number, along with its thermal health factors, maximum operating temperature, and safe operating temperature. A dimensional vector is represented as: in, Represents a vector of thermal health factors. Represents the limiting operating temperature vector. This represents the safe operating temperature vector. The safe operating temperature and the extreme operating temperature are set separately according to the joint number. Joints with poor heat dissipation or high power density can use a lower safe operating temperature.

[0046] In another embodiment, the thermal health factor can also decrease as the comprehensive thermal state index reaches its maximum value in the form of a piecewise constant. This involves dividing the interval between the safe operating temperature and the extreme operating temperature into several temperature ranges, each corresponding to a fixed thermal health factor value. The thermal health factor decreases progressively from low to high temperature ranges to the minimum constraint value. In this form, the thermal health factor value does not change continuously with temperature, and the torque boundary experiences a small jump when crossing temperature ranges. Its effect on suppressing sudden torque changes is slightly less than the aforementioned linearly decreasing form, while requiring less computation.

[0047] In one embodiment, the torque constraint boundary includes the maximum allowable output torque and the maximum allowable output torque variation. The process of correcting the torque constraint boundary in step S103 includes: using the product of the thermal health factor of each joint and the rated torque of the motor of each joint as the maximum allowable output torque of each joint; and using the product of the thermal health factor of each joint and the rated torque variation of the motor of each joint within a unit control cycle as the maximum allowable output torque variation of each joint. The calculation formula is as follows: in, Indicates the maximum permissible output torque. This indicates the rated torque of the joint motor, the value of which is determined based on the motor's rated output capacity. This indicates the maximum permissible variation in output torque. It represents the change in rated torque within a unit control cycle, and its value is obtained by multiplying the rated torque change rate by the control cycle.

[0048] In application, the torque amplitude boundary and torque change boundary respectively constrain the magnitude of the joint's output torque and its rate of change between adjacent control cycles, both of which are reduced by a thermal health factor. When only the amplitude boundary is reduced, the joint may still increase the torque directly from the value of the previous cycle to the reduced upper limit within a control cycle, and the rate of change of torque is unrestricted. After both boundaries are reduced simultaneously, the torque decrease process during the reduction of the joint's output torque is also limited to the allowable change within each control cycle, resulting in a smoother torque curve.

[0049] In one embodiment, the process of predicting the thermal state of each joint in the prediction time domain in step S102 includes: using the optimal torque sequence obtained from the previous control cycle as the torque input for each prediction step in the prediction time domain, and determining the motor heat loss of each joint in each prediction step based on the torque input and the predicted joint angular velocity. The motor heat loss includes winding copper loss proportional to the square of the torque input and friction loss proportional to the joint angular velocity; using the collected winding temperature and joint assembly temperature as initial values, and based on the thermal network model, motor heat loss, and the obtained joint ambient temperature, progressively recursively obtaining the thermal state of each joint in the prediction time domain. Definition The thermal state variables at time t are: Utilize the current Given the state and input at time step, predict the first... +1 step, its recursive formula is: in, This represents the heat loss vector of the motor. This represents the joint ambient temperature vector. , and For multi-joint thermal state equations according to control period The state matrix obtained after discretization. Similarly, it can be predicted that: In application, the optimal torque sequence solved in the previous control cycle has a total of The number of terms is less than the prediction time domain length, and the portion of the prediction step size exceeding the control time domain length is taken as the last term of the sequence as the torque input. In the first control cycle after the controller is powered on, the optimal torque sequence of the previous control cycle does not yet exist. At this time, the currently measured output torque of each joint is used as the torque input for each prediction step in the prediction time domain, or zero can be used as the torque input. From the second control cycle onwards, the value is taken according to the optimal torque sequence of the previous control cycle. The joint ambient temperature is measured by an ambient temperature sensor placed near the joint. In situations where the ambient temperature changes slowly, a fixed ambient temperature setpoint can also be used.

[0050] In one embodiment, each joint is driven by a permanent magnet synchronous motor, and the formula for calculating the motor heat loss of each joint in each prediction step is as follows: in, Indicates the first Heat loss of the motor in each joint, Indicates the first The resistance of the motor windings at each joint, Indicates the first Motor torque coefficient of each joint Indicates the first Torque input of each joint, Indicates the first The frictional torque of each joint, Indicates the first The joint angular velocity of each joint. The first term in the formula is the winding copper loss. The ratio of the phase current of the permanent magnet synchronous motor to its output torque is the reciprocal of the motor torque coefficient. Therefore, the copper loss of the three-phase winding is directly proportional to the square of the torque input and inversely proportional to the square of the motor torque coefficient. The second term in the formula is the friction loss, the magnitude of which is the product of the friction torque and the joint angular velocity.

[0051] In the application, taking the first knee joint 9 as an example, the motor winding resistance is 0.15 ohms, the motor torque coefficient is 2.5 Nm per ampere, the torque input of the joint in a certain prediction step is 50 Nm, the friction torque is 1.2 Nm, and the predicted joint angular velocity is 3 radians per second. Then, the winding copper loss in this prediction step is 180 watts, the friction loss is 3.6 watts, and the motor heat loss is 183.6 watts.

[0052] In one embodiment, the above-mentioned thermal network model is a two-node thermal network model, where the two nodes are the winding node and the joint assembly node. Motor heat loss is injected into the winding node, and heat is transferred from the winding node to the joint assembly node through the winding thermal resistance. The joint assembly node dissipates heat to the environment through the thermal resistance from the joint assembly to the environment. Within time t, the... The thermal balance relationship of each joint is as follows: in, and These represent the heat capacity of the winding and the heat capacity of the joint assembly, respectively. and These represent the winding temperature and the joint assembly temperature, respectively. Indicates the winding thermal resistance. This indicates the thermal resistance of the joint assembly to the environment. Indicates the heat loss of the motor. This indicates the ambient temperature of the joint.

[0053] With the first The winding temperature of each joint and the joint assembly temperature constitute the thermal state parameter variables of the joint, expressed as: From the above thermal equilibrium relationship, the thermal equation of state for this joint can be obtained as follows: in, Indicates the first Thermal state parameter variables of each joint , and This represents the state matrix of the thermal network model of the joint.

[0054] From single joints to multiple joints, The winding temperatures and joint assembly temperatures of each joint are arranged in order. A dimensional vector is represented as: The motor heat loss and joint ambient temperature of each joint are also arranged separately. A dimensional vector is represented as: The thermal state parameter variables of the multi-joint system are formed by the winding temperature vectors of each joint and the temperature vector of the joint assembly, and are expressed as follows: Based on this, the thermal state equation for many joints can be obtained as follows: In applications, the temperature of the winding node is the winding temperature, and the temperature of the joint assembly node is the joint assembly temperature; both constitute the thermal state of the joint. The heat capacity of the winding is less than that of the joint assembly. Under the same heat loss injection, the winding temperature rises before the joint assembly temperature. Therefore, the two-node division can simultaneously reflect the rapid temperature rise of the winding and the slow heat accumulation of the joint assembly. In cases where the internal structure of the joint is relatively simple and the temperature difference between the winding and the shell is small, the thermal network model can also be simplified to a single-node model retaining only one temperature node. In this case, the thermal state of the joint is characterized only by the temperature of that node, reducing the model order and consequently decreasing its ability to characterize the instantaneous temperature rise of the winding.

[0055] In one embodiment, the optimization objective in step S104 is to minimize the weighted sum of the motion state tracking term and the control torque smoothing term. The motion state tracking term represents the deviation between the predicted motion state and the acquired reference motion state, and the control torque smoothing term represents the torque change between adjacent control cycles in the control time domain. The objective function is: in, Describe the objective function. This indicates the motion tracking item. This indicates the control torque smoothing term. This indicates a torque smoothing weight that is greater than zero. The torque sequence to be optimized in the time domain of the control moment is represented as follows: in, This represents the torque sequence to be optimized. This represents the motor output torque. The motion state tracking term is calculated using the following formula: The control torque smoothing term is calculated using the following formula: in, Indicates trajectory error. This represents the state tracking weight matrix, used to adjust the weights of joint angle error and angular velocity error in the objective function; This indicates the torque change between adjacent control cycles. This represents the control increment weight matrix, used to adjust the weight of the joint torque change rate in the objective function. The trajectory error and torque change are expressed as follows: In the formula, The reference motion state variable, composed of the reference joint angle and the reference joint angular velocity, is represented as: With the goal of minimizing both motion tracking error and torque smoothness, the optimization objective function can be constructed as follows: In the above equation, the first term on the right-hand side is the aforementioned motion state tracking term, and the second term is the product of the torque smoothing weight and the aforementioned control torque smoothing term. The above optimization is performed under the following constraints: The maximum permissible output torque and its variation in the above constraints are obtained by subtracting the rated torque from the rated torque and the variation in rated torque per unit control cycle using the thermal health factor. When the minimum constraint value of the thermal health factor is zero, the thermal health factor degenerates to: The maximum value of the comprehensive thermal state index in the formula is taken from the comprehensive thermal state index of each prediction step within the prediction time domain, that is: In application, a larger torque smoothing weight value results in a more favorable objective function that emphasizes torque smoothness, leading to gentler changes in joint torque but a slight increase in tracking error; conversely, a smaller value has the opposite effect. The solution involves substituting the motion prediction model into the motion state tracking term, transforming the objective function into a quadratic form with the torque sequence to be optimized in the control time domain as the independent variable. The Hessian matrix of this quadratic form consists of the state tracking weight matrix, the control increment weight matrix, and the coefficient matrix of the prediction model. This quadratic form is convex when both weight matrices are positive definite. Both the torque magnitude constraint and the torque change constraint are linear inequality constraints, making the optimization problem a convex quadratic programming problem. Iterative solutions are found using the effective set method or the interior point method within each control cycle. The result of shifting the optimal torque sequence of the previous control cycle forward by one control cycle is used as the initial value for iteration. Iteration stops when the difference between the objective function values ​​of two adjacent iterations is less than the convergence threshold or the number of iterations reaches the upper limit. The result of the current iteration is taken as the optimal torque sequence for the current control cycle.

[0056] In one embodiment, the process of predicting the motion state of each joint in the prediction time domain in step S102 includes: using the sum of the load torque and friction torque of each joint as the external disturbance of each joint, and keeping the external disturbance unchanged at the value of the current control cycle in the prediction time domain; using the collected joint angles and joint angular velocities as initial values, and based on the joint dynamics model, gradually recursively obtaining the joint angles and joint angular velocities of each joint in the prediction time domain from the torque sequence to be optimized in the control time domain and the external disturbance. The dynamic model of each joint is as follows: in, This represents the equivalent moment of inertia of the joint. , and These represent the angle, angular velocity, and angular acceleration of the joint, respectively. This indicates the motor output torque of the joint. This indicates the load torque on the joint. This represents the frictional torque of the joint.

[0057] With the first The joint angles and angular velocities of each joint constitute the motion state parameter variables of that joint, and the sum of the load torque and friction torque of that joint is used as the external disturbance parameter of that joint, expressed as follows: From the above dynamic model, the motion state equation of the joint can be obtained as follows: in, Indicates the first Motion state parameter variables of each joint Indicates the first External disturbances of each joint , and The state matrix represents the motion model of the joint.

[0058] From single joints to multiple joints, The joint angles, joint angular velocities, and motor output torques of each joint are arranged in a [formula / structure]. A dimensional vector is represented as: The load torque and friction torque of each joint are also arranged in order. For a multi-dimensional vector, the external perturbation of a multi-joint vector is the sum of the two, expressed as: in, Represents the load torque vector. This represents the friction torque vector. The motion state parameters of the multi-joint system are constructed from the joint angle vectors and joint angular velocity vectors of each joint, and are expressed as follows: Based on this, the motion equations for many joints can be obtained as follows: definition The motion state variables at time t are: The above motion state equations are calculated according to the control period. Discretization, utilizing current Given the state and input at time step, predict the first... +1 step, its recursive formula is: in, , and This is the state matrix obtained by discretizing the state equations of multi-joint motion according to the control period. Similarly, it can be predicted that: In the prediction time domain, since the load change of the robot joints is relatively slow compared to the control cycle, it is assumed that the external disturbance remains constant, i.e.: External disturbance is the sum of load torque and friction torque, neither of which can be directly measured by sensors. Its value in the current control cycle is obtained by inverse solving of the dynamic model. That is, by substituting the actual output torque of the previous control cycle and the joint angular acceleration obtained by quadratic difference of the joint angle into the dynamic model, the external disturbance of the current control cycle can be solved. In the case of a joint torque sensor, it can also be obtained by subtracting the motor output torque from the measured value of the torque sensor. The load change of the robot joint is relatively slow compared to the control cycle. In the prediction time domain, the external disturbance is kept constant at the value of the current control cycle. The prediction of the motion state has a definite disturbance input from the second prediction step onwards.

[0059] This application embodiment also provides a robot joint drive control system, which includes a state acquisition module, a state prediction module, a torque boundary correction module, and a torque solution module. Each module is connected to... Figure 2 The four levels correspond to each other. The state acquisition module is used to acquire the motion and thermal states of each joint. Its input is connected to the position and temperature sensors of each joint, and its output is connected to the input of the state prediction module. The state prediction module is used to predict the motion and thermal states of each joint in the prediction time domain based on the acquisition results. The motion prediction branch and the thermal state prediction branch share the same set of acquisition results. The torque input of the thermal state prediction branch receives the optimal torque sequence output by the torque solving module in the previous control cycle. The torque boundary correction module is used to determine the thermal health factor of each joint based on the predicted thermal state and to adjust the thermal health factor accordingly. The factor corrects the torque constraint boundary of each joint. The thermal health factor is negatively correlated with the predicted thermal state. Its input receives the predicted thermal state output by the state prediction module, and its output sends the corrected torque constraint boundary to the constraint input of the torque solving module. The torque solving module is used to solve the optimal torque sequence in the control time domain within the corrected torque constraint boundary with the optimization objective of minimizing the deviation between the predicted motion state and the obtained reference motion state. It drives each joint according to the first term of the optimal torque sequence. Its output is connected to the motor driver of each joint and simultaneously fed back to the torque input of the state prediction module.

[0060] In application, the above modules can be software program modules, or implemented through different logic circuits integrated in the joint controller, or through multiple distributed processors. Since data is transmitted unidirectionally between the state acquisition module and the state prediction module, between the state prediction module and the torque boundary correction module, and between the torque boundary correction module and the torque solving module in the aforementioned order, and the output of the torque solving module is fed back to the state prediction module, the system completes a closed loop from acquisition to drive in each control cycle. Therefore, thermal state information can be applied to the torque solution in each control cycle.

[0061] The following example, using a 12-DOF humanoid robot performing a transport task, further illustrates the implementation process of the robot joint drive control method provided in this application. Figure 4 As shown, the humanoid robot comprises twelve joint drive units: first shoulder joint 1, second shoulder joint 2, first elbow joint 3, second elbow joint 4, first wrist joint 5, second wrist joint 6, first hip joint 7, second hip joint 8, first knee joint 9, second knee joint 10, first ankle joint 11, and second ankle joint 12. The first shoulder joint 1 and second shoulder joint 2, the first elbow joint 3 and second elbow joint 4, and the first wrist joint 5 and second wrist joint 6 correspond to the left and right upper limb joints, respectively. The first hip joint 7 and second hip joint 8, the first knee joint 9 and second knee joint 10, and the first ankle joint 11 and second ankle joint 12 correspond to the left and right lower limb joints, respectively. Each joint uses a permanent magnet synchronous motor as the drive actuator. The joint angle, joint angular velocity, winding temperature, joint assembly temperature, and motor output torque of each joint are collected in real time by the control system in each control cycle. Specifically, the second elbow joint 4 and the second wrist joint 6... Figure 4 The center is blocked by the load being transported, and its lead wires point to the two joints. Figure 4 The location within.

[0062] This humanoid robot performs a 10-kilogram load transport task, with a task duration set at 120 seconds. The task process includes three phases: grasping the object with both arms, walking while carrying the load, and placing the object. During the grasping phase, the first shoulder joint 1, second shoulder joint 2, first elbow joint 3, and second elbow joint 4 overcome the load's gravity and maintain the grasping posture. The output torque of each joint in the upper limbs increases from zero to the value required to maintain the load, and the winding temperature begins to rise. During the walking phase while carrying the load, the first hip joint 7, second hip joint 8, first knee joint 9, second knee joint 10, first ankle joint 11, and second ankle joint 12 alternately bear the load during the support and swing phases in each gait cycle. The knee joint on the support phase bears the torque generated by the sum of the robot's total weight and the load weight. This phase lasts the longest, and the winding temperature rise of the lower limb joints is greater than that of the upper limb joints. During the object placement phase, the output torque of the upper limb joints gradually unloads, and the lower limb joints enter a static support state. The winding temperature of each joint tends to level off as the motor's heat loss decreases. The controller executes steps S101 to S104 in all three stages, and the torque constraint boundary of each joint is updated periodically according to its own predicted thermal state.

[0063] Different joints use different drive units according to their load requirements. The rated torque and peak torque of each joint are shown in the table below: The safe operating temperature for each joint is uniformly set at 80 degrees Celsius, and the extreme operating temperature for each joint is uniformly set at 120 degrees Celsius. In practical applications, each joint can also be set individually according to the specific thermal design parameters of its actuator. Control cycle Set to 10 milliseconds, prediction time domain length Setting it to 20 controls the time domain length. Set to 5, adjustment factor Set to 0.7, minimum constraint value for thermal health factor. The value is set to 0.3, and the rated torque change rate is taken as 2000 Nm per second. Based on this, the rated torque change of each joint in a unit control cycle is 20 Nm.

[0064] Taking the first knee joint 9 during the load-bearing walking phase as an example, the rated torque of this joint is 70 Nm. During a certain control cycle, the winding temperature is 95 degrees Celsius, the joint assembly temperature is 70 degrees Celsius, the winding heat capacity is 120 joules per degree Celsius, the thermal resistance from the winding to the joint assembly is 0.35 degrees Celsius per watt, the joint assembly heat capacity is 900 joules per degree Celsius, the joint assembly thermal resistance to the environment is 0.8 degrees Celsius per watt, and the joint ambient temperature is 30 degrees Celsius. Using the optimal torque sequence from the previous control cycle as the torque input, the motor heat loss of this joint in each prediction step is approximately 183.6 watts. Substituting this into the thermal network model and recursively extrapolating, we can obtain that the winding temperature rises at a rate of approximately 0.93 degrees Celsius per second, and the joint assembly temperature rises at a rate of approximately 0.02 degrees Celsius per second. At the end of the prediction time domain of 20 control cycles, or 0.2 seconds, the winding temperature is approximately 95.19 degrees Celsius, and the joint assembly temperature is approximately 70.01 degrees Celsius.

[0065] Weighted summing the two temperatures with an adjustment factor of 0.7, the comprehensive thermal state index for this prediction step is approximately 87.63 degrees Celsius, which is the maximum value within the prediction time domain. This means the maximum comprehensive thermal state index of the first knee joint 9 is approximately 87.63 degrees Celsius. This value is higher than the safe operating temperature of 80 degrees Celsius and lower than the extreme operating temperature of 120 degrees Celsius. Substituting this value into the calculation formula for the thermal health factor, the thermal health factor of the first knee joint 9 is approximately 0.87. Multiplying this thermal health factor by the rated torque of 70 Nm yields a maximum permissible output torque of approximately 60.9 Nm for the first knee joint 9 in this control cycle. Multiplying this thermal health factor by the rated torque change of 20 Nm within a unit control cycle yields a maximum permissible output torque change of approximately 17.4 Nm. The control layer uses 60.9 Nm and 17.4 Nm as the torque amplitude constraint and torque change constraint for this joint, and substitutes them together with the constraints of the other joints into the optimization problem to obtain the optimal torque sequence in the control time domain. The first term is then taken and output to the motor driver of the first knee joint 9.

[0066] Within the same control cycle, the winding temperature of the first wrist joint 5 is 52 degrees Celsius, and the joint assembly temperature is 45 degrees Celsius. Its comprehensive thermal state index is approximately 49.9 degrees Celsius, which is lower than the safe operating temperature of 80 degrees Celsius. The thermal health factor of this joint is set to 1, and its maximum allowable output torque remains at the rated torque of 5 Nm. The torque constraint boundary is not tightened. It can be seen that the thermal health factor is calculated for each joint individually. The first knee joint 9, which has a heavier load and a faster temperature rise, has its maximum allowable output torque reduced first, while the first wrist joint 5, which has a slower temperature rise, still outputs torque at its rated capacity. The movement of the entire machine is not interrupted by the reduction of the torque upper limit of a single joint.

[0067] As the handling task progresses, the maximum value of the comprehensive thermal state index of the first knee joint 9 continues to rise. When this value reaches 120 degrees Celsius, which is the limit operating temperature, its thermal health factor is set to the minimum constraint value of 0.3, and the maximum allowable output torque drops to 21 Nm. The joint still retains some of the torque output capacity required to support the body, and the controller does not reduce the output torque of the joint to zero due to thermal protection. The above parameters are only used to illustrate the implementation process of the method provided in the embodiments of this application. Specific parameters can be adjusted according to the specifications of different models of humanoid robot joint actuators.

[0068] In another embodiment, the robot joint drive control method provided in this application is applied to a multi-joint robotic arm. The robotic arm has six joints driven by permanent magnet synchronous motors via reducers. From the base to the end effector, these joints are, in order: base rotation joint, upper arm pitch joint, forearm pitch joint, wrist rotation joint, wrist pitch joint, and end effector rotation joint. The rated torque of each joint decreases sequentially from the base to the end effector. During palletizing operations, the robotic arm reciprocates to perform picking, moving, and placing actions. The base rotation joint and upper arm pitch joint bear larger load torques, and their winding temperature rises faster than that of the wrist joints. The controller also executes steps S101 to S104, collecting the joint angle, joint angular velocity, winding temperature, and joint assembly temperature of the six joints in each control cycle. Based on the thermal network model of each joint, the winding temperature and joint assembly temperature in the prediction time domain are predicted. This yields the maximum value of the comprehensive thermal state index and the thermal health factor for each joint, and the torque constraint boundaries of each joint are corrected accordingly. The safe operating temperature and extreme operating temperature of each joint are set according to the thermal design of its actuator. For example, the base rotary joint has a safe operating temperature of 75 degrees Celsius and an extreme operating temperature of 115 degrees Celsius, while the wrist rotary joint has a safe operating temperature of 85 degrees Celsius and an extreme operating temperature of 125 degrees Celsius. During continuous palletizing, the thermal health factor of the base rotary joint decreases first, and its maximum allowable output torque and the change in maximum allowable output torque decrease accordingly. The movement cycle of the robotic arm slows down accordingly, and the palletizing operation is not interrupted. The thermal health factor of each joint of the wrist remains constant, and its output capacity is not affected.

[0069] The robot joint drive control method provided in this application is applicable to humanoid robots with upper and lower limb joints, as well as multi-joint robotic arms with joints arranged in series along a single kinetic chain. Its implementation does not depend on the specific arrangement of the joints on the body.

[0070] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps in the above-described method embodiments.

[0071] This application provides a computer program product, including a computer program, which, when run on an electronic device, enables the electronic device to implement the steps described in the various method embodiments above.

[0072] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0073] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A robot joint drive control method, characterized in that, include: Collect the motion and thermal states of each joint; Based on the collected results, predict the motion and thermal states of each joint in the prediction time domain; The thermal health factor of each joint is determined based on the predicted thermal state. The thermal health factor is negatively correlated with the predicted thermal state, and the torque constraint boundary of each joint is corrected according to the thermal health factor. With the goal of minimizing the deviation between the predicted motion state and the obtained reference motion state, the optimal torque sequence in the control time domain is solved within the corrected torque constraint boundary, and each joint is driven according to the first term of the optimal torque sequence.

2. The robot joint drive control method as described in claim 1, characterized in that, The acquisition of the motion and thermal states of each joint includes: Collect the joint angles and angular velocities of each joint to determine their motion state. The winding temperature and joint assembly temperature of each joint are collected as the thermal state of each joint.

3. The robot joint drive control method as described in claim 2, characterized in that, The determination of thermal health factors for each joint based on the predicted thermal state includes: The predicted winding temperature and the joint assembly temperature are weighted and summed to obtain the comprehensive thermal state index of each joint at each prediction step in the prediction time domain. The calculation formula is as follows: in, This indicates the comprehensive thermal state index. This indicates the predicted winding temperature. This indicates the predicted temperature of the joint assembly. Indicates the regulating factor; The maximum value of the comprehensive thermal state index of each joint in each prediction step within the prediction time domain is taken as the maximum value of the comprehensive thermal state index of each joint. The thermal health factor of each joint is determined based on the maximum value of the comprehensive thermal state index of each joint. The calculation formula is as follows: in, Indicates the joint number. Indicates the first Thermal health factors of each joint Indicates the first The maximum value of the comprehensive thermal state index of each joint. Indicates the first The safe operating temperature of each joint. Indicates the first The maximum operating temperature of each joint. This represents the minimum constraint value for thermal health factors, and Greater than zero.

4. The robot joint drive control method as described in claim 1, characterized in that, The torque constraint boundary includes the maximum permissible output torque and the maximum permissible output torque variation. The step of correcting the torque constraint boundaries of each joint according to the thermal health factor includes: The maximum permissible output torque of each joint is the product of the thermal health factor of each joint and the rated torque of the motor of each joint. The maximum permissible output torque variation of each joint is calculated by multiplying the thermal health factor of each joint by the rated torque variation of the motor of each joint within a unit control cycle.

5. The robot joint drive control method as described in claim 2, characterized in that, The prediction of the thermal state of each joint in the prediction time domain includes: The optimal torque sequence obtained from the previous control cycle is used as the torque input for each prediction step in the prediction time domain. Based on the torque input and the predicted joint angular velocity, the motor heat loss of each joint in each prediction step is determined. The motor heat loss includes winding copper loss proportional to the square of the torque input and friction loss proportional to the joint angular velocity. Using the collected winding temperature and joint assembly temperature as initial values, the thermal state of each joint in the prediction time domain is gradually derived based on the thermal network model, the motor heat loss, and the obtained joint ambient temperature.

6. The robot joint drive control method as described in claim 5, characterized in that, Each joint is driven by a permanent magnet synchronous motor, and the formula for calculating the motor heat loss of each joint in each prediction step is as follows: in, Indicates the first Heat loss of the motor in each joint, Indicates the first The resistance of the motor windings at each joint, Indicates the first Motor torque coefficient of each joint Indicates the first Torque input of each joint, Indicates the first The frictional torque of each joint, Indicates the first The joint angular velocity of each joint.

7. The robot joint drive control method as described in claim 5, characterized in that, The thermal network model is a two-node thermal network model, wherein the two nodes of the two-node thermal network model are the winding node and the joint assembly node, respectively. The heat loss of the motor is injected into the winding node, the winding node transfers heat to the joint assembly node through the winding thermal resistance, and the joint assembly node dissipates heat to the environment through the thermal resistance from the joint assembly to the environment.

8. The robot joint drive control method according to any one of claims 1 to 7, characterized in that, The optimization objective of minimizing the deviation between the predicted motion state and the acquired reference motion state includes: The optimization objective is to minimize the weighted sum of the motion state tracking term and the control torque smoothing term, whereby the motion state tracking term represents the deviation and the control torque smoothing term represents the torque change between adjacent control cycles in the control time domain.

9. The robot joint drive control method as described in claim 2, characterized in that, The prediction of the motion state of each joint in the prediction time domain includes: The sum of the load torque and friction torque of each joint is used as the external disturbance of each joint, and the external disturbance is kept constant at the value of the current control cycle within the prediction time domain. Using the acquired joint angles and joint angular velocities as initial values, and based on the joint dynamics model, the joint angles and joint angular velocities of each joint in the prediction time domain are gradually derived from the torque sequence to be optimized in the control time domain and the external disturbance.

10. A robot joint drive control system, characterized in that, include: The status acquisition module is used to collect the motion and thermal states of each joint. The state prediction module is used to predict the motion state and thermal state of each joint in the prediction time domain based on the acquisition results. The torque boundary correction module is used to determine the thermal health factor of each joint based on the predicted thermal state, and to correct the torque constraint boundary of each joint according to the thermal health factor. The thermal health factor is negatively correlated with the predicted thermal state. The torque solving module is used to solve for the optimal torque sequence in the control time domain within the corrected torque constraint boundary, with the optimization objective of minimizing the deviation between the predicted motion state and the acquired reference motion state, and to drive each joint according to the first term of the optimal torque sequence.

Images