Humanoid robot bionic skin heating adaptive power distribution method

Abstract

This invention belongs to the field of robot control, specifically relating to an adaptive power allocation method for biomimetic skin heating in humanoid robots. It aims to address the problem of inadequate heating control for humanoid robots performing complex dynamic tasks. The method includes: acquiring target data; inputting the target data into a pre-constructed digital twin model to predict the thermal state change trend of each skin region within a preset future period, and identifying key areas for priority heating based on the predicted motion state; constructing and solving a multi-objective optimization model using the heating power allocated to each skin region as a decision variable; and determining the heating control command for each skin region based on the obtained power allocation value. This invention utilizes a digital twin model and predictive motion state to anticipate thermal state trends caused by future robot actions and environmental changes, effectively improving the heating control performance of robots under complex working conditions.

Description

Technical Field

[0001] This invention belongs to the field of robot control, and specifically relates to an adaptive power distribution method for bionic skin heating of a humanoid robot. Background Technology

[0002] Humanoid robots are increasingly being deployed in complex and dynamic unstructured environments such as outdoor patrols, disaster relief, and human-robot collaboration. Their bionic skin, as a key medium for interaction with the environment, is crucial for the accuracy of embedded sensors, the physical properties of bionic materials, and the safety of human-robot interaction, as its temperature stability is essential.

[0003] Existing robot heating control technologies, whether threshold-based PID (proportional-integral-derivative) control or schemes with preset fixed power levels, essentially simplify thermal management into a localized, reactive temperature maintenance problem. These methods are adequate for simple working conditions where the robot is static or in a stable environment, but their inherent limitations become apparent when humanoid robots perform complex dynamic tasks. For example, robot movements (such as high-speed arm swings) cause drastic and uneven changes in the convective heat transfer coefficients of various parts of the surface; tasks (such as grasping cold tools) introduce strong localized contact heat conduction, resulting in highly spatiotemporal heterogeneous and dynamically changing surface heat demands. Therefore, reactive control, due to its inherent lag, cannot effectively compensate for significant heat loss before it occurs; while static power allocation strategies cannot adapt to dynamically changing heat demands, leading to waste in low-demand areas and insufficient heating in high-demand areas. Summary of the Invention

[0004] To address the aforementioned problems in existing technologies, namely the difficulty in providing reasonable heating control for humanoid robots performing complex dynamic tasks, this invention proposes an adaptive power allocation method for heating biomimetic skin in humanoid robots. The method includes:

[0005] Acquire target data, which includes the surface temperature of each skin region of the robot, environmental parameters of the environment, and the robot's real-time and predictive motion state;

[0006] The target data is input into a pre-built digital twin model to predict the thermal state change trend of each skin region within a future preset period, and the key areas that are preferentially heated are identified in combination with the predictive motion state.

[0007] A multi-objective optimization model is constructed and solved using the heating power allocated to each skin region as the decision variable. The multi-objective optimization model uses the power allocation value of each skin region as the decision variable, takes the temperature of the key region not being lower than a preset threshold as the core constraint, and takes minimizing the total energy consumption and maximizing the temperature field uniformity of the robot's overall surface as the optimization objective. Its optimization constraints are the upper limit of the total power of each skin region and the power limit of the heating unit corresponding to each skin region.

[0008] Based on the power distribution values ​​obtained from the solution, the heating control commands for each skin region are determined.

[0009] In some preferred embodiments, acquiring the target data includes:

[0010] By pre-integrating temperature sensors under each skin area, real-time temperature readings of N independent areas on the robot surface are collected concurrently at a preset sampling frequency, forming an N-dimensional temperature state vector, where N is a positive integer;

[0011] Collect ambient temperature, relative humidity, and triaxial air velocity data to construct an environmental parameter vector;

[0012] A unified global timestamp is added to the temperature state vector and the environmental parameter vector, encapsulated into a data structure that meets a preset format, and the data structure is used as input to transmit to the computing unit running the digital twin model.

[0013] In some preferred embodiments, acquiring the target data further includes:

[0014] Access the robot’s main motion controller and extract the predetermined task trajectory within the next T seconds. The predetermined task trajectory includes the target angle, angular velocity and angular acceleration commands in the robot’s joint space.

[0015] Using a robot forward kinematics model established based on DH parameters, the predetermined task trajectory is calculated frame by frame and converted into time series data of three-dimensional coordinates, linear velocity, and linear acceleration of M control points preset on the robot's bionic skin surface in the world coordinate system. The time series data is then used as the predictive motion state of the robot.

[0016] In some preferred embodiments, identifying key areas for preferential heating by incorporating the predictive motion state includes:

[0017] Based on the predicted motion state, the maximum linear velocity of each skin region control point within a future preset period is determined, and regions with maximum linear velocities exceeding a first preset threshold are marked as candidate key regions.

[0018] Based on the predicted motion state, determine the skin area that will come into contact with an object with a temperature below a second preset threshold within a future preset period, and add it to the candidate key area;

[0019] Perform a logical union operation on each skin region within the candidate key region to determine the key region.

[0020] In some preferred embodiments, the digital twin model is a multi-layered three-dimensional transient thermodynamic finite element model, and the prediction of the thermal state change trend of each skin region within a preset period includes:

[0021] The surface temperature data of each skin region is used as the initial temperature field condition of the digital twin model, and the environmental parameters are used to define the external boundary conditions of the model.

[0022] The heat generated by the movement of the robot's internal joints is used as an internal heat source. By solving the unsteady heat conduction partial differential equation, forward integration is performed in the time domain to output a predicted sequence of temperature changes of each grid node over time within a preset period.

[0023] In some preferred embodiments, the temperature field uniformity of the robot's overall surface is maximized, and the methods for achieving this include:

[0024] In the objective function of the multi-objective optimization model, a non-uniformity evaluation function is defined, which is used to calculate the overall variance of the set of predicted temperature values ​​for all N skin regions at the end of the prediction cycle.

[0025] Minimizing the non-uniformity evaluation function is used as the optimization objective of the multi-objective optimization model, and is processed in parallel with the objective of minimizing total energy consumption.

[0026] Under the premise of satisfying the core constraints, determine the power allocation decision variables that enable the two-dimensional objective vector consisting of minimizing the non-uniformity evaluation function and minimizing total energy consumption to achieve Pareto optimality.

[0027] In some preferred embodiments, the solution process of the multi-objective optimization model includes:

[0028] A population is randomly initialized, where each individual is a real-number encoded chromosome containing N skin region heating power allocation values;

[0029] For each individual in the population, the temperature field is predicted using a digital twin model based on its corresponding power allocation value, and two objective function values, total energy consumption and temperature field variance, are calculated.

[0030] Based on the calculated objective function value, the population is subjected to non-dominated sorting and crowding calculation. Based on the sorting results and crowding, the population is iterated through a polynomial mutation genetic operator to generate offspring populations.

[0031] Merge the parent and child generations, repeat the above steps until the preset number of iterations or convergence conditions are met, and output the optimal solution set of the current model.

[0032] In some preferred embodiments, determining the heating control command for each skin region based on the power allocation value obtained from the solution includes:

[0033] From the set of optimal solutions obtained by solving the problem, one of the power allocation vectors is selected as the execution scheme for the current control cycle according to the preset preference strategy;

[0034] Each determined power allocation value is converted into the duty cycle of a pulse width modulation signal based on the electrical characteristics of its corresponding heating unit.

[0035] The calculated N duty cycle values ​​are then sent to the corresponding N skin area heating units.

[0036] In some preferred embodiments, the preset threshold of the key region is set by the following process:

[0037] At the beginning of each control cycle, query the robot's task scheduling system for the currently executing or upcoming task identifiers.

[0038] Using the queried task identifier as the key, the corresponding temperature lower limit structure is retrieved from the preset task-threshold mapping table, and the values ​​corresponding to each key area are extracted as the preset threshold. The task-threshold mapping table is a key-value pair data structure, where the key is the robot's task identifier and the value is the temperature lower limit structure corresponding to different parts of the robot.

[0039] The beneficial effects of this invention are:

[0040] This invention utilizes digital twin models and predictive motion states to anticipate thermal trends caused by future robot actions and environmental changes. This allows power allocation decisions to be based not on past temperature deviations, but on accurate predictions of future heat demands. Furthermore, this invention effectively overcomes the response lag problem of traditional control methods, ensuring that the temperature in critical areas remains stable even under severe dynamic disturbances, thus significantly improving the robot's heating control performance under complex operating conditions.

[0041] Meanwhile, this invention constructs the thermal management problem as a multi-objective optimization model that includes core constraints and dual optimization objectives (total energy consumption and temperature uniformity). By solving its optimal set, it provides a series of trade-off solutions, solving the problem that multiple conflicting performance indicators are difficult to balance. Attached Figure Description

[0042] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0043] Figure 1 This is a flowchart illustrating an adaptive power allocation method for biomimetic skin heating of a humanoid robot proposed in an embodiment of the present invention.

[0044] Figure 2 This is a schematic diagram of the framework of a humanoid robot bionic skin heating adaptive power distribution system proposed in an embodiment of the present invention.

[0045] Figure 3 This is a schematic diagram of the structure of a computer system proposed in an embodiment of the present invention. Detailed Implementation

[0046] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0047] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0048] Please refer to Figure 1 The first embodiment of the present invention provides a method for adaptive power allocation for heating the bionic skin of a humanoid robot, comprising:

[0049] Step S10: Obtain target data, which includes the surface temperature of each skin area of ​​the robot, environmental parameters of the environment, and the real-time and predictive motion state of the robot.

[0050] Step S20: Input the target data into a pre-constructed digital twin model to predict the thermal state change trend of each skin region within a future preset period, and identify the key areas that are preferentially heated by combining the predictive motion state.

[0051] Step S30: Using the heating power allocated to each skin region as the decision variable, construct and solve a multi-objective optimization model; wherein, the multi-objective optimization model uses the power allocation value of each skin region as the decision variable, takes the temperature of the key region not being lower than a preset threshold as the core constraint, and takes minimizing the total energy consumption and maximizing the temperature field uniformity of the robot's overall surface as the optimization objective, and its optimization constraints are the upper limit of the total power of each skin region and the power limit of the heating unit corresponding to each skin region;

[0052] Step S40: Determine the heating control command for each skin region based on the power distribution value obtained from the solution.

[0053] The method proposed in this embodiment can be implemented by a humanoid robot that has bionic skin and can adaptively heat the bionic skin and control its heating power.

[0054] For example, this method is used to heat the bionic skin on a six-degree-of-freedom collaborative robot arm. The robot arm is deployed in an automated warehouse environment with an ambient temperature of 5°C and irregular airflow. Its task is to identify and grasp metal and plastic parts of different materials. The bionic skin of the robot arm is divided into 50 independent heating zones (N=50), each zone is equipped with an independent thin-film resistance heating unit and a microelectromechanical system (MEMS) temperature sensor.

[0055] The specific process of the method in this embodiment will be described below in conjunction with the robot described above.

[0056] Specifically, this method can be executed by the robot's control system, which first performs data acquisition operations in a loop with a control cycle of 20 milliseconds.

[0057] For example, a MEMS temperature sensor array deployed under a silicone substrate in 50 skin regions concurrently acquires real-time temperature readings of each region at a sampling frequency of 50 Hz. These readings constitute a 50-dimensional real-time temperature vector.

[0058] At the same time, the environmental perception module installed on the robot base is invoked. This module integrates a high-precision digital temperature and humidity sensor and a three-axis ultrasonic anemometer to obtain the current ambient temperature, relative humidity H_env, and three-dimensional air velocity vector in real time.

[0059] Simultaneously, the current angles and angular velocities of all joints are directly read from the robot's motion controller. For example, the planned trajectory for the next 5 seconds (i.e., a preset period T=5s) can be extracted by accessing the robot's task planner. This trajectory is given as a series of timestamp-joint angle target points. Using the established robot forward kinematics model (based on the DH parameter method), the joint spatial trajectory is solved into a time series of the three-dimensional spatial position, linear velocity, and linear acceleration of the center point of each skin region in the world coordinate system.

[0060] In this embodiment, the pre-constructed digital twin model is a three-dimensional transient thermodynamic model based on the finite element method (FEM). The model is structurally identical to the physical entity of the robot and includes the precise geometry and material thermophysical properties (thermal conductivity, specific heat capacity, density, etc.) of the skin layer, heating layer, insulation layer, and internal skeleton.

[0061] At the beginning of each control cycle, the real-time temperature vector obtained in the preceding steps is used as the initial temperature field condition for the finite element model, and the environmental parameters and the predicted relative velocity sequence are used as the boundary conditions of the model. The model solves the unsteady heat conduction equation, performs forward integration in the time domain, and outputs the temperature change trend curves of all skin regions within the next 5 seconds.

[0062] Simultaneously, based on the aforementioned predictive motion states, key areas in multiple skin regions are identified and determined, including:

[0063] Analyze the velocity prediction sequence for the next 5 seconds and mark any skin area with a predicted peak linear velocity exceeding 0.8 m / s (e.g., the end of the arm during rapid swinging) as a candidate key region;

[0064] Parsing task instructions, for example, when a grasping instruction is recognized, based on the target object attributes identified by the vision system, if the target is a metal part (whose temperature is close to the ambient temperature by default, i.e., 5°C), then the skin area of ​​the palm and fingertips of the hand performing the grasping task will be marked as a candidate key area.

[0065] All marked candidate regions are merged to form the final list of key regions.

[0066] Based on the above steps, a multi-objective optimization model is constructed and solved, specifically including:

[0067] The decision variables, namely the heating power allocation values ​​for 50 skin regions, are determined and formed into a vector P;

[0068] For all regions i identified as critical regions in the list, their lowest predicted temperature within the next 5-second prediction period must not be lower than a preset threshold (set to 20°C in this embodiment).

[0069] Its optimization objectives include:

[0070] Objective 1 (Energy Consumption): Minimize total heating power;

[0071] Objective 2 (Uniformity): Maximize the uniformity of the temperature field, which is equivalent to minimizing the variance of the predicted temperature of all 50 regions at the end of the prediction period;

[0072] Its optimization constraints include:

[0073] Maximum total power: For example, the maximum available heating power provided by the robot's battery management system is 200 watts;

[0074] Individual power limits: For example, the rated maximum power of each heating unit is 8 watts.

[0075] For example, the non-dominated sorting genetic algorithm with an elitist strategy (NSGA-II) can be used to solve this multi-objective optimization problem. Each individual in the algorithm's population is a 50-dimensional power allocation vector P. During the iteration process, for each individual, a digital twin model is used as the fitness function evaluator to calculate its corresponding total energy consumption and temperature variance. After selection, crossover, and mutation operations, the algorithm finally converges and outputs a set of Pareto optimal solutions. This set of solutions represents the optimal power allocation schemes with different trade-offs between energy consumption and uniformity under all constraints.

[0076] As a feasible implementation method, taking the Pareto optimal solution set as an example, a final execution solution is selected based on the robot's current working mode. For example:

[0077] Energy-saving cruise mode: Selects the solution with the lowest total energy consumption in the solution set;

[0078] Human-computer interaction mode: Select the solution with the minimum temperature variance in the solution set;

[0079] Standard operating mode: Select the inflection point solution on the Pareto front, that is, the solution that best balances energy consumption and uniformity.

[0080] Assuming that the optimal power allocation vector is selected in the above process, the system converts each power value into the duty cycle of the corresponding pulse width modulation (PWM) signal;

[0081] Subsequently, the main controller sends 50 duty cycle values ​​to 50 slave microcontrollers (MCUs) controlling the corresponding skin area heating units via the Controller Area Network (CAN) bus. Upon receiving the instructions, each MCU immediately generates a PWM waveform with the specified duty cycle to precisely drive the thin-film heater connected to it.

[0082] In the above implementation, by repeatedly executing the above steps, the robot can proactively and preferentially ensure the temperature stability of key parts in a dynamic and complex environment in the most energy-efficient way, while also taking into account the thermal uniformity of the overall surface.

[0083] Furthermore, in the above embodiments, acquiring the target data includes:

[0084] Temperature sensors pre-integrated under each skin region are used to concurrently collect real-time temperature readings from N independent areas on the robot's surface at a preset sampling frequency, forming an N-dimensional temperature state vector where N is a positive integer. Environmental temperature, relative humidity, and triaxial airflow velocity data are collected to form an environmental parameter vector. A unified global timestamp is added to the temperature state vector and the environmental parameter vector, and the data is encapsulated into a data structure that meets a preset format. The data structure is then used as input and transmitted to the computing unit that runs the digital twin model.

[0085] The data structure is used to encapsulate the robot's thermal state information, and to aggregate the scattered sensor data in a standardized way and add a unified timestamp to ensure that the data input to the digital twin model has time synchronization and atomicity.

[0086] It is easy to understand that this data structure logically consists of three core components:

[0087] Global timestamp: Records the precise moment when this data snapshot was created;

[0088] Environmental parameters: These encapsulate all thermodynamic parameters that describe the external environment in which the robot operates.

[0089] Skin temperature status: An embedded substructure that encapsulates temperature readings for all independently heated areas on the robot's surface.

[0090] Furthermore, in the above embodiments, acquiring the target data further includes:

[0091] Access the robot's main motion controller to extract the predetermined task trajectory within the next T seconds. The predetermined task trajectory includes the target angle, angular velocity, and angular acceleration commands in the robot's joint space. Using the robot's forward kinematics model established based on DH parameters, perform frame-by-frame calculations on the predetermined task trajectory to convert it into time-series data of the three-dimensional coordinates, linear velocity, and linear acceleration of M control points preset on the robot's bionic skin surface in the world coordinate system. Use the time-series data as the robot's predictive motion state.

[0092] Specifically, taking the six-degree-of-freedom robot in the above embodiment as an example, its standard Denavit-Hartenberg (DH) parameter table is established according to its mechanical structure manual.

[0093] For example, the table contains six rows, each corresponding to a joint, and columns containing four parameters: link twist angle, link length, link offset, and joint angle. Based on this DH table, the robot's forward kinematics function is constructed, where the input is a 6-dimensional joint angle vector, and the output is a 4x4 homogeneous transformation matrix that describes the pose of the robot's end effector (sixth link) relative to the robot's base (world coordinate system).

[0094] Meanwhile, through the communication interface of the robot operating system, the robot subscribes to messages published by the main motion controller. These messages indicate a predetermined task trajectory for a future period of time (in this embodiment, the preset period T = 5 seconds). The trajectory data is organized into a series of waypoints with timestamps.

[0095] For example, assuming 500 waypoints are extracted, batch processing is performed on each of the 500 extracted waypoints, frame by frame (i.e., time point by time). For each time point t in the trajectory:

[0096] First, calculate the pose of each link in the coordinate system. Using the forward kinematics model, calculate the homogeneous transformation matrix from the base to each link based on the joint angles of the current frame.

[0097] For the Control points (Assuming it is attached to the i-th link), its three-dimensional coordinates in the world coordinate system Calculated using the following formula:

[0098] ;

[0099] in It is the constant coordinate vector of the control point in the local coordinate system of its corresponding link i;

[0100] Then, using the robot's geometric Jacobian matrix, the linear velocity of each control point is calculated. :

[0101] ;

[0102] in, It is the velocity component of the Jacobian matrix associated with control point Pⱼ. It is the joint angular velocity vector of the current frame. The Jacobian matrix itself is also the joint angle. The function.

[0103] linear acceleration of control points The calculation formula is:

[0104] ;

[0105] in It is the time derivative of the Jacobian matrix, and its calculation depends on and ; It is the joint angular acceleration vector of the current frame.

[0106] Furthermore, in the above embodiments, by combining the predictive motion state, key areas that are preferentially heated are identified, including:

[0107] Based on the predicted motion state, determine the maximum linear velocity of each skin region control point within a future preset period, and mark the region with the maximum linear velocity exceeding a first preset threshold as a candidate key region; based on the predicted motion state, determine the skin regions that will come into contact with objects with a temperature lower than a second preset threshold within a future preset period, and add them to the candidate key regions; perform a logical union operation on each skin region within the candidate key regions to determine the key region.

[0108] The first preset threshold is the maximum linear velocity threshold, which is set to, for example, 1.2 m / s. Its value can be obtained from experimental data. When the movement speed of the skin area exceeds this value, the convective exchange effect between the skin and the ambient air will be significantly enhanced, resulting in rapid heat loss.

[0109] The second preset threshold is the target object temperature threshold. If it is set to 10.0 degrees Celsius, in a warehouse environment, objects with a temperature lower than this (usually metal parts) will cause a drastic local temperature drop (conductive cooling) in the skin area that comes into contact with them, affecting the robot's tactile perception accuracy.

[0110] Furthermore, in the above embodiments, the digital twin model is a multi-layered three-dimensional transient thermodynamic finite element model, and the prediction of the thermal state change trend of each skin region within a future preset period includes:

[0111] The surface temperature data of each skin region is used as the initial temperature field condition of the digital twin model, and the environmental parameters are used to define the external boundary conditions of the model. The heat generated by the movement of the robot's internal joints is used as the internal heat source. By solving the unsteady heat conduction partial differential equation, forward integration is performed in the time domain to output the predicted sequence of temperature changes of each grid node over time within a preset period.

[0112] The construction of the digital twin model can be based on the robot's CAD model, extracting the precise geometric shape of its outer skin and internal support structure. For example, a hybrid tetrahedral and hexahedral mesh can be used to spatially discretize this geometric domain to generate a three-dimensional finite element mesh. It includes at least two material layers: a surface biomimetic skin material layer and an inner structural support layer, and each layer is assigned corresponding material thermal properties, including density, specific heat capacity, and anisotropic thermal conductivity.

[0113] The surface temperature data of each skin region collected at the current moment through the distributed temperature sensor network is mapped to the corresponding surface nodes of the finite element model through an interpolation algorithm, and this is used as the initial condition for the entire three-dimensional temperature field transient simulation.

[0114] Meanwhile, based on the measured ambient wind speed and air temperature, the surface convection heat transfer coefficient is calculated according to an empirical formula and applied to all unit surfaces of the robot's outer surface that come into contact with the air. This embodiment does not limit the specific empirical formula.

[0115] It is easy to understand that changes in the robot's motion state will update the intensity and distribution of the internal heat source in real time. Therefore, in this embodiment, the heat generated by the robot's internal joint motors, reducers, and other moving parts during operation is equivalent to an internal heat source. The power of this internal heat source can be determined by consulting a mapping table of the relationship between motor current, speed, and efficiency, and then distributed to the corresponding mesh elements within the finite element model according to their spatial location.

[0116] After completing the above settings, the three-dimensional unsteady heat conduction partial differential equation is solved and numerically integrated. Starting from the initial moment, the calculation is gradually advanced with a fixed time step to simulate the temperature field evolution within a preset period (e.g., the next 5-10 minutes). Finally, a complete predicted sequence of temperature changes over time for each skin region during this period is output.

[0117] Furthermore, in the above embodiments, the temperature field uniformity of the robot's overall surface is maximized, and the method for achieving this includes:

[0118] In the objective function of the multi-objective optimization model, a non-uniformity evaluation function is defined, which is used to calculate the overall variance of the set of predicted temperature values ​​for all N skin regions at the end of the prediction period. The minimization of the non-uniformity evaluation function is taken as the optimization objective of the multi-objective optimization model and is processed in parallel with the total energy consumption minimization objective. Under the premise of satisfying the core constraints, the power allocation decision variable that makes the two-dimensional objective vector composed of the minimization of the non-uniformity evaluation function and the minimization of total energy consumption Pareto optimal is determined.

[0119] The non-uniformity evaluation function is defined as the total variance of the predicted temperature values ​​of all N skin regions at the end of the prediction time domain. Minimizing the non-uniformity evaluation function is taken as an independent optimization objective, which together with the original total energy consumption minimization objective forms a two-dimensional objective vector. In the optimization model, this two-dimensional objective vector and the core constraints participate in the optimization solution.

[0120] For example, in this embodiment, a weighted summation method can be used to transform the multi-objective problem into a single-objective problem for processing. For instance, a comprehensive objective function can be constructed, where the weight coefficients reflect the degree of importance attached to the energy consumption objective and the uniformity objective, respectively. By adjusting the ratio of the weight coefficients, a series of Pareto optimal solutions can be generated.

[0121] Within each control cycle, under the conditions of satisfying the total power constraint and the power limits of each region, the power allocation scheme that minimizes the comprehensive objective function is solved. A sequential quadratic programming algorithm can be used in the solution process to find the optimal power allocation decision variables that satisfy all constraints through iterative calculation. Finally, the obtained power allocation values ​​are converted into control command outputs for each heating unit.

[0122] Furthermore, in the above embodiments, the solution process of the multi-objective optimization model includes:

[0123] A population is randomly initialized, where each individual is a real-number encoded chromosome containing N skin region heating power allocation values. For each individual in the population, the temperature field is predicted using a digital twin model based on its corresponding power allocation value, and two objective function values ​​are calculated: total energy consumption and temperature field variance. Based on the calculated objective function values, the population is subjected to non-dominated sorting and crowding calculation. According to the sorting results and crowding, the population is iterated using a multinomial mutation genetic operator to generate offspring. The parent and offspring generations are merged, and the above steps are repeated until the preset number of iterations or convergence conditions are met, and the optimal solution set of the current model is output.

[0124] In this embodiment, the population size is first set to M, and each individual is represented by a real number encoding method to represent a complete power allocation scheme, that is, the chromosome form is [P1,P2,...,P_N], where P_i represents the heating power value of the i-th skin region. Under the condition of satisfying the total power constraint of the system and the power limit of each region, the initial population is randomly generated.

[0125] For example, the following calculation is performed for each individual in the population:

[0126] The chromosome is decoded into a specific power allocation value; this power allocation is used as input to perform forward simulation through a digital twin model to obtain the temperature field distribution at the end of the predicted time domain; two objective function values ​​are calculated: total energy consumption and temperature field non-uniformity.

[0127] The population is non-dominated and ordered according to the objective function value, dividing it into multiple frontier levels. Individuals in the first frontier are considered optimal, followed by the second frontier, and so on. Within each frontier, the crowding degree of each individual is calculated to evaluate the distribution density of individuals in the objective space.

[0128] Based on this, a binary tournament selection mechanism can be adopted, prioritizing individuals with high frontier rank and high crowding as parents. Gene recombination is performed by simulating binary crossover operators, and perturbations are introduced using polynomial mutation operators to generate offspring populations. Parents and offspring are then merged to form a new population of size 2M.

[0129] The merged population is re-sorted for non-dominated ordering and crowding calculation. Based on the sorting results, the top M individuals are selected as the new generation population. This iterative process is repeated until the preset maximum number of iterations G_max is reached, or the quality of the solution set no longer improves significantly in consecutive iterations. Finally, all solutions of the first frontier are output as the Pareto optimal solution set.

[0130] Furthermore, in the above embodiments, the process of setting the preset threshold of the key area includes:

[0131] At the beginning of each control cycle, the task identifier currently being executed or about to be executed is queried from the robot's task scheduling system. Using the queried task identifier as the key, the corresponding temperature lower limit structure is retrieved from the preset task-threshold mapping table, and the values ​​corresponding to each key area are extracted as the preset threshold. The task-threshold mapping table is a key-value pair data structure, where the key is the robot's task identifier and the value is the temperature lower limit structure corresponding to different parts of the robot.

[0132] Specifically, at the beginning of each control cycle, the identifier of the currently executing task is obtained in real time through the robot's task scheduling system interface. This identifier uses a unified task coding format and can uniquely identify the robot's working status and task type.

[0133] In this embodiment, a task-threshold mapping table is pre-established. This table uses a hash table data structure to achieve fast lookup. Each entry in the table contains two fields: a task identifier (key), a string type, storing a standardized task name; and a temperature threshold structure (value), a structure type containing multiple subfields, each corresponding to the minimum temperature requirement of different key parts of the robot.

[0134] Using the acquired task identifier as the query key, a lookup operation is performed in the task-threshold mapping table. If a matching entry is found, the corresponding temperature threshold structure is read; otherwise, the default temperature threshold structure is used. The specific temperature threshold values ​​for each key region are extracted from the retrieved structure.

[0135] The second embodiment of the present invention proposes a humanoid robot bionic skin heating adaptive power distribution system, comprising:

[0136] The data acquisition module 210 is used to acquire target data, which includes the surface temperature of each skin area of ​​the robot, environmental parameters of the environment, and the real-time and predictive motion state of the robot.

[0137] The model prediction module 220 is used to input the target data into a pre-built digital twin model to predict the thermal state change trend of each skin region within a future preset period, and to identify key areas that are preferentially heated in combination with the predictive motion state.

[0138] The model solving module 230 is used to construct and solve a multi-objective optimization model with the heating power allocated to each skin region as the decision variable. The multi-objective optimization model uses the power allocation value of each skin region as the decision variable, takes the temperature of the key region not being lower than a preset threshold as the core constraint, and takes minimizing the total energy consumption and maximizing the temperature field uniformity of the robot's overall surface as the optimization objective. Its optimization constraints are the upper limit of the total power of each skin region and the power limit of the heating unit corresponding to each skin region.

[0139] The heating control module 240 is used to determine the heating control command for each skin area based on the power distribution value obtained from the solution.

[0140] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the system described above can be found in the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0141] The following is for reference. Figure 3 It shows a schematic diagram of the structure of a computer system suitable for implementing the system and method embodiments of the present invention. Figure 3 The server shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0142] like Figure 3As shown, the computer system includes a Central Processing Unit (CPU) 301, which can perform various appropriate actions and processes based on programs stored in Read Only Memory (ROM) 302 or programs loaded from storage section 308 into Random Access Memory (RAM) 303. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An Input / Output (I / O) interface 305 is also connected to the bus 304.

[0143] The following components are connected to the input / output interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the input / output interface 305 as needed. A removable medium 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 310 as needed so that computer programs read from it can be installed into the storage section 308 as needed.

[0144] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by central processing unit 301, it performs the functions defined in the methods of the present invention. It should be noted that the computer-readable medium described above in the present invention can be a computer-readable signal medium or a computer-readable storage medium or any combination thereof. The computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof.

[0145] More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fiber, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can transmit, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0146] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including local area networks (LANs) or wide area networks (WANs), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0147] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0148] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0149] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.

[0150] The technical solution of the present invention has now been described in conjunction with the preferred embodiments shown in the accompanying drawings.

[0151] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for adaptive power distribution in the heating of bionic skin for humanoid robots, characterized in that, The method includes: Acquire target data, which includes the surface temperature of each skin region of the robot, environmental parameters of the environment, and the robot's real-time and predictive motion state; By pre-integrating temperature sensors under each skin area, real-time temperature readings of N independent areas on the robot surface are collected concurrently at a preset sampling frequency, forming an N-dimensional temperature state vector, where N is a positive integer; Collect ambient temperature, relative humidity, and triaxial air velocity data to construct an environmental parameter vector; A unified global timestamp is added to the temperature state vector and the environmental parameter vector, encapsulated into a data structure that meets a preset format, and the data structure is used as input to transmit to the computing unit running the digital twin model; Access the robot’s main motion controller and extract the predetermined task trajectory within the next T seconds. The predetermined task trajectory includes the target angle, angular velocity and angular acceleration commands in the robot’s joint space. Using the robot's forward kinematics model established based on DH parameters, the predetermined task trajectory is calculated frame by frame and converted into time series data of three-dimensional coordinates, linear velocity and linear acceleration of M control points preset on the robot's bionic skin surface in the world coordinate system. The time series data is used as the robot's predictive motion state. The target data is input into a pre-built digital twin model to predict the thermal state change trend of each skin region within a future preset period, and the key areas that are preferentially heated are identified in combination with the predictive motion state. A multi-objective optimization model is constructed and solved using the heating power allocated to each skin region as the decision variable. The multi-objective optimization model uses the power allocation value of each skin region as the decision variable, takes the temperature of the key region not being lower than a preset threshold as the core constraint, and takes minimizing the total energy consumption and maximizing the temperature field uniformity of the robot's overall surface as the optimization objective. Its optimization constraints are the upper limit of the total power of each skin region and the power limit of the heating unit corresponding to each skin region. Based on the power distribution values ​​obtained from the solution, the heating control commands for each skin region are determined.

2. The adaptive power distribution method for bionic skin heating of a humanoid robot according to claim 1, characterized in that, The process of identifying key areas for preferential heating by combining the predicted motion state includes: Based on the predicted motion state, the maximum linear velocity of each skin region control point within a future preset period is determined, and regions with maximum linear velocities exceeding a first preset threshold are marked as candidate key regions. Based on the predicted motion state, determine the skin area that will come into contact with an object with a temperature below a second preset threshold within a future preset period, and add it to the candidate key area; Perform a logical union operation on each skin region within the candidate key region to determine the key region.

3. The adaptive power distribution method for bionic skin heating of a humanoid robot according to claim 1, characterized in that, The digital twin model is a multi-layered three-dimensional transient thermodynamic finite element model. The prediction of the thermal state change trend of each skin region within a preset future period includes: The surface temperature data of each skin region is used as the initial temperature field condition of the digital twin model, and the environmental parameters are used to define the external boundary conditions of the model. The heat generated by the movement of the robot's internal joints is used as an internal heat source. By solving the unsteady heat conduction partial differential equation, forward integration is performed in the time domain to output a predicted sequence of temperature changes of each grid node over time within a preset period.

4. The adaptive power distribution method for bionic skin heating of a humanoid robot according to claim 1, characterized in that, The method for maximizing the temperature field uniformity of the robot's overall surface includes: In the objective function of the multi-objective optimization model, a non-uniformity evaluation function is defined, which is used to calculate the overall variance of the set of predicted temperature values ​​for all N skin regions at the end of the prediction cycle. Minimizing the non-uniformity evaluation function is used as the optimization objective of the multi-objective optimization model, and is processed in parallel with the objective of minimizing total energy consumption. Under the premise of satisfying the core constraints, determine the power allocation decision variables that enable the two-dimensional objective vector consisting of minimizing the non-uniformity evaluation function and minimizing total energy consumption to achieve Pareto optimality.

5. The adaptive power distribution method for bionic skin heating of a humanoid robot according to claim 1, characterized in that, The solution process of the multi-objective optimization model includes: A population is randomly initialized, where each individual is a real-number encoded chromosome containing N skin region heating power allocation values; For each individual in the population, the temperature field is predicted using a digital twin model based on its corresponding power allocation value, and two objective function values, total energy consumption and temperature field variance, are calculated. Based on the calculated objective function value, the population is subjected to non-dominated sorting and crowding calculation. Based on the sorting results and crowding, the population is iterated through a polynomial mutation genetic operator to generate offspring populations. Merge the parent and child generations, repeat the above steps until the preset number of iterations or convergence conditions are met, and output the optimal solution set of the current model.

6. The adaptive power distribution method for bionic skin heating of a humanoid robot according to claim 1, characterized in that, The step of determining the heating control command for each skin region based on the power allocation value obtained from the solution includes: From the set of optimal solutions obtained by solving the problem, one of the power allocation vectors is selected as the execution scheme for the current control cycle according to the preset preference strategy; Each determined power allocation value is converted into the duty cycle of a pulse width modulation signal based on the electrical characteristics of its corresponding heating unit. The calculated N duty cycle values ​​are then sent to the corresponding N skin area heating units.

7. The adaptive power distribution method for bionic skin heating of a humanoid robot according to claim 1, characterized in that, The process of setting the preset threshold for the key area includes: At the beginning of each control cycle, query the robot's task scheduling system for the currently executing or upcoming task identifiers. Using the queried task identifier as the key, the corresponding temperature lower limit structure is retrieved from the preset task-threshold mapping table, and the values ​​corresponding to each key area are extracted as the preset threshold. The task-threshold mapping table is a key-value pair data structure, where the key is the robot's task identifier and the value is the temperature lower limit structure corresponding to different parts of the robot.

Images