Mechanical arm self-adaptive trajectory control system and method based on neural network

Abstract

The invention provides a mechanical arm adaptive trajectory control system and method based on a neural network. The mechanical arm adaptive trajectory control method comprises the steps that mechanical arm state information of a music robot is collected in real time, and playing task information is loaded; generating a playing expected trajectory of each mechanical arm node according to the playing task information, and determining a trajectory control bias domain of each mechanical arm node according to the mechanical arm state information and the corresponding playing expected trajectory; bias approximation is conducted on the mechanical arm control parameters through a pre-trained neural network model, and track dead zone bias compensation of a mechanical arm control model is obtained; and a mechanical arm control torque vector is generated according to the mechanical arm track dead zone offset compensation and the track control offset domain of each mechanical arm node, and self-adaptive track control is conducted on the mechanical arm through the mechanical arm control torque vector. According to the technical scheme provided by the invention, an error constraint mechanism and dead zone modeling compensation can be optimized through a lightweight network structure, so that the trajectory control accuracy of the mechanical arm of the music robot is improved.

Description

Technical Field

[0001] The present application relates to the field of robotic arm control technology, and more specifically, to a neural network-based robotic arm adaptive trajectory control system and method. Background Art

[0002] In recent years, neural network-based adaptive trajectory control technology for robotic arms has demonstrated great potential in high-precision, multi-degree-of-freedom complex manipulation tasks. It is particularly suitable for applications such as musical robots, which have extremely high requirements for trajectory, timing, and power control. For example, in automatic playing systems for instruments such as pianos and guzhengs, the robotic arm must not only achieve precise keystrokes and string plucking, but also accurately control the timing, intensity, and dynamic performance of each note. Such tasks place extremely high demands on the trajectory control of the robotic arm, which traditional rigid control methods find difficult to meet. In this context, neural network-based adaptive trajectory control methods have become an important means to overcome modeling uncertainty, actuator nonlinearities (such as dead zones and friction), and complex control environments.

[0003] However, music performance is a high-frequency, rhythmically rigorous operation process, and especially during high-speed continuous performance (such as playing fast passages or rolls), the response speed of the control system is extremely demanding. In the existing technology, neural networks need to perform a large number of matrix operations when performing online learning and updating weights. Especially when the number of network nodes is large, this will significantly increase the system's computing load, resulting in real-time control lag. In addition, because the performance task involves frequently switching trajectories and changes in force, the convergence speed of the neural network under different working conditions may be inconsistent, and even overfitting or instability may occur, resulting in amplified trajectory tracking errors, which in turn affects the accuracy of the note timing and the performance expressiveness. Therefore, how to optimize the error constraint mechanism and dead zone modeling compensation through a lightweight network structure to improve the trajectory control accuracy of the music robot's manipulator arm is a difficult problem faced by the industry. Summary of the Invention

[0004] The present application provides a neural network-based adaptive trajectory control system and method for a robotic arm, which can optimize the error constraint mechanism and dead zone modeling compensation through a lightweight network structure to improve the trajectory control accuracy of the musical robot's robotic arm.

[0005] In a first aspect, the present application provides a method for adaptive trajectory control of a robotic arm based on a neural network, comprising the following steps:

[0006] Collect the status information of the music robot's mechanical arm in real time and load the music robot's performance task information;

[0007] generating an expected performance trajectory of each mechanical arm node of the music robot according to the performance task information, and determining a trajectory control bias domain of each mechanical arm node of the music robot during the performance according to the mechanical arm state information and the corresponding expected performance trajectory;

[0008] Obtaining a pre-trained neural network model, performing bias approximation on the control parameters of the musical robot's robotic arm using the neural network model, and thereby obtaining trajectory dead zone bias compensation for the robotic arm control model;

[0009] The manipulator control torque vector of the music robot is generated according to the manipulator trajectory dead zone bias compensation and the trajectory control bias domain of each manipulator node, and the manipulator trajectory is adaptively controlled by the manipulator control torque vector.

[0010] In some embodiments, the state information of the mechanical arm of the music robot is collected in real time through an intelligent sensor array.

[0011] In some embodiments, the performance task information is a structured data set used to guide the music robot's mechanical arm to complete the performance action.

[0012] In some embodiments, generating the expected performance trajectory of each mechanical arm node of the music robot according to the performance task information specifically includes:

[0013] Converting the start time and duration of each note in the performance task information into a corresponding action execution time window;

[0014] Mapping each note in the performance task information to a corresponding end space position, and then generating an end space smooth trajectory of each mechanical arm node of the music robot based on all the end space positions;

[0015] The expected performance trajectory of each robotic arm node of the music robot is generated according to all action execution time windows and the corresponding terminal spatial smooth trajectory.

[0016] In some embodiments, determining the trajectory control bias domain of each robotic arm node of the music robot during the performance according to the robotic arm state information and the corresponding expected performance trajectory specifically includes:

[0017] For each mechanical arm node of the music robot during the performance, the real-time trajectory position and real-time node speed of the mechanical arm node are extracted from the mechanical arm state information;

[0018] Determining a trajectory position bias coefficient of the robot arm node according to the real-time trajectory position and the corresponding expected performance trajectory;

[0019] Designing a virtual controller based on an obstacle Lyapunov function, and obtaining a virtual trajectory velocity of the robotic arm node according to the virtual controller and the corresponding expected performance trajectory;

[0020] Obtaining a trajectory velocity bias coefficient of the robot arm node from the real-time node velocity and the virtual trajectory velocity;

[0021] The trajectory control bias domain of the robotic arm node is constructed according to the trajectory position bias coefficient and the trajectory speed bias coefficient, thereby obtaining the trajectory control bias domain of each robotic arm node of the music robot during the performance.

[0022] In some embodiments, performing bias approximation on the control parameters of the musical robot's robotic arm through the neural network model to obtain trajectory dead zone bias compensation of the robotic arm control model specifically includes:

[0023] The state information of the music robot's robotic arm and the control parameters of the robotic arm form a network input vector;

[0024] Inputting the network input vector into the neural network model, the neural network model automatically selects the center node of the receptive field based on the Gaussian kernel function, and continuously adjusts the network output in the control loop through the weight update law, thereby dynamically learning and approximating the nonlinear deviation term caused by the dead zone in the robot arm control model;

[0025] The compensation term output by the neural network model is used as trajectory dead zone bias compensation of the robotic arm control model.

[0026] In some embodiments, performing adaptive trajectory control on the robotic arm by controlling the robotic arm torque vector specifically includes:

[0027] The robotic arm control torque vector is applied in real time to the joint actuators corresponding to each robotic arm node of the music robot. The joint actuator converts the corresponding trajectory control torque into the corresponding joint driving current, and sends it to the robotic arm joint through the interface, so that the robotic arm can complete the angle and position adjustment, thereby realizing adaptive trajectory control of the musical robot's robotic arm.

[0028] In a second aspect, the present application provides a neural network-based adaptive trajectory control system for a robotic arm, which is used to execute a neural network-based adaptive trajectory control method for a robotic arm, comprising:

[0029] The information collection module is used to collect the status information of the music robot's mechanical arm in real time and load the music robot's performance task information;

[0030] a bias determination module, configured to generate an expected performance trajectory for each mechanical arm node of the music robot according to the performance task information, and determine a trajectory control bias domain for each mechanical arm node of the music robot during the performance according to the mechanical arm state information and the corresponding expected performance trajectory;

[0031] A bias compensation module is used to obtain a pre-trained neural network model, and to perform bias approximation on the control parameters of the musical robot's manipulator arm through the neural network model, thereby obtaining trajectory dead zone bias compensation of the manipulator arm control model;

[0032] The trajectory control module is used to generate the manipulator control torque vector of the music robot according to the manipulator trajectory dead zone offset compensation and the trajectory control offset domain of each manipulator node, and perform adaptive trajectory control on the manipulator through the manipulator control torque vector.

[0033] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a code, and the processor is configured to obtain the code and execute the above-mentioned neural network-based robotic arm adaptive trajectory control method.

[0034] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned neural network-based robotic arm adaptive trajectory control method.

[0035] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0036] In the neural network-based adaptive trajectory control system and method for a robotic arm provided in the present application, the robotic arm state information of the music robot is collected in real time, and the performance task information of the music robot is loaded; the expected performance trajectory of each robotic arm node of the music robot is generated according to the performance task information, and the trajectory control bias domain of each robotic arm node of the music robot during the performance is determined according to the robotic arm state information and the corresponding expected performance trajectory; a pre-trained neural network model is obtained, and the robotic arm control parameters of the music robot are biased and approximated by the neural network model, thereby obtaining the trajectory dead zone bias compensation of the robotic arm control model; the robotic arm control torque vector of the music robot is generated according to the robotic arm trajectory dead zone bias compensation and the trajectory control bias domain of each robotic arm node, and the robotic arm is adaptively trajectory controlled by the robotic arm control torque vector.

[0037] It can be seen that in this application, first, the expected performance trajectory of each robotic arm node of the music robot is generated according to the performance task information, and the trajectory control bias domain is dynamically constructed in combination with the actual robotic arm state information, which not only realizes the real-time quantification of the trajectory execution error, but also provides a clear and adjustable error boundary constraint for the subsequent controller, which helps to improve the system's response sensitivity to dynamic errors and enhance its adaptability to complex rhythms and fine movements; then, by obtaining a pre-trained neural network model, the parameters of the musical robot's robotic arm are biased and approximated, and trajectory dead zone bias compensation is implemented, which can significantly improve the control system's adaptability to nonlinear uncertainties, especially in complex performance scenarios with input dead zones and dynamic modeling errors. It is more critical to utilize a lightweight network structure to dynamically approximate the nonlinear deviation terms in the system while ensuring low computational overhead, thereby achieving efficient real-time error compensation; finally, by combining the robotic arm trajectory dead zone bias compensation with the trajectory control bias domain of each robotic arm node to generate a control torque vector, and implementing adaptive trajectory control accordingly, it is possible to achieve high-precision dynamic control of the music robot in complex performance scenarios.

[0038] To sum up, the technical solution adopted in this application can optimize the error constraint mechanism and dead zone modeling compensation through a lightweight network structure to improve the trajectory control accuracy of the music robot's manipulator arm. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0040] Figure 1 is an exemplary flow chart of a method for adaptive trajectory control of a robotic arm based on a neural network according to some embodiments of the present application;

[0041] Figure 2 is an exemplary flow chart for generating the expected performance trajectory of each mechanical arm node of a music robot according to some embodiments of the present application;

[0042] Figure 3 is a schematic structural diagram of a neural network-based adaptive trajectory control system for a robotic arm according to some embodiments of the present application;

[0043] Figure 4 It is a structural diagram of a computer device for implementing a neural network-based adaptive trajectory control method for a robotic arm according to some embodiments of the present application. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0045] The embodiment of the present application provides a neural network-based adaptive trajectory control system and method for a manipulator, the core of which is to collect the manipulator state information of a music robot in real time and load the performance task information of the music robot; generate the expected performance trajectory of each manipulator node of the music robot based on the performance task information, and determine the trajectory control bias domain of each manipulator node of the music robot during the performance according to the manipulator state information and the corresponding expected performance trajectory; obtain a pre-trained neural network model, and use the neural network model to perform bias approximation on the manipulator control parameters of the music robot, thereby obtaining the trajectory dead zone bias compensation of the manipulator control model; generate the manipulator control torque vector of the music robot based on the manipulator trajectory dead zone bias compensation and the trajectory control bias domain of each manipulator node, and use the manipulator control torque vector to perform adaptive trajectory control on the manipulator. The above scheme can be used to optimize the error constraint mechanism and dead zone modeling compensation through a lightweight network structure to improve the trajectory control accuracy of the manipulator arm of the music robot.

[0046] In order to better understand the above technical solution, the following will be described in detail with reference to the accompanying drawings and specific implementation methods. Figure 1 , which is an exemplary flow chart of a method for adaptive trajectory control of a robotic arm based on a neural network according to some embodiments of the present application, and mainly includes the following steps:

[0047] In step S101 , the state information of the mechanical arm of the music robot is collected in real time, and the performance task information of the music robot is loaded.

[0048] In some embodiments, the state information of the musical robot's robotic arm is collected in real time through an intelligent sensor array. In specific implementation, the intelligent sensor array integrated on each joint or end effector of the robotic arm can be used to perform millisecond-level sampling and transmission of the motion state of the robotic arm, thereby obtaining the state information of the musical robot's robotic arm, wherein the robotic arm state information includes the real-time trajectory position, real-time node speed, and node torque of each robotic arm node, and each joint on the robotic arm can be used as a robotic arm node.

[0049] It should be noted that in the present application, the performance task information is a structured data set used to guide the music robot's mechanical arm to complete the performance action; in specific implementation, the task scheduler can read the specified performance segment from the local trajectory database or external interface, thereby parsing the performance segment to obtain the performance task information, such as the performance start time and duration of each note, the performance intensity level and performance type identification, etc.

[0050] In step S102, the expected performance trajectory of each mechanical arm node of the music robot is generated according to the performance task information, and the trajectory control bias domain of each mechanical arm node of the music robot during the performance is determined according to the mechanical arm state information and the corresponding expected performance trajectory.

[0051] In some embodiments, reference Figure 2 As shown in FIG. 1 , this figure is an exemplary flow chart of generating the expected performance trajectory of each mechanical arm node of the music robot according to some embodiments of the present application. In this embodiment, generating the expected performance trajectory of each mechanical arm node of the music robot according to the performance task information can be achieved by using the following steps:

[0052] In step S1021, the start time and duration of each note in the performance task information are converted into a corresponding action execution time window;

[0053] In step S1022, each note in the performance task information is mapped to a corresponding end space position, and then a smooth end space trajectory of each mechanical arm node of the music robot is generated based on all the end space positions;

[0054] In step S1023 , the expected performance trajectory of each mechanical arm node of the music robot is generated according to all action execution time windows and the corresponding terminal spatial smooth trajectory.

[0055] In specific implementation, first, the start time and duration of each note in the performance task information can be converted into a corresponding action execution time window, that is, the start time of the note is used as the lower bound of the action execution time window, and the sum of the start time and duration of the note is used as the upper bound of the action execution time window. The action execution time window corresponding to each note can be obtained in the above manner; then, each note in the performance task information can be mapped to a corresponding end space position, that is, a note space mapping table is established according to the instrument layout (such as the position of the keys or strings), so that each note is mapped to a target point in three-dimensional space, that is, the end space position corresponding to each note mapping, and the mapping rule is based on the static transformation matrix between the instrument geometry, the key / string number and the spatial coordinates (such as a simulation or calibration table); secondly, the end space smooth trajectory of each robotic arm node of the music robot can be generated based on all the end space positions, that is, the spatial position sequence of each robotic arm node of the music robot when the robotic arm end is at each end space position can be obtained, and spline interpolation (cubic spline) or minimum acceleration trajectory to construct the terminal space smooth trajectory of each robotic arm node; finally, the expected performance trajectory of each robotic arm node of the music robot can be generated according to all action execution time windows and the corresponding terminal space smooth trajectory, that is, for each robotic arm node of the music robot, a timestamp can be added to the terminal space smooth trajectory of the robotic arm node through all action execution time windows, and the terminal space smooth trajectory can be converted into the angular trajectory of the robotic arm node, that is, the expected performance trajectory, through the above method.

[0056] In some embodiments, the trajectory control bias domain of each robotic arm node of the music robot during the performance is determined according to the robotic arm state information and the corresponding expected performance trajectory, specifically in the following manner, namely:

[0057] For each mechanical arm node of the music robot during the performance, the real-time trajectory position and real-time node speed of the mechanical arm node are extracted from the mechanical arm state information;

[0058] Determining a trajectory position bias coefficient of the robot arm node according to the real-time trajectory position and the corresponding expected performance trajectory;

[0059] Designing a virtual controller based on an obstacle Lyapunov function, and obtaining a virtual trajectory velocity of the robotic arm node according to the virtual controller and the corresponding expected performance trajectory;

[0060] Obtaining a trajectory velocity bias coefficient of the robot arm node from the real-time node velocity and the virtual trajectory velocity;

[0061] The trajectory control bias domain of the robotic arm node is constructed according to the trajectory position bias coefficient and the trajectory speed bias coefficient, thereby obtaining the trajectory control bias domain of each robotic arm node of the music robot during the performance.

[0062] In the specific implementation, first, for each robotic arm node of the music robot during the performance, the real-time trajectory position and real-time node speed of the robotic arm node can be extracted from the robotic arm state information; then, the trajectory position bias coefficient of the robotic arm node can be determined through the real-time trajectory position and the corresponding expected performance trajectory, wherein the trajectory position bias coefficient represents the overall difference deviation between the real-time motion trajectory of the robotic arm node and the expected performance trajectory during the performance, and the Euclidean distance between the real-time trajectory position corresponding to each note and the corresponding trajectory position in the expected performance trajectory can be calculated, and the standard deviation of all Euclidean distances is used as the trajectory position bias coefficient of the robotic arm node; secondly, a virtual controller can be designed based on the obstacle Lyapunov function, so that the virtual trajectory speed of the robotic arm node can be obtained according to the virtual controller and the corresponding expected performance trajectory. When establishing a stable and adjustable control framework, a virtual controller designed based on the obstacle Lyapunov function is introduced. The role of the virtual controller is to dynamically construct an ideal virtual speed trajectory according to the current error state and the expected performance index, so as to guide the actual speed to gradually approach the expected performance trajectory and ensure the trajectory error. The difference will not exceed the set boundary. The virtual trajectory speed not only takes into account the current expected performance trajectory, but also introduces the error convergence speed and adjustable performance parameters, so that the system stability and responsiveness can be maintained even in scenarios with rapid changes or continuous performance. Furthermore, the trajectory speed bias coefficient of the robot node can be obtained from the real-time node speed and the virtual trajectory speed. The trajectory speed bias coefficient represents the overall deviation between the real-time node speed and the virtual trajectory speed of the robot node during the performance. The Euclidean distance between the real-time node speed corresponding to each note and the corresponding point in the virtual trajectory speed can be calculated, and the standard deviation of all Euclidean distances is used as the trajectory speed bias coefficient of the robot node. Finally, the trajectory control bias domain of the robot node can be constructed based on the trajectory position bias coefficient and the trajectory speed bias coefficient, where the trajectory control bias domain represents the overall deviation of the running trajectory under the current robot control model. The data composed of the trajectory position bias coefficient and the trajectory speed bias coefficient can be combined as the trajectory control bias domain of the robot node. The trajectory control bias domain of each robot node of the music robot during the performance can be obtained in the above manner.

[0063] It should be noted that generating the expected performance trajectory of each robotic arm node of the music robot based on the performance task information and dynamically constructing the trajectory control bias domain in combination with the actual robotic arm state information can not only achieve real-time quantification of trajectory execution errors, but also provide clear and adjustable error boundary constraints for subsequent controllers, which helps to improve the system's response sensitivity to dynamic errors and enhance its adaptability to complex rhythms and fine movements (such as piano fast keys and guzheng glissando).

[0064] In step S103, a pre-trained neural network model is obtained, and bias approximation is performed on the control parameters of the musical robot's manipulator arm through the neural network model, thereby obtaining trajectory dead zone bias compensation of the manipulator arm control model.

[0065] In some embodiments, the pre-trained neural network model may be obtained in the following manner:

[0066] The node structure of the neural network model is optimized through width learning, and the weight of the neural network model is updated using the adaptive update law to obtain a pre-trained neural network model.

[0067] In the specific implementation, first of all, when facing complex dynamic systems, traditional neural networks (such as standard radial basis neural networks) often have low training efficiency and weak generalization ability due to redundant nodes or rigid structure. In the application, the node structure of the neural network model can be optimized through width learning. The width learning mechanism is introduced to dynamically optimize the node structure of the neural network. That is, in the learning process, according to the distribution of input data, it is judged whether a new node needs to be added. Only when the distance between the new input sample and the center of the existing node exceeds a certain threshold and the current number of nodes does not exceed the preset upper limit, a new node will be created, thereby avoiding structural bloat. In addition, in each forward calculation, the network only activates the few nodes closest to the current input (such as the previous m ) nodes participate in the calculation, significantly reducing the amount of computation and improving real-time performance. For activated nodes, their receptive field center and width parameters are adjusted according to the new input samples, so that the network structure maintains a high degree of coupling with the input features throughout the training process, thereby enhancing its ability to express the robot arm's state. Then, the weights of the neural network model can be updated using an adaptive update law. That is, based on the optimization of the network structure, an adaptive update law designed based on Lyapunov stability theory is further used to achieve online update and stable convergence of the neural network weights. During operation, the controller continuously collects the error between the actual state (position, velocity) and the desired state of the robot arm nodes. This error information is combined with the activation characteristics of the network as input to drive the weight adjustment equation, adjusting the connection weights of the neural network in real time. The update law design ensures error-guided weight learning. The Lyapunov function construction ensures that the update process does not cause system instability, with global convergence guaranteed. Weight updates can continue until the network output stably approximates the trajectory dead zone bias term and the model uncertainty, thus forming an accurate and highly generalized neural network model, that is, the pre-trained neural network model.

[0068] In some embodiments, the neural network model is used to bias approximate the control parameters of the musical robot's robotic arm, thereby obtaining the trajectory dead zone bias compensation of the robotic arm control model, specifically in the following manner, namely:

[0069] The state information of the music robot's robotic arm and the control parameters of the robotic arm form a network input vector;

[0070] Inputting the network input vector into the neural network model, the neural network model automatically selects the center node of the receptive field based on the Gaussian kernel function, and continuously adjusts the network output in the control loop through the weight update law, thereby dynamically learning and approximating the nonlinear deviation term caused by the dead zone in the robot arm control model;

[0071] The compensation term output by the neural network model is used as trajectory dead zone bias compensation of the robotic arm control model.

[0072] In the specific implementation, first, the current robotic arm state information of the music robot and the known robotic arm control parameters are combined to form the input vector of the neural network. The control parameters of the robotic arm include auxiliary information such as inertia matrix estimation, joint friction coefficient, historical control torque, etc. The robotic arm state information and robotic arm control parameters are fused to form a high-dimensional feature vector as the input of the neural network to fully reflect the current control state and dynamic change characteristics; then, the network input vector can be input into the neural network model. The neural network model is based on the Gaussian kernel function and automatically selects the center node of the receptive field. That is, the neural network dynamically selects the closest m center nodes for activation and calculation according to the distance between the current input sample and the center of the existing node, thereby improving the operation efficiency. rate and avoid overfitting, and continuously adjust the network output in the control loop through the weight update law, that is, the weight update law driven by the control error continuously adjusts the network connection weights, so that the neural network output gradually approaches the trajectory deviation term caused by nonlinearities such as dead zones, thereby dynamically learning and approximating the nonlinear deviation term caused by dead zones in the manipulator control model; finally, the compensation term output by the neural network model can be used as the trajectory dead zone bias compensation of the manipulator control model, where the trajectory dead zone bias compensation represents the deviation caused by the dead zone effect in the current manipulator trajectory control. If the trajectory dead zone bias compensation is applied to the actual control torque command generation module, the dead zone error can be dynamically corrected, so that the actual output torque is more in line with the ideal trajectory requirement.

[0073] It should be noted that by using a pre-trained neural network model to bias approximate the parameters of the musical robot's manipulator and achieve trajectory dead zone bias compensation, the control system's adaptability to nonlinear uncertainties can be significantly improved. This is particularly critical in complex performance scenarios with input dead zones and dynamic modeling errors. By utilizing a lightweight network structure, the nonlinear bias terms in the system are dynamically approximated while ensuring low computational overhead, enabling efficient real-time error compensation. Compared to traditional rigid control strategies, this adaptive compensation mechanism not only enhances the flexibility of trajectory error constraints but also effectively addresses performance tasks such as rapid piano tapping and guzheng glissando, which require extremely high response speed and trajectory accuracy. This ensures that keystrokes are not delayed or misaligned, and that playing dynamics and rhythm are precisely consistent, thereby improving the fidelity and expressiveness of the musical robot's performance.

[0074] In step S104, a manipulator control torque vector of the music robot is generated according to the manipulator trajectory dead zone bias compensation and the trajectory control bias domain of each manipulator node, and the manipulator trajectory is adaptively controlled by the manipulator control torque vector.

[0075] In some embodiments, the following method can be used to generate the manipulator control torque vector of the music robot based on the manipulator trajectory dead zone offset and the trajectory control offset domain of each manipulator node, namely:

[0076] Design of robotic arm controller based on Lyapunov stability theory;

[0077] Inputting the robot arm trajectory dead zone offset and the trajectory control offset domain of each robot arm node into the robot arm controller, thereby obtaining the trajectory control torque of each robot arm node;

[0078] The manipulator control torque vector of the music robot is generated through the trajectory control torque of all manipulator nodes.

[0079] In the specific implementation, first, the manipulator controller can be designed based on the Lyapunov stability theory, that is, a manipulator controller framework based on the Lyapunov stability theory is constructed. The manipulator controller not only takes into account the feedback adjustment mechanism of the conventional trajectory tracking error, but also integrates the modeling uncertainty and dead zone nonlinear effects in the model. The design goal of the manipulator controller is to ensure that all closed-loop error systems are asymptotically stable under the action of the control input, that is, all position and velocity errors can gradually converge within the bounded performance function, while avoiding overshoot or jitter, to ensure the stability and smoothness of the performance trajectory; then, the manipulator trajectory dead zone bias and the trajectory control bias domain of each manipulator node can be input into the manipulator controller, that is, in the control execution stage, the trajectory dead zone bias predicted by the neural network model and the trajectory control bias domain calculated in real time by each manipulator node are input as key parameters into the designed manipulator controller. In the device, the manipulator controller will comprehensively consider these inputs, that is, use the trajectory dead zone bias information for nonlinear compensation, use the trajectory control bias domain information to perform boundary constraints and corrections on the current joint errors, and dynamically adjust the control gain to achieve personalized trajectory repair for each manipulator node, thereby outputting the control torque of each manipulator node, that is, obtaining the trajectory control torque of each manipulator node. In this process, the manipulator controller uses the obstacle Lyapunov function to construct the constraint domain to ensure that all biases are adjusted within a safe range to prevent action out-of-limit or keystroke offset, which is particularly suitable for playing actions such as guzheng glissando and piano fast keys that require extremely high trajectory continuity and triggering accuracy; finally, the manipulator control torque vector of the music robot can be generated through the trajectory control torque of all manipulator nodes, that is, the characteristic vector composed of the trajectory control torque of all manipulator nodes in the order of nodes is used as the manipulator control torque vector of the music robot.

[0080] In some embodiments, the adaptive trajectory control of the manipulator by the manipulator control torque vector may be performed in the following manner, namely:

[0081] The robotic arm control torque vector is applied in real time to the joint actuators corresponding to each robotic arm node of the music robot. The joint actuator converts the corresponding trajectory control torque into the corresponding joint driving current, and sends it to the robotic arm joint through the interface, so that the robotic arm can complete the angle and position adjustment, thereby realizing adaptive trajectory control of the musical robot's robotic arm.

[0082] In specific implementation, first, the manipulator control torque vector can be applied in real time to the joint actuators corresponding to each manipulator node of the music robot, that is, the trajectory control torque corresponding to each dimension in the manipulator control torque vector is applied in real time to the joint actuators corresponding to each manipulator node of the music robot. Each trajectory control torque is bound to a specific joint motion axis to drive it to perform precise rotation or linear displacement movements; then, the joint actuator converts the received trajectory control torque signal into the corresponding drive current instruction, and through the torque-current mapping model of the joint actuator (such as the torque constant inside the servo motor or direct drive motor), an efficient conversion from control quantity to physical driving force is achieved; secondly, the converted drive current is transmitted through a high-speed real-time communication interface (such as USB, Ethernet) The interface supports low-latency, high-bandwidth data interaction, ensuring that each trajectory control instruction is issued and executed within milliseconds, meeting the response speed requirements of complex playing movements. Finally, after receiving the control current, the joint control module drives the motor or servo mechanism to complete the corresponding angle and position adjustment, so as to accurately track the desired trajectory under the current control cycle. The adjustment process is continuous and driven by closed-loop feedback, which can correct the trajectory error caused by dead zone deviation, dynamic interference or structural friction in real time, ensuring the accuracy and consistency of each keystroke, glissando, finger lift and other actions. Through the above steps, the robotic arm completes a response control of the target trajectory, thereby realizing the adaptive trajectory control of the musical robot arm.

[0083] It should be noted that by combining dead-zone bias compensation for the robot trajectory with the trajectory control bias domain of each robot node to generate a control torque vector and implementing adaptive trajectory control accordingly, high-precision dynamic control of the musical robot in complex performance scenarios can be achieved. This significantly enhances the model's ability to identify and compensate for nonlinear dead-zone effects, model uncertainty, and real-time errors. This makes it particularly suitable for performance tasks such as rapid piano legato and guzheng glissando, which require extremely high motion responsiveness and trajectory accuracy. By introducing a lightweight neural network structure, the system achieves online learning and rapid correction of dead-zone modeling and error constraints while maintaining controllable computational overhead. This effectively reduces trajectory drift and motion delay caused by model lag or input saturation, ensuring that each keystroke of the robot is accurately positioned, rhythmically coherent, and with consistent force, enhancing the naturalness and musical expressiveness of the performance. This is a key control strategy for achieving high-fidelity anthropomorphic performance.

[0084] It can be seen that in this application, first, the expected performance trajectory of each robotic arm node of the music robot is generated according to the performance task information, and the trajectory control bias domain is dynamically constructed in combination with the actual robotic arm state information, which not only realizes the real-time quantification of the trajectory execution error, but also provides a clear and adjustable error boundary constraint for the subsequent controller, which helps to improve the system's response sensitivity to dynamic errors and enhance its adaptability to complex rhythms and fine movements; then, by obtaining a pre-trained neural network model, the parameters of the musical robot's robotic arm are biased and approximated, and trajectory dead zone bias compensation is implemented, which can significantly improve the control system's adaptability to nonlinear uncertainties, especially in complex performance scenarios with input dead zones and dynamic modeling errors. It is more critical to utilize a lightweight network structure to dynamically approximate the nonlinear deviation terms in the system while ensuring low computational overhead, thereby achieving efficient real-time error compensation; finally, by combining the robotic arm trajectory dead zone bias compensation with the trajectory control bias domain of each robotic arm node to generate a control torque vector, and implementing adaptive trajectory control accordingly, it is possible to achieve high-precision dynamic control of the music robot in complex performance scenarios.

[0085] To sum up, the technical solution adopted in this application can optimize the error constraint mechanism and dead zone modeling compensation through a lightweight network structure to improve the trajectory control accuracy of the music robot's manipulator arm.

[0086] In addition, in another aspect of the present application, in some embodiments, the present application provides a neural network-based adaptive trajectory control system for a robotic arm, referring to Figure 3 , which is a schematic diagram of the structure of a neural network-based adaptive trajectory control system for a manipulator according to some embodiments of the present application. The neural network-based adaptive trajectory control system for a manipulator includes: an information acquisition module 201, a bias determination module 202, a bias compensation module 203, and a trajectory control module 204, which are described as follows:

[0087] The information collection module 201 is used to collect the status information of the mechanical arm of the music robot in real time and load the performance task information of the music robot;

[0088] The bias determination module 202 is configured to generate an expected performance trajectory for each mechanical arm node of the music robot according to the performance task information, and determine a trajectory control bias domain for each mechanical arm node of the music robot during the performance according to the mechanical arm state information and the corresponding expected performance trajectory;

[0089] The bias compensation module 203 is used to obtain a pre-trained neural network model, and perform bias approximation on the mechanical arm control parameters of the music robot through the neural network model, thereby obtaining trajectory dead zone bias compensation of the mechanical arm control model;

[0090] The trajectory control module 204 is used to generate the manipulator control torque vector of the music robot according to the manipulator trajectory dead zone offset compensation and the trajectory control offset domain of each manipulator node, and perform adaptive trajectory control on the manipulator through the manipulator control torque vector.

[0091] In addition, the present application also provides a computer device, which includes a memory and a processor, the memory storing a code, and the processor being configured to obtain the code and execute the above-mentioned neural network-based robotic arm adaptive trajectory control method.

[0092] In some embodiments, reference Figure 4 , which is a schematic diagram of the structure of a computer device for implementing a neural network-based adaptive trajectory control method for a manipulator according to some embodiments of the present application. The neural network-based adaptive trajectory control method for a manipulator in the above embodiment can be achieved by Figure 4 The computer device shown in FIG3 is implemented as shown in FIG3 , which includes at least one processor 301 , a communication bus 302 , a memory 303 and at least one communication interface 304 .

[0093] The processor 301 can be a general-purpose central processing unit (CPU), or an application-specific integrated circuit (ASIC) or one or more components for controlling the execution of the neural network-based robotic arm adaptive trajectory control method in the present application.

[0094] The communication bus 302 may be used to transmit information between the aforementioned components.

[0095] Memory 303 may be, but is not limited to, a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer. Memory 303 may be independent and connected to processor 301 via communication bus 302. Memory 303 may also be integrated with processor 301.

[0096] The memory 303 is used to store program code for executing the solution of the present application, and is controlled by the processor 301. The processor 301 is used to execute the program code stored in the memory 303. The program code may include one or more software modules. The determination of the neural network-based robotic arm adaptive trajectory control method in the above embodiment can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.

[0097] The communication interface 304 uses any device such as a transceiver to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), etc.

[0098] In a specific implementation, as an example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0099] The aforementioned computer device can be a general-purpose computer device or a dedicated computer device. In a specific implementation, the computer device can be a desktop computer, a portable computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. The embodiments of this application do not limit the type of computer device.

[0100] In addition, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned neural network-based robotic arm adaptive trajectory control method.

[0101] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0102] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for adaptive trajectory control of a robotic arm based on a neural network, characterized in that: The steps include: Collect the status information of the music robot's mechanical arm in real time and load the music robot's performance task information; generating an expected performance trajectory of each mechanical arm node of the music robot according to the performance task information, and determining a trajectory control bias domain of each mechanical arm node of the music robot during the performance according to the mechanical arm state information and the corresponding expected performance trajectory; Obtaining a pre-trained neural network model, performing bias approximation on the control parameters of the musical robot's robotic arm using the neural network model, and thereby obtaining trajectory dead zone bias compensation for the robotic arm control model; The manipulator control torque vector of the music robot is generated according to the manipulator trajectory dead zone bias compensation and the trajectory control bias domain of each manipulator node, and the manipulator trajectory is adaptively controlled by the manipulator control torque vector.

2. The method for adaptive trajectory control of a robotic arm based on a neural network according to claim 1, wherein: The status information of the music robot's robotic arm is collected in real time through an intelligent sensor array.

3. The method for adaptive trajectory control of a robotic arm based on a neural network according to claim 1, wherein: The performance task information is a structured data set used to guide the music robot's mechanical arm to complete the performance action.

4. The method for adaptive trajectory control of a robotic arm based on a neural network according to claim 1, wherein: Generating the expected performance trajectory of each mechanical arm node of the music robot according to the performance task information specifically includes: Converting the start time and duration of each note in the performance task information into a corresponding action execution time window; Mapping each note in the performance task information to a corresponding end space position, and then generating an end space smooth trajectory of each mechanical arm node of the music robot based on all the end space positions; The expected performance trajectory of each robotic arm node of the music robot is generated according to all action execution time windows and the corresponding terminal spatial smooth trajectory.

5. The method for adaptive trajectory control of a robotic arm based on a neural network according to claim 1, wherein: Determining the trajectory control bias domain of each mechanical arm node of the music robot during the performance according to the mechanical arm state information and the corresponding expected performance trajectory specifically includes: For each mechanical arm node of the music robot during the performance, the real-time trajectory position and real-time node speed of the mechanical arm node are extracted from the mechanical arm state information; Determining a trajectory position bias coefficient of the robot arm node according to the real-time trajectory position and the corresponding expected performance trajectory; Designing a virtual controller based on an obstacle Lyapunov function, and obtaining a virtual trajectory velocity of the robotic arm node according to the virtual controller and the corresponding expected performance trajectory; Obtaining a trajectory velocity bias coefficient of the robot arm node from the real-time node velocity and the virtual trajectory velocity; The trajectory control bias domain of the robotic arm node is constructed according to the trajectory position bias coefficient and the trajectory speed bias coefficient, thereby obtaining the trajectory control bias domain of each robotic arm node of the music robot during the performance.

6. The method for adaptive trajectory control of a robotic arm based on a neural network according to claim 1, wherein: The neural network model is used to bias the control parameters of the music robot's manipulator arm, thereby obtaining the trajectory dead zone bias compensation of the manipulator arm control model. Specifically, the method includes: The state information of the music robot's robotic arm and the control parameters of the robotic arm form a network input vector; Inputting the network input vector into the neural network model, the neural network model automatically selects the center node of the receptive field based on the Gaussian kernel function, and continuously adjusts the network output in the control loop through the weight update law, thereby dynamically learning and approximating the nonlinear deviation term caused by the dead zone in the robot arm control model; The compensation term output by the neural network model is used as trajectory dead zone bias compensation of the robotic arm control model.

7. The method for adaptive trajectory control of a robotic arm based on a neural network according to claim 1, wherein: The adaptive trajectory control of the manipulator by the manipulator control torque vector specifically includes: The robotic arm control torque vector is applied in real time to the joint actuators corresponding to each robotic arm node of the music robot. The joint actuator converts the corresponding trajectory control torque into the corresponding joint driving current, and sends it to the robotic arm joint through the interface, so that the robotic arm can complete the angle and position adjustment, thereby realizing adaptive trajectory control of the musical robot's robotic arm.

8. A neural network-based adaptive trajectory control system for a robotic arm, used to execute the neural network-based adaptive trajectory control method for a robotic arm according to any one of claims 1 to 7, characterized in that: include: The information collection module is used to collect the status information of the music robot's mechanical arm in real time and load the music robot's performance task information; a bias determination module, configured to generate an expected performance trajectory for each mechanical arm node of the music robot according to the performance task information, and determine a trajectory control bias domain for each mechanical arm node of the music robot during the performance according to the mechanical arm state information and the corresponding expected performance trajectory; A bias compensation module is used to obtain a pre-trained neural network model, and to perform bias approximation on the control parameters of the musical robot's manipulator arm through the neural network model, thereby obtaining trajectory dead zone bias compensation of the manipulator arm control model; The trajectory control module is used to generate the manipulator control torque vector of the music robot according to the manipulator trajectory dead zone offset compensation and the trajectory control offset domain of each manipulator node, and perform adaptive trajectory control on the manipulator through the manipulator control torque vector.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a code, and the processor is configured to obtain the code and execute the neural network-based robotic arm adaptive trajectory control method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the neural network-based adaptive trajectory control method for a robotic arm is implemented.

Images