Master-slave operating arm with electromagnetic force feedback for precision assembly and bidirectional transparent control method

Abstract

The application belongs to the technical field of robot control, and particularly relates to a master-slave operating arm with electromagnetic force feedback for precision assembly and a bidirectional transparent control method, which comprises a master-slave robot unit, an information transmission unit, a master-slave control unit, an electromagnetic force feedback unit, a bidirectional transparent control unit, a singular point processing unit, an adaptive path planning unit and a master-slave proportion matching unit. The electromagnetic force feedback unit obtains environmental force information through a magnetic coupling component and transmits the information to the master control unit; the bidirectional transparent control unit constructs a master-slave control closed loop, balances the system transparency and stability under communication delay, detects and avoids joint singular points through a Jacobian matrix, and constructs a potential field function to generate an optimal motion path. The application realizes high-precision control and force feedback under a communication delay environment, and improves the accuracy and safety of remote operation.

Description

Technical Field

[0001] This invention belongs to the field of robot control, specifically to a master-slave manipulator with electromagnetic force feedback for precision assembly and a two-way transparent control method. Background Technology

[0002] With the rapid development of robotics technology, master-slave manipulator systems are increasingly widely used in remote operation, hazardous environment work, and precision surgery. These systems precisely transmit the operator's movements to a remote actuator and send force feedback information from the remote environment back to the operator, enabling remote, precise human-machine collaborative operation. In these systems, force feedback technology is a key factor in improving operational accuracy and user experience.

[0003] Traditional master-slave operating systems primarily employ a position-position control architecture, where the master controller sends position commands, and the slave controller receives and executes the corresponding actions. However, this unidirectional control method cannot provide feedback on environmental force information to the operator, leading to increased blind operation and limited accuracy. To address this issue, researchers have proposed a bidirectional control architecture. By installing force / torque sensors on the slave end, environmental contact force information is transmitted back to the master end, achieving force feedback functionality. For example, CN117840968A discloses a bidirectional control system and method for robot teleoperation based on master-slave force feedback. This system receives force feedback information from the slave end control unit through the master end control unit, calculates the force control information required for torque reproduction, and reproduces the feedback torque of each joint on the slave end at the master end.

[0004] In the area of ​​master-slave mapping control, CN104440864B proposes a master-slave teleoperated industrial robot system and its control method. This method analyzes the geometric configuration of the master and slave hands, designs a master-slave motion mapping strategy, adopts a master-slave control algorithm based on the inverse Jacobian matrix, and eliminates master-slave position following errors through proportional-derivative control. Furthermore, CN107717994B discloses a general control method and system for master-slave heterogeneous robots based on master-slave spatial mapping. This system maps and transforms the position, velocity, and force of the master and slave hands according to the link parameters of the master and slave hands using the DH parameter method and the Jacobian matrix method, achieving precise control of the slave hand by the master hand.

[0005] To improve operational efficiency and reduce operational burden, WO2025179628A1 proposes a multimodal shared teleoperation system and method for a three-armed space robot. This system integrates multimodal teleoperation control methods, including pose control, voice control, and force control, with the robot's autonomous control through a shared control algorithm, enabling human-machine collaboration in controlling the position, posture, and contact force of the robotic arm. Regarding force feedback, CN108284425A introduces a collaborative force feedback master-slave control method and system for a live-line working robot. This system uses a six-degree-of-freedom force / torque sensor installed at the end of the robotic arm to calculate the torque data of each joint of the master operator based on the force data at the end of the robotic arm, thus achieving force feedback control.

[0006] The main drawbacks of the relevant technologies are as follows: 1. Traditional master-slave operating systems struggle to achieve high transparency and stability in environments with communication latency. Even a minimum latency of 100ms can significantly impact the accuracy of feedback and the user experience. Existing technologies typically employ a four-channel controller or a two-channel control architecture, combined with damped injection and force reflection filters. However, these methods often require performance trade-offs, making it difficult to simultaneously ensure system transparency and stability.

[0007] 2. Existing force feedback systems are prone to insufficient or distorted force feedback due to factors such as sensor noise, bandwidth limitations, and actuator dynamic characteristics. Furthermore, the size difference between the master and slave systems makes it difficult to precisely match the position and force response scaling factors, further reducing the accuracy of force feedback. Although the reaction observer (RFOB) technique can achieve force feedback without sensors, its accuracy is often lower than that of direct measurement methods.

[0008] 3. Robotic arms are prone to singularities in certain configurations, leading to unstable control or abnormal joint speeds. Traditional singularity handling methods are computationally intensive, have low accuracy, and are difficult to respond in real time. Furthermore, traditional path planning methods are prone to getting trapped in local minima in complex environments, and closed-chain singularities reduce path planning efficiency and increase errors.

[0009] 4. Existing master-slave operating systems still have room for improvement in terms of algorithm and parameter optimization for force feedback control, especially in the limited research on the application of electromagnetic force feedback technology in master-slave manipulator systems. Electromagnetic force feedback has advantages over traditional mechanical force feedback, such as faster response speed and simpler structure, but its application in master-slave manipulator systems is still insufficient.

[0010] 5. Existing bidirectional transparent control methods have shortcomings in handling communication delays, singularity avoidance, and adaptive path planning, making it difficult to meet the requirements of high-precision and high-stability master-slave operations. In particular, the overall performance of existing solutions still has considerable room for improvement in key technical aspects such as master-slave ratio matching, electromagnetic force feedback signal processing, and real-time singularity detection and processing.

[0011] Therefore, to address the above issues, a master-slave manipulator with electromagnetic force feedback for precision assembly and a two-way transparent control method are proposed. Summary of the Invention

[0012] The purpose of this invention is to address the shortcomings of the prior art by setting up a master-slave manipulator with electromagnetic force feedback for precision assembly and a two-way transparent control method, thereby solving the technical problems mentioned in the background art.

[0013] To address the above technical issues, the following technical solution is adopted: a master-slave manipulator system with electromagnetic force feedback for precision assembly, comprising a master robot, a slave robot, an information transmission channel, an operator interface, and a control system, wherein:

[0014] The main robot includes a base, a gripping device, and joint structures required to achieve at least a preset number of degrees of freedom of motion. The joint structures correspond to the motion dimensions related to shoulder, elbow, wrist, and hand operations, including the motion dimensions corresponding to the fingers, middle finger, and ring finger, to meet the needs of complex operation posture adjustment. The gripping device is mounted on the end joint. The base is provided with multiple casters at the bottom for easy movement.

[0015] The robot includes a base, a gripping device, and a joint structure that matches the motion degrees of freedom of the main robot. The joint structure is adapted to the joint structure of the main robot and corresponds to the same motion dimensions related to shoulder, elbow, wrist, and hand operations, so as to synchronize the operation actions of the main robot. The gripping device is mounted on the end joint. The bottom of the base is provided with multiple casters for easy movement.

[0016] The information transmission channel includes wired or wireless communication modules for real-time transmission of status information between the master robot and the slave robot.

[0017] The operator interface includes a control handle, a display screen, and control buttons. The control handle is identical in shape to the main robot handle for easy operation. The display screen shows the current status information and force feedback information of the master and slave robots. The control buttons include a stop button, an emergency stop button, and a mode switch button.

[0018] The control system includes a master controller and slave controllers, wherein:

[0019] The main controller adopts an FPGA and ARM architecture design and includes a main control module, a communication module, a force feedback processing module, and a path planning module. The main control module is responsible for coordinating the movement of each joint to achieve position control of the main robot. The communication module is responsible for data transmission with the slave controller and the operator interface. The force feedback processing module processes the received force feedback signals to generate corresponding control signals. The path planning module generates a motion path based on the current position and the target position.

[0020] The controller includes a drive module, a sensing module, and a feedback module. The drive module receives control signals sent by the main controller and drives the robot to move. The sensing module includes a position sensor and a force sensor to detect the robot's state information. The feedback module feeds back the sensed information to the main controller through an information transmission channel.

[0021] Both the master controller and the slave controller adopt a modular design, which facilitates maintenance and upgrades.

[0022] Preferably, the electromagnetic force feedback system includes multiple magnetic coupling modules, each corresponding to a joint, wherein:

[0023] Each magnetic coupling module includes an electromagnet unit consisting of a permanent magnet and a coil, a magnetic field sensor, and a drive circuit. The permanent magnet is made of N52 grade NdFeB material with a magnetic energy product of 9000-11000 kJ / m³. The coil is made of copper wire with a diameter of 0.5-0.8 mm and is wound in a multi-layer stacked manner. The magnetic field sensor is a high-precision triaxial fluxgate sensor with a resolution of 0.1 μT. The drive circuit uses PWM waveform drive with a frequency of 20-25 kHz.

[0024] Furthermore, the joints of the master robot and the slave robot adopt a brushless DC motor drive system, including a motor, a motor driver and an encoder; the motor is a high-precision brushless motor with a rated voltage of 12-24V and a rated speed of 3000-10000rpm; the encoder is a high-resolution optical encoder with a measurement range of 360 degrees and a resolution of 0.005°.

[0025] Preferably, the system employs a two-way transparent control method, specifically including:

[0026] The master controller sends position commands to the slave robot and simultaneously receives position information from the slave robot. The slave robot moves according to the received position commands and simultaneously feeds back the actual position information to the master controller through the communication channel. The master controller calculates the position and velocity gain matrices Kp and Kd based on the current position error and velocity error. The master controller sends the calculated Kp and Kd to the slave robot through the communication channel, and the slave robot performs position and velocity control based on the received Kp and Kd.

[0027] Furthermore, the main controller calculates the target pose based on the operator's intention and generates motion commands based on the current pose. After receiving the motion commands from the robot, it begins to execute the motion and collects environmental feedback force information through the magnetic coupling module. The robot feeds back its position, velocity, acceleration, and environmental feedback force information to the main controller through the communication channel. The main controller corrects the motion commands based on the feedback information to ensure stable system operation.

[0028] Preferably, the system also includes a singularity processing module, comprising:

[0029] The singularity detection unit calculates the Jacobian matrix in real time to determine whether a singularity exists. When a singularity is detected, the singularity processing unit rearranges the joints using SVD decomposition to generate a new control input. The control input is then linearly combined with the original control input using a weighted average method to generate the final control signal, thus avoiding control instability caused by singularities.

[0030] Furthermore, the path planning module employs an adaptive path planning algorithm based on the potential field method, including:

[0031] First, construct the potential field function U(q), where q is the joint angle vector; then calculate the potential field gradient -▽U(q); next, calculate the damping term k▽V(q); finally, use the total potential field gradient as the control input to drive the system motion. The specific expression for the potential field function U(q) is as follows:

[0032] ;

[0033] in, To assist the target point, As a reference point for obstacle avoidance in Cartesian space, and These are weighting coefficients, adjusted in real time to avoid local minima.

[0034] Used to enhance the attraction of the target joint angle to the robot; Used to dynamically adjust the strength of obstacle avoidance constraints; Used to adapt to joint angle deviation correction in different directions. Among them, The angle vectors from the robot joints are acquired in real time by the joint encoder. Generated by mapping the main robot's operational intent. The obstacle boundary is identified by the vision module or generated by the preset assembly path. The three are related through kinematic relationships to ensure that the path optimization meets the requirements of joint motion constraints and Cartesian space obstacle avoidance at the same time.

[0035] Preferably, the system also includes a master-slave system proportional matching module, which is used to automatically adjust the position and force feedback proportional factors according to the size difference between the master and slave robots, thereby improving the force feedback accuracy.

[0036] The beneficial effects of this invention are:

[0037] 1. This invention employs a bidirectional transparent control principle, utilizing a master-slave two-channel control law design combined with electromagnetic force feedback technology to effectively solve the problem of balancing transparency and stability in traditional four-channel controllers under communication delay environments. Even in complex environments with communication delays, the system maintains high transparency and good stability, significantly improving the accuracy of force feedback and the user experience.

[0038] 2. This invention optimizes the electromagnetic force feedback principle, employs a combination of permanent magnets and coils, and achieves high-precision force feedback through precise magnetic field strength control. This design effectively overcomes the influence of factors such as sensor noise, bandwidth limitations, and actuator dynamic characteristics, significantly improving the accuracy and real-time performance of force feedback.

[0039] 3. This invention proposes a novel singularity handling and path planning method, which performs adaptive path planning based on the potential field method and incorporates a damping term to avoid local minima. This method not only has low computational cost and high accuracy, but also can respond in real time, effectively solving the problem of traditional methods getting trapped in local minima in complex environments, and greatly improving the efficiency and accuracy of path planning.

[0040] 4. This invention ensures system stability and accuracy through the rational design of the structural parameters and control system of the master and slave manipulators. When the size difference between the master and slave systems makes it difficult to precisely match the position and force response scaling factors, an appropriate scaling factor matching strategy is adopted, effectively improving the accuracy of force feedback.

[0041] 5. This invention realizes a two-way transparent control and force feedback mechanism for the master-slave manipulator system, enabling the operator to accurately perceive the force information of the remote environment, realize fine operation in complex environments, and significantly improve the application capability of the master-slave operating system in high-risk environments. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] In the attached diagram:

[0044] Figure 1 This is a flowchart illustrating the control system of the present invention;

[0045] Figure 2 This is a flowchart illustrating the control method in this invention. Detailed Implementation

[0046] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0047] Specific implementation examples are given below.

[0048] Example 1

[0049] The master-slave manipulator system with electromagnetic force feedback for precision assembly includes a master robot unit, a slave robot unit, an information transmission unit, a master control unit, a slave control unit, an electromagnetic force feedback unit, a bidirectional transparent control unit, a singularity processing unit, an adaptive path planning unit, and a master-slave proportional matching unit.

[0050] The main robot unit provides six degrees of freedom (DOF) operation and end effector gripping functionality. It comprises six joints and an end effector gripping device. The six joints correspond to the movement dimensions of the shoulder, elbow, wrist, fingers, middle finger, and ring finger, enabling adjustments for complex operational postures. The end effector gripping device features a three-finger adaptive configuration with curved, non-slip fingertips, adaptable to parts ranging from 10mm to 300mm in size. It is made of lightweight, high-strength aluminum alloy and incorporates a micro-displacement sensor with a resolution of 0.001mm. The gripping force ranges from 0.1N to 50N and is adjustable in 10 levels, with a response time of no more than 5ms. It supports both position and force control modes, is compatible with master-slave proportional matching units, and features anti-collision protection exceeding a 50N threshold. It also works in conjunction with an electromagnetic force feedback unit to achieve a closed-loop force feedback mechanism. The main robot unit's base is equipped with multiple omnidirectional wheels for easy movement. The main robot unit's joint drive system employs a combination of a brushless DC motor and a high-resolution encoder. The encoder has an angular resolution of 0.01°. The joint rotation data acquired by the encoder and the motor drive signal form a closed-loop feedback, achieving high-precision closed-loop control of joint movement. The brushless DC motor is a high-precision core LESS motor with a rated voltage of 12-24V and a rated speed of 3000-10000rpm.

[0051] The joint motion stroke of the main robot unit has been optimized, and its workspace is spherical with a radius of 200-300mm. This can cover the operation range of mainstream assembly objects in precision assembly scenarios, ensuring that the operator can reach all target assembly positions through the main end operation.

[0052] The slave robot unit responds to the master robot unit's operation commands and executes remote operations. Its structure is adapted to the master robot unit, also including six joints and an end effector to match the master robot unit's degrees of freedom. Multiple casters are also provided on the bottom of the slave robot unit's base for easy movement. The slave robot unit's joint drive system also uses a combination of brushless DC motors and high-resolution encoders. The encoder's angular resolution is 0.01°. The joint angle data collected by the encoder and the motor drive signals form a closed-loop feedback, achieving high-precision closed-loop control of joint movement.

[0053] The spherical workspace radius of the robot unit is 250-350mm. The size design ensures that the slave end workspace completely covers the master end spherical workspace, and reserves 50-50mm of operation redundancy space. It can adapt to assembly objects of different specifications (such as electronic components and micro mechanical parts) with diameters of 10mm-300mm, meet the high requirements for spatial adaptability in precision assembly, and at the same time ensure the spatial coordination consistency between the master end operation intention and the slave end execution action.

[0054] The information transmission unit establishes real-time data interaction channels between the master robot unit and the slave robot unit, and between the master control unit and the slave control unit, transmitting position commands, status information, and force feedback information. The information transmission unit includes wired or wireless communication modules; communication protocol selection is as follows:

[0055] Wired communication: EtherCAT protocol is used, with a transmission rate of 100Mbps and a periodic communication time of no more than 1ms, ensuring that the delay between modules is no more than 50ms;

[0056] Wireless communication: Adopts WiFi 6E protocol, 5.9GHz frequency band, MIMO technology to enhance anti-interference capability, transmission latency no more than 20ms, packet loss rate no more than 0.1%;

[0057] Data frame structure: frame header (2 bytes) + instruction type (1 byte) + data segment (no more than 64 bytes) + checksum (2 bytes) + frame trailer (1 byte), the checksum uses the CRC16 algorithm;

[0058] Anti-interference measures:

[0059] Hardware aspects: Shielded twisted-pair cable and low-noise amplifier are used;

[0060] At the software level: data retransmission mechanism (retransmission within 10ms when packets are lost), timestamp synchronization (clock deviation between master and slave controllers is no more than 1μs); used to transmit the status information of master and slave robots in real time.

[0061] The main control unit coordinates the overall system control logic, possessing high real-time computing capabilities and multi-module collaborative scheduling capabilities. It receives operator intentions, processes feedback information, and generates motion commands, while providing computing and control support for other functional units. The main control unit adopts an FPGA and ARM architecture to achieve high real-time computing and multi-module collaborative scheduling functions. The FPGA module is specifically designed to handle computing tasks with real-time requirements of no less than 1ms, including gain matrix calculation and potential field function gradient solution.

[0062] The FPGA adaptation logic for solving the gradient of the potential function is as follows: Piecewise gradient calculation uses a hardware comparator to determine x and... The gradient calculation path is switched according to the magnitude relationship; the squaring operation and addition and subtraction operation are implemented through a dedicated arithmetic logic unit (ALU), and the time for a single gradient calculation is no more than 200μs; the angular velocity norm calculation of the damping compensation term is implemented through a pipeline adder and a square root IP core to ensure that the overall algorithm meets the requirement of real-time solution within 1ms and does not affect the closed-loop cycle of the system control.

[0063] The ARM module is responsible for data interaction between units, force feedback signal preprocessing, and motion command generation. The interaction interface uses the AXI4-Lite bus interface, with a data width of 32 bits, a clock frequency of 100MHz, and a read / write response time of no more than 10ns. 2. Data transmission format:

[0064] FPGA to ARM: Gain matrix data, for example, each element is a 32-bit floating-point number, totaling 6×6×4 bytes + status flag bits, for example, 1 byte, 0x00 for normal and 0x01 for abnormal;

[0065] ARM to FPGA: Joint angle instructions (32-bit floating-point number for each angle, totaling 6×4 bytes) + control enable bit (1 byte);

[0066] Synchronization mechanism: Hardware interrupt triggering is adopted. After the FPGA completes real-time calculation, it outputs an interrupt signal (active high level). The ARM responds to the interrupt and reads the data. The interrupt interval is consistent with the control cycle (1ms).

[0067] The main control unit includes a main control module, a communication module, a force feedback processing module, and a path planning module. The main control module coordinates the movement of each joint, achieving position control of the main robot. The communication module handles data transmission with the slave controllers and the operator interface. The force feedback processing module processes the received force feedback signals and generates corresponding control signals. The path planning module generates a motion path based on the current position and the target position.

[0068] The slave control unit drives the movement of the slave robot unit and collects its state information. It includes a sensing module and a drive module. The sensing module acquires the position, velocity, and environmental force information of the slave robot unit, while the drive module controls the joint movement of the slave robot unit according to commands. The slave control unit includes a drive module, a sensing module, and a feedback module. The drive module receives control signals from the main controller and drives the slave robot to move. The sensing module includes position and force sensors to detect the state information of the slave robot. The feedback module feeds back the sensed information to the main controller through an information transmission channel.

[0069] The electromagnetic force feedback unit generates high-precision environmental force feedback signals, including magnetic coupling components corresponding one-to-one with each joint of the master and slave robot units. The magnetic coupling components acquire environmental force information through magnetic field sensing and transmit this information to the main control unit to achieve force feedback reproduction. In the magnetic coupling components of the electromagnetic force feedback unit, this system uses an N52-grade NdFeB permanent magnet with a residual magnetic induction intensity Br=1.42T, coupled with an AWG24 copper coil. Experimental verification shows that the maximum output feedback force of a single magnetic coupling module can reach 15N, the force feedback response time is less than 2ms, and the peak power consumption is no more than 25W, meeting the requirements for high-precision, real-time force feedback. The magnetic field sensor is a three-axis fluxgate sensor with a resolution of 0.1μT; the magnetic coupling gap between the coil and the permanent magnet is adapted to the joint movement stroke, and the output frequency of the drive circuit matches the sampling frequency of the magnetic field sensor to ensure the real-time performance and accuracy of the force feedback signal. Specifically, the electromagnetic force feedback system comprises multiple magnetic coupling modules, each corresponding to a joint. Each magnetic coupling module includes an electromagnet unit combining a permanent magnet and a coil, a magnetic field sensor, and a drive circuit. The permanent magnet uses N52 grade NdFeB material with a magnetic energy product of 9000-11000 kJ / m³. The coil uses copper wire with a diameter of 0.5-0.8 mm, wound using a multi-layer stacked winding process. The magnetic field sensor employs a high-precision triaxial fluxgate sensor with a resolution of 0.1 μT. The drive circuit uses PWM waveform drive with a frequency of 20-25 kHz.

[0070] The core principle of electromagnetic force feedback is based on the Lorentz force law, and the formula for calculating the feedback force is: Where I is the current flowing through the coil (range 0-5A), L is the effective conductor length of the coil (designed according to the joint size, L=20-50mm), and B is the magnetic induction intensity generated by the permanent magnet.

[0071] The joint feedback torque is calculated using the following formula: ,in The joint lever arm (corresponding to the design parameters of each joint, r=15-30mm). V is the coil current density, and V is the coil volume.

[0072] A bidirectional transparent control unit constructs a master-slave control closed loop to balance system transparency and stability under communication delay. By receiving position commands generated by the master control unit and status information fed back by the slave control unit, it calculates the position gain matrix and velocity gain matrix, and sends the gain matrix to the slave control unit to adjust the motion control parameters. The calculation of the gain matrix is ​​based on the data-driven iterative least squares idea. The core is to dynamically optimize the position gain matrix and velocity gain matrix to offset the master-slave synchronization error caused by communication delay (no more than 50ms) while ensuring the transparency of force feedback. The algorithm input data are all measured technical data from the system: the target pose of the main robot, acquired by the main robot joint encoder with an accuracy of 0.001 rad; the real-time position of the slave robot, output from the control unit sensing module with a sampling rate of 1 kHz; the angular velocities of the master and slave joints, calculated through joint angle difference with a filter window length of 5; the environmental force feedback signal Fe, acquired by the electromagnetic force feedback unit magnetic coupling component with a sampling rate of 10 kHz and a resolution of 0.01 N; and the force applied by the operator, data from the end effector force sensor of the main robot with a range of ±50 N. The output is a 6×6 dimensional gain matrix adapted to a 6-DOF manipulator, which directly affects the PID controller parameter adjustment of the slave robot joint drive module. The key parameters of the algorithm are defined as follows: number of joints n=6, matching the 6-DOF motion dimensions of the shoulder, elbow, wrist, and hand; state weight matrix P, initially a 6×6 identity matrix; control weight matrix R=0.01, balancing control energy consumption and error correction intensity; force feedback gain coefficient (adapting to different assembly force feedback requirements); comprehensive error index, where position error (unit: mm), velocity error (unit: rad / s), and force feedback deviation (unit: N), with initial weighting coefficients that can be dynamically adjusted according to the assembly scenario; forming a closed-loop process of command sending, state feedback, parameter adjustment, and motion control. The comprehensive error calculation of the bidirectional transparent control unit includes three types of parameters: position error, velocity error, and force feedback deviation. The weighting coefficients of each parameter are dynamically adjusted according to the real-time operating status of the system; the adjustment direction of the position gain matrix Kp and the velocity gain matrix Kd is adapted to the trend of comprehensive error change, that is, the gain adaptation increases when the comprehensive error increases and decreases when the comprehensive error decreases, in order to balance the system response speed and stability. The specific control method includes: the master controller sends position commands to the slave robot and simultaneously receives position information fed back by the slave robot. The robot moves according to the received position commands and simultaneously feeds back its actual position information to the main controller via the communication channel. The main controller calculates the position and velocity gain matrices Kp and Kd based on the current position and velocity errors. The position gain matrix Kp and the velocity gain matrix Kd are solved using a data-driven iterative least squares algorithm, with the specific steps as follows:

[0073] Define the dimensionality-upgrading feature function ( (for the number of joints), mapping the joint space state to a high-dimensional feature space;

[0074] Build function: ,in This is the state weight matrix (initialized as the identity matrix). To control the weight matrix (value 0.01);

[0075] Sample data of the system operation:

[0076] ;

[0077] Where (m is not less than 1000);

[0078] Update via iterative formula :

[0079] ;

[0080] in, This is the updated state weight matrix. For the current iteration period matrix, For the first The upgraded feature vector of each sample.

[0081] Calculate the gain matrix:

[0082] ;

[0083] in This is the joint drive matrix (obtained from motor parameter calibration).

[0084] Define the comprehensive error The initial weighting coefficients for α, β, and γ are 0.5, 0.3, and 0.2, respectively. When ε is greater than 0.1 mm, the weighting coefficients will be increased by 20%; when ε is less than 0.01 mm, the weighting coefficients will be decreased by 15%, thereby achieving dynamic adaptation of the weighting coefficients.

[0085] The dynamic equations of a two-way transparent control system can be expressed as:

[0086] ;

[0087] ;

[0088] in, , Master-slave inertia matrix, , For Coriolis force and centrifugal force terms, , To control the input, Apply force to the operator, As an environmental reaction force, , These are the joint angle vectors of the master and slave robots, respectively. , These represent the end-effector linear velocities of the master and slave robots, respectively.

[0089] This system adopts a two-channel control law design, the specific expression of which is as follows:

[0090] ;

[0091] ;

[0092] in, and The position and velocity gain matrices mentioned above, This is the force feedback gain coefficient (range 0.8-1.2, dynamically adapted according to the force feedback accuracy). , These are the coordinates of the end effector positions of the master and slave robots, respectively.

[0093] For communication latency scenarios, system stability must meet the following conditions:

[0094] ;

[0095] ;

[0096] in Maximum communication delay (this system) (No more than 50ms) , These are the minimum eigenvalues ​​of the master and slave inertia matrices, obtained through joint parameter calibration. Not less than 0.01 kg·m².

[0097] The master controller sends the calculated Kp and Kd to the slave robot via the communication channel. The slave robot then controls its position and velocity based on the received Kp and Kd. The master controller calculates the target pose based on the operator's intention and generates motion commands based on the current pose. After receiving the motion commands, the slave robot begins to execute the motion and collects environmental feedback force information through the magnetic coupling module. The slave robot feeds back its position, velocity, acceleration, and environmental feedback force information to the master controller via the communication channel. The master controller corrects the motion commands based on the feedback information to ensure stable system operation.

[0098] The singularity processing unit detects and avoids singularities at the joints of the master and slave robot units. It determines the presence of singularities by calculating the Jacobian matrix of joint motion in real time. When a singularity is detected, SVD decomposition is used to rearrange the joint motion parameters to generate new control inputs. These new control inputs are then weighted and combined with the original control inputs to output the final control signal, thus preventing control instability. The singularity processing unit determines the singularity state based on the condition number of the Jacobian matrix. The weighting of the new and original control inputs is dynamically adapted according to the severity of the singularity. The singularity determination criteria are as follows:

[0099] Calculate the condition number of the Jacobian matrix J. ( For the maximum singular value, (Minimum singular value);

[0100] Threshold setting: This is the normal state. Not greater than It is a near-singular state. Not less than It is a singular state;

[0101] SVD decomposition operation: for Perform singular value decomposition ,reserve For singular values ​​of σ_i that are not less than 0.01σ_max, set elements less than this value to zero and reconstruct the Jacobian matrix. Calculate the new control input ( (for false rebellion)

[0102] Weighted weight calculation: Let the weight coefficients be set. Then the final control input ,in :

[0103] Normal state The original control input is used;

[0104] Near-singular state ( Not greater than Less than ): It increases linearly from 0 to 0.8;

[0105] Strange state ( Not less than ): It adopts a completely new control input;

[0106] Furthermore, the weight adjustment process is continuous and smooth, avoiding joint motion shock or control instability caused by sudden changes in control signals. The core algorithm for singular point detection and processing is based on singular value decomposition (SVD), and the Jacobian matrix decomposition form is: Where U and V are orthogonal matrices, and Σ is a diagonal matrix (the diagonal elements are singular values). Not less than Not less than... not less than Not less than 0). When the minimum singular value When the value approaches zero, the system is determined to be approaching or entering a singular state.

[0107] To avoid control failure caused by singularities, the damped least squares method is used to calculate joint velocities: ;in The dynamic damping coefficient is adjusted according to the following logic: when Not less than In the formula Basic damping coefficient , Threshold for identifying singularities ( Simulation results show that the algorithm enables the robotic arm to smoothly pass through the singularity at 80% of its normal operating speed (normal speed not exceeding 5 rad / s), and the position error after passing through is controlled within 0.2 mm, meeting the requirements for high-precision operation.

[0108] The specific processing steps include: the singularity detection unit calculates the Jacobian matrix in real time to determine whether singularities exist. When a singularity is detected, the singularity processing unit rearranges the joints using SVD decomposition to generate new control inputs. The control inputs are then linearly combined with the original control inputs using a weighted average method to generate the final control signal, thus avoiding control instability caused by singularities.

[0109] The adaptive path planning unit generates the optimal motion path for the master and slave robot units. It constructs a potential field function containing target position and weight adjustment terms, calculates the gradient of the potential field function and the damping compensation term, and uses the total potential field gradient as the control input to drive the robot's motion, avoiding path planning from getting trapped in local minima. The path planning module employs an adaptive path planning algorithm based on the potential field method, specifically including: first, constructing the potential field function U(q), where q is the joint angle vector; then calculating the potential field gradient -▽U(q); and finally calculating the damping term k▽V(q), with the following variable definitions:

[0110] x: Position coordinates of the robot's end effector in Cartesian space (Unit: mm);

[0111] : Target joint angle vector at the end effector of the main robot (Unit: rad)

[0112] : Target position coordinates in Cartesian space (Depend on (obtained by solving the forward kinematics);

[0113] Damping term calculation:

[0114] definition Joint velocity energy function:

[0115] ;

[0116] in Joint angular velocity (unit: rad / s);

[0117] Dynamic adjustment of the damping coefficient k: ,in The maximum angular velocity of the joint is set to 5 rad / s to ensure that the greater the joint velocity, the stronger the damping compensation.

[0118] Weighting coefficients applicable range: When the distance d between the robot and the target point is greater than 100mm, When d is less than 20mm, Finally, the total potential gradient is used as the control input to drive the system's motion. The potential field function is... The specific expression is:

[0119] ;

[0120] in, For the target point, The target joint angle vector is set by the operator or preset in the assembly path, with an accuracy of ±0.001 rad. The obstacle avoidance reference point in Cartesian space is generated by the vision module recognizing obstacle boundaries, or preset as a key positioning point for assembly, with a positioning accuracy of ±0.1mm and a weighting coefficient. The straight-line distance between the end effector of the follower robot and the target point Dynamic adaptation ensures a balance between target attraction and obstacle avoidance constraints; and These are weighting coefficients, adjusted in real time to avoid local minima. In a preferred embodiment, , .

[0121] Dynamic adaptation rule for weighting coefficients: When d is greater than 100mm, set... , 1. Enhance the attractiveness of the target location item to improve path planning efficiency; when 20mm is not greater than d and 100mm, set... , Balance attraction and constraint; when d is less than 20mm, set , This weakens the influence of the absolute value constraint term and avoids overshoot or oscillation near the target point.

[0122] In this embodiment, the quadratic term of the joint angle target Attraction of the target position through positive kinematic mapping to Cartesian space; absolute value adjustment term for obstacle avoidance reference point. The Cartesian obstacle avoidance point is converted into a joint angle constraint using inverse kinematics; obstacle avoidance reference point sign adjustment term. This adapts to the deviation direction of the joint angle on both sides of the obstacle avoidance point. For example, when considering the robot's joint angle... and When the corresponding joint angle deviation is in the positive direction, this adjustment item is positive, guiding the joint to correct in the negative direction to ensure a smooth obstacle avoidance path.

[0123] The master-slave proportional matching unit adjusts the position and force feedback ratio between the master and slave robot units. It automatically corrects the ratio factors of position response and force feedback based on the size differences between the master and slave robot units to improve force feedback accuracy. The master-slave system proportional matching module automatically adjusts the position and force feedback ratio factors based on the size differences between the master and slave robots. Key dimensional parameters include the arm length parameters of each joint of the master and slave robots. ( (For the number of joints), calculate the size ratio factor. Master_i / L Slave_i; 2. Position scaling factor K_pos: calculated using a weighted average. Where w_i is the joint weight (shoulder w=0.3, elbow w=0.25, wrist w=0.2, hand w=0.25); 3. Force feedback proportional factor K_force: based on the principle of power conservation ,in The system efficiency coefficient (calibrated experimentally) 4. Dynamic correction: When the master-slave motion error |e_pos| is greater than 0.1mm, K_pos is updated in real time. , To improve the accuracy of force feedback, the maximum permissible error is set to 1mm.

[0124] Both the main and slave control units are modularly designed, supporting independent upgrades of functional modules, with inter-module communication latency not exceeding 50ms, facilitating maintenance and upgrades. The system also includes an operator interface, comprising a control handle, a display screen, and control buttons. The control handle is identical in shape to the main robot's handle for easy operator use. The display screen shows the current status information and force feedback information of both the main and slave robots. Control buttons include a stop button, an emergency stop button, and a mode switch button.

[0125] Example 2

[0126] This embodiment provides an electronic device applied to a master-slave manipulator system with electromagnetic force feedback for precision assembly. The electronic device includes at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor.

[0127] This electronic device is applied to the master-slave manipulator system with electromagnetic force feedback for precision assembly described in Embodiment 1. The system includes a master robot unit, a slave robot unit, an information transmission unit, a master control unit, a slave control unit, an electromagnetic force feedback unit, a bidirectional transparent control unit, a singularity processing unit, an adaptive path planning unit, and a master-slave proportional matching unit. The specific structures and functions of these units are as described in Embodiment 1.

[0128] In this embodiment, the processor of the electronic device can be a single-core or multi-core processor, and it adopts an FPGA and ARM architecture design, the same as the main control unit described in Embodiment 1. The processor has high real-time computing capabilities and multi-module collaborative scheduling capabilities. The FPGA module is specifically designed to handle computing tasks with real-time requirements of not less than 1ms, including gain matrix calculation and potential field function gradient solution.

[0129] The ARM module is responsible for data interaction between units, preprocessing of force feedback signals, and generation of motion commands.

[0130] The memory of an electronic device may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may include non-volatile random access memory. The memory stores program modules such as the operating system, network communication module, electromagnetic force feedback processing module, bidirectional transparent control module, singularity processing module, adaptive path planning module, and master-slave proportional matching module.

[0131] When the electronic device is activated, the processor executes a computer program stored in memory to perform the following functions: receiving the operator's intentions, processing feedback information, and generating motion commands, while providing computational and control support for other functional units. Specifically, when the processor executes the computer program, it implements the functions of the main control unit, bidirectional transparent control unit, singularity processing unit, adaptive path planning unit, and master-slave proportional matching unit described in Embodiment 1.

[0132] The electronic equipment also includes a communication interface for exchanging data with the master robot unit and slave robot units. The communication interface can be wired or wireless, supporting multiple communication protocols to ensure real-time and stable data transmission.

[0133] In a preferred embodiment, the electronic device further includes a display unit for displaying the current status information and force feedback information of the master and slave robots, and a user input unit for receiving user operation instructions.

[0134] When the computer program in the electronic device is executed by the processor, it implements the bidirectional transparent control method, singularity handling method, and adaptive path planning algorithm described in Embodiment 1. Through these algorithms, the electronic device can effectively control the master-slave manipulator system, achieving precise position control and force feedback, while avoiding singularity problems and improving system stability and operational accuracy.

[0135] Example 3

[0136] A master-slave manipulator with electromagnetic force feedback for precision assembly and a bidirectional transparent control method are applied to the master-slave manipulator system with electromagnetic force feedback for precision assembly described in the referenced embodiment. Through multi-step collaborative operation, high-precision control and force feedback are achieved under communication delay conditions. Specifically, the method includes the following steps:

[0137] S1. Master-Slave Proportional Initialization and Adjustment Step: The master-slave proportional matching unit automatically adjusts the position response proportional factor and force feedback proportional factor based on the size difference between the master and slave robot units, ensuring that the master-slave motion dimensions and force feedback signal parameters are matched, and establishing a baseline for basic control parameters. This step is identical in function to the proportional matching module of the master-slave system in the referenced embodiment, used to automatically adjust the position and force feedback proportional factors based on the size difference between the master and slave robots, thereby improving the accuracy of force feedback.

[0138] S2. Control Command Generation and Transmission Steps: The main control unit receives the operator's intention, calculates the target operation pose based on the current pose of the main robot unit, and generates position control commands; the information transmission unit transmits the position control commands to the slave control unit in real time. This step utilizes the main control module function of the main control unit in the referenced embodiment, which is responsible for coordinating the movement of each joint to achieve position control of the main robot.

[0139] S3. Robot Motion Execution and Status Acquisition Steps: After receiving position control commands from the control unit, the drive module drives the six joints of the robot unit to move to execute the target operation. Simultaneously, the sensing module of the control unit collects the position information, velocity information, and environmental feedback force information of the robot unit in real time, and transmits them back to the main control unit through the information transmission unit. This step is the same as the function of the drive module, sensing module, and feedback module of the control unit in the referenced embodiment.

[0140] S4. Electromagnetic force feedback signal processing steps: The electromagnetic force feedback unit receives environmental feedback force information transmitted from the control unit, filters and optimizes this information to generate a reproducible force feedback signal, and transmits it to the main control unit to enable the operator to perceive the remote environmental force. This step is implemented using the electromagnetic force feedback system and force feedback processing module in the referenced embodiment.

[0141] In this embodiment, the electromagnetic force feedback signal optimization processing adopts a nonlinear correction method, which is based on the magnetic field strength-force value mapping relationship between the permanent magnet and the coil. The calibration method of the magnetic field strength-force value mapping relationship is as follows:

[0142] Experimental setup: A standard force sensor (accuracy 0.01N) is fixed to the end effector of the robot, and a known load F = [0.5, 1.0, ..., 50.0]N is applied to the magnetic coupling assembly;

[0143] Data acquisition: The three-dimensional magnetic field strength H(x,y,z) and standard force value F output by the magnetic field sensor were recorded simultaneously. Each set of data was collected 10 times and the average value was taken to obtain the sample set. ;

[0144] Fitting Modeling: A mapping model is constructed using multiple linear regression. ,in , , The fitting coefficients (obtained using the least squares method) (Not less than 0.99), b is the zero-point offset (not greater than 0.02N); , , As an indirect representation of environmental forces, it reflects the change in the magnetic field generated by the relative motion between the coil and the permanent magnet at the joint. It should be noted that in this system... , , This refers to the local magnetic field components of the robot joints, used for force feedback calculations, and is not a component of the Earth's magnetic field (geomagnetic field).

[0145] Environmental compensation logic:

[0146] Temperature compensation: Introducing a correction term ( (This represents the difference between ambient temperature and calibration temperature, in °C).

[0147] External magnetic field interference: Static interference is filtered out by a high-pass filter (cutoff frequency 0.1Hz), and dynamic interference is corrected in real time by a Kalman filter with a filter gain of Q=0.01 and R=0.1.

[0148] This mapping relationship was obtained through pre-experimental calibration and includes environmental factor compensation logic, specifically including three aspects: temperature change compensation, external magnetic field interference suppression, and mechanical vibration filtering. Furthermore, the output amplitude of the force feedback signal is dynamically adjusted according to the operator's operating speed. When the operating speed is low, the force feedback signal amplitude is appropriately increased to improve the perception of fine operation; when the operating speed is high, the force feedback signal amplitude is appropriately reduced to avoid operational instability caused by sudden force feedback, thus adapting to the balance requirements of real-time force feedback and operational safety.

[0149] S5. Two-way closed-loop parameter calculation and correction steps: The two-way transparent control unit receives the position information and velocity information of the slave robot unit forwarded by the main control unit, as well as the force feedback signal output by the electromagnetic force feedback unit, and calculates the current position error and velocity error; based on the error, it generates a position gain matrix (Kp) and a velocity gain matrix (Kd), and sends the gain matrix to the slave control unit through the information transmission unit; the slave control unit adjusts the drive parameters according to the gain matrix, corrects the motion trajectory of the slave robot unit, and forms a control closed loop; this step is the same as the control method of the two-way transparent control unit in the referenced embodiment.

[0150] The adaptive calculation sub-process for the gain matrix is ​​as follows:

[0151] Step 1, Dimensional Upgrading Feature Mapping, Define Feature Function:

[0152] ;

[0153] 6-dimensional joint space state:

[0154] ;

[0155] Mapping to a 12-dimensional feature space improves the accuracy of error fitting and avoids gain adjustment from getting trapped in local optima;

[0156] Step 2, Function construction, establishing the optimization objective function:

[0157] ;

[0158] in This function balances minimizing state error and optimizing control energy consumption, taking the robot joint control input as the basis.

[0159] Step 3: Sample data acquisition. During system initialization, the master and slave manipulators traverse typical assembly postures (covering 80% of the joint range of motion) to collect a sample set.

[0160] ;

[0161] in Not less than 1000 To determine the joint state at the next moment, ensuring that the samples cover different error scenarios;

[0162] Step 4, iterative update of the state weight matrix. The P matrix is ​​optimized using an iterative formula to adapt to changes in the system's dynamic characteristics:

[0163] ;

[0164] The iteration termination condition is: This ensures matrix convergence;

[0165] Step 5, Solve for the gain matrix, based on the P matrix after iterative convergence and the joint driving matrix. ;

[0166] ;

[0167] Calculate the final gain:

[0168] ;

[0169] Step 6, Dynamic gain adjustment: Correct the gain amplitude based on the overall error ε.

[0170] When ε is greater than 0.1 mm, Kp and Kd are increased by 20% to enhance the system response speed; when ε is less than 0.01 mm, the gain is reduced by 15% to avoid system oscillation; when the force feedback deviation ef is greater than 0.5 N, γ is increased to 0.4 to improve the force feedback adaptability.

[0171] The sub-process for solving the adaptive potential gradient is as follows:

[0172] Step 1, Distance Calculation and Parameter Adaptation: Real-time calculation of the distance between the robot's end effector and the target point.

[0173] ;

[0174] The three-dimensional coordinates of the end point. Given the three-dimensional coordinates of the target point, based on Numerical range adaptation weight coefficient , ;

[0175] Step 2, solve for the piecewise gradient. Since the potential field function contains an absolute value term, ... Calculate the gradient for each case at the boundary point. The gradient direction is the direction of decreasing potential energy, which is the direction guiding the robot's motion. hour, The potential gradient is:

[0176] ;

[0177] Step 3, calculate the damping compensation term and define the joint velocity energy function:

[0178] ;

[0179] in For the first The angular velocity of each joint is obtained through the joint angle difference; the dynamic adjustment formula for the damping coefficient is:

[0180] ;

[0181] In the formula The maximum permissible angular velocity of the joint is determined by the motor's performance.

[0182] ;

[0183] The norm of the angular velocity vector; the final damping compensation term is... It is used to suppress oscillations during high-speed motion;

[0184] Step 4: Solve for the total potential field gradient. The final control input is the reversed potential field gradient (for guiding motion) + damping compensation (for stabilizing motion), expressed as:

[0185] ;

[0186] The total gradient is directly used as the control input for the robot's joint motion, driving the robot to move along the optimal path.

[0187] S6. Singularity Real-time Detection and Processing Steps: During the movement of the robot unit, the singularity processing unit calculates the Jacobian matrix of the joint motion of the robot unit in real time, and determines whether there are singularities by using the matrix condition number. If a singularity is detected, SVD decomposition is used to rearrange the joint motion parameters to generate new joint control inputs. The new control inputs are then weighted and averaged with the control inputs corresponding to the original position control commands, and the final control signal is output to the slave control unit to avoid abnormal joint speeds or unstable control. This step is the same as the function of the singularity processing module in the referenced embodiment.

[0188] S7. Adaptive Path Optimization Steps: The adaptive path planning unit constructs a potential field function containing a target position term and a weight adjustment term based on the target pose generated by the main control unit and the real-time position of the slave robot unit; calculates the gradient of the potential field function and superimposes a damping compensation term. The total potential field gradient is obtained; this gradient is then transmitted to the main control unit as an auxiliary control signal. The main control unit uses this signal to optimize subsequent position control commands, preventing the motion path from getting trapped in local minima. This step is the same as the adaptive path planning algorithm in the path planning module of the referenced embodiment.

[0189] In this embodiment, the weighting coefficients of the potential function in the adaptive path optimization step... , This is linked to the distance from the robot unit to the target point and the distance to obstacles, forming a coordinated adjustment relationship. Specifically, when the robot unit is far from the target point, A larger value is used to enhance the target's gravitational pull; when approaching the target point, The value is gradually decreased to avoid overshoot. Similarly, when the robot unit is far from the obstacle, The value is relatively small; when approaching an obstacle, The value is gradually increased to enhance obstacle avoidance capabilities. This dynamic adaptation mechanism ensures the flexibility and safety of path planning. Furthermore, the amplitude adjustment of the damping compensation term is consistent with the changing trend of joint motion velocity. When the joint velocity increases, the damping compensation is correspondingly enhanced to avoid system oscillations; when the joint velocity decreases, the damping compensation is appropriately weakened to ensure a smooth transition of motion states during path optimization.

[0190] Through the coordinated implementation of the above seven steps, this method achieves high-precision control and force feedback under communication delay conditions, effectively improving the operating accuracy and stability of the master-slave manipulator system.

[0191] In the description of this invention, it should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solutions provided in this disclosure can be achieved, and no limitation is imposed herein.

[0192] The above description is merely a preferred embodiment of the present invention and does not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A master-slave manipulator system with electromagnetic force feedback for precision assembly, characterized in that, include: The main robot unit is used to provide operation and end-effector gripping functions with preset degrees of freedom of motion. It includes joint structures and end-effector gripping devices required to realize the preset degrees of freedom of motion. The joint structures correspond to at least the motion dimensions related to shoulder, elbow, wrist and hand operations, so as to realize complex operation posture adjustment. The slave robot unit is used to respond to the operation commands of the master robot unit and perform remote operations. Its structure is adapted to the master robot unit and includes an adapter structure that matches the motion degrees of freedom of the master robot unit and an end-effector gripping device to synchronize the operation actions of the master robot unit. The information transmission unit is used to establish a real-time data interaction channel between the master robot unit and the slave robot unit, and between the master control unit and the slave control unit, and to transmit position commands, status information and force feedback information. The main control unit is used to coordinate the overall control logic of the system; The control unit is used to drive the movement of the robot unit and collect its status information. The electromagnetic force feedback unit is used to generate high-precision environmental force feedback signals. It includes magnetic coupling components that correspond one-to-one with each joint of the master and slave robot units. The magnetic coupling components obtain environmental force feedback information through magnetic field induction and transmit the environmental force feedback information to the master control unit to realize force feedback reproduction. The bidirectional transparent control unit is used to construct a master-slave control closed loop to balance system transparency and stability under communication delay. It receives position commands generated by the master control unit and status information fed back by the slave control unit, calculates position gain matrix and velocity gain matrix, and sends the position gain matrix and velocity gain matrix to the slave control unit to adjust motion control parameters. The singularity processing unit is used to detect and avoid singularities at the joints of the master and slave robot units. It determines whether there are singularities by calculating the Jacobian matrix of joint motion in real time. When a singularity is detected, SVD decomposition is used to rearrange the joint motion parameters to generate new control inputs. The new control inputs are then weighted and combined with the original control inputs to output the final control signal to avoid control instability. The adaptive path planning unit is used to generate the optimal motion path for the master and slave robot units. It constructs a potential field function containing the target position term and the weight adjustment term, calculates the gradient of the potential field function and the damping compensation term, and uses the total potential field gradient as the control input to drive the robot motion, thus avoiding the path planning from getting trapped in local minima. The master-slave proportional matching unit is used to adjust the position and force feedback ratio between the master robot unit and the slave robot unit, and automatically corrects the proportional factor of position response and force feedback according to the size difference between the master and slave robot units. The adaptive path planning unit adopts an adaptive path planning algorithm based on the potential field method. First, it constructs a potential field function U(q), where q is a joint angle vector; second, it calculates the potential field gradient ∇U(q); third, it calculates the damping term k∇V(q); and finally, it uses the total potential field gradient as the control input to drive the system motion. The specific expression for the potential field function U(q) is as follows: ; in, To assist the target point, As a reference point for obstacle avoidance in Cartesian space, and These are weighting coefficients, adjusted in real time to avoid local minima. The singularity processing unit determines the singularity state using the condition number of the Jacobian matrix. The weighted combination of the new control input and the original control input adopts a linear weighting method, with the weights... ,in To calculate the condition number of the Jacobian matrix J, For the maximum singular value, It is the minimum singular value, and the weights are dynamically adapted according to the severity of the singularity.

2. The master-slave manipulator system for precision assembly with electromagnetic force feedback according to claim 1, characterized in that, The magnetic coupling component of the electromagnetic force feedback unit includes a coil, a permanent magnet, a magnetic field sensor, and a driving circuit. The magnetic coupling gap between the coil and the permanent magnet is adapted to the joint movement stroke. The PWM output frequency of the driving circuit matches the sampling frequency of the magnetic field sensor, and the PWM output frequency of the driving circuit is adapted to the magnetic response characteristics of the coil to ensure the real-time performance and accuracy of the force feedback signal.

3. The master-slave manipulator system for precision assembly with electromagnetic force feedback according to claim 1, characterized in that, The main control unit adopts an FPGA and ARM architecture. This architecture is combined to match the high real-time computing capabilities and multi-module collaborative scheduling requirements of the system. It also includes an FPGA module, which is used to carry out real-time computing tasks, specifically for computing tasks with real-time requirements of no less than 1ms. The ARM module is responsible for data interaction between units, force feedback signal preprocessing, and motion command generation. Both the main control unit and the slave control unit are modularly designed, supporting independent upgrades of functional modules.

4. The master-slave manipulator system for precision assembly with electromagnetic force feedback according to claim 1, characterized in that, Both the master robot unit and the slave robot unit's joint drive system adopt a combination structure of brushless DC motor and high-resolution encoder; the rated voltage and speed of the brushless DC motor are adapted to the power output requirements of joint movement, the angle resolution of the encoder is adapted to the high-precision closed-loop control requirements of joint movement, and the encoder measurement range covers the full range of joint movement requirements; the joint rotation angle data collected by the encoder and the motor drive signal form a closed-loop feedback to realize high-precision closed-loop control of joint movement.

5. The master-slave manipulator system for precision assembly with electromagnetic force feedback according to claim 1, characterized in that, The comprehensive error calculation of the bidirectional transparent control unit includes at least three types of parameters: position error, velocity error, and force feedback deviation. The weighting coefficients of each parameter are dynamically adjusted according to the real-time operating status of the system. The adjustment directions of the position gain matrix and velocity gain matrix are adapted to the trend of the overall error change. That is, when the overall error increases, the gain adaptation increases, and when the overall error decreases, the gain adaptation decreases, so as to balance the system response speed and stability. Moreover, when the main control unit generates motion commands, it needs to combine the acceleration information fed back from the robot unit to correct the target operation pose.

6. An electronic device, characterized in that: The electronic device is applied to a master-slave manipulator system for precision assembly with electromagnetic force feedback as described in any one of claims 1 to 5; the electronic device includes at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor.

7. A bidirectional transparent control method, applied to the master-slave manipulator system for precision assembly with electromagnetic force feedback as described in any one of claims 1-5, achieves high-precision control and force feedback under communication delay conditions through multi-step collaboration, specifically including the following steps: S1. Master-slave ratio initialization and adjustment steps: The master-slave ratio matching unit automatically adjusts the position response ratio factor and force feedback ratio factor according to the size difference between the master robot unit and the slave robot unit, so as to adapt the master-slave motion dimensions and force feedback signal parameters and establish a basic control parameter benchmark. S2. Control command generation and transmission steps: The main control unit receives the operator's operation intention, calculates the target operation posture and generates position control commands based on the current posture of the main robot unit; the information transmission unit transmits the position control commands to the slave control unit in real time. S3. Steps for executing robot motion and acquiring status: After receiving position control commands from the control unit, the robot unit's joints are driven to move through the drive module to execute the target operation; at the same time, the robot unit's position information, speed information and environmental feedback force information are acquired in real time by the control unit's sensing module and transmitted back to the main control unit through the information transmission unit. S4. Electromagnetic force feedback signal processing steps: The electromagnetic force feedback unit receives the environmental force feedback information transmitted back from the control unit, performs filtering and signal optimization processing on the environmental force feedback information, generates a reproducible force feedback signal, and transmits it to the main control unit to enable the operator to perceive the remote environmental force. S5. Two-way closed-loop parameter calculation and correction steps: The two-way transparent control unit receives the position information and velocity information of the slave robot unit forwarded by the main control unit, as well as the force feedback signal output by the electromagnetic force feedback unit, and calculates the current position error and velocity error; based on the error, it generates a position gain matrix and a velocity gain matrix, and sends the position gain matrix and velocity gain matrix to the slave control unit through the information transmission unit; The control unit adjusts the drive parameters according to the gain matrix to correct the motion trajectory of the robot unit, thus forming a control closed loop. S6. Singularity Real-time Detection and Processing Steps: During the movement of the robot unit, the singularity processing unit calculates the Jacobian matrix of the joint movement of the robot unit in real time, and then processes it using the matrix condition number. Determine if a singularity exists; If a singularity is detected, the joint motion parameters are rearranged using SVD decomposition to generate a new joint control input. The new control input is then weighted and averaged with the control input corresponding to the original position control command, and the final control signal is output to the slave control unit to avoid abnormal joint speed or unstable control. S7. Adaptive Path Optimization Steps: The adaptive path planning unit constructs a potential field function, including a quadratic term for the target position, a first weight adjustment absolute value term, and a second weight adjustment sign term, based on the target pose generated by the main control unit and the real-time position of the slave robot unit. The joint angle target quadratic term = Obstacle avoidance reference point absolute value adjustment term = Obstacle avoidance reference point symbol adjustment item = ,in This is a vector from the robot's joint angles. The joint angle vector of the main robot's end effector target. Using the Cartesian space obstacle avoidance reference point, the gradient of the potential field function is calculated, and a damping compensation term is superimposed to obtain the total potential field gradient. The total potential field gradient is transmitted to the main control unit as an auxiliary control signal. The main control unit combines this signal to optimize subsequent position control commands and avoid the motion path from getting trapped in local minima.

Images