Robot adjustable joint system

Abstract

The invention provides a robot adjustable joint system, and belongs to the field of robot adjustable joint systems. The method specifically comprises the following steps: S1, signal acquisition: acquiring a joint target angle instruction, a joint output end real-time rotation angle, an external load signal of a joint connection part and a joint internal environment temperature signal in real time through a sensor module arranged in a joint; s2, signal processing: carrying out filtering and noise reduction processing on various collected signals; s3, performing two-parameter parallel adjustment based on accurate angle deviation and load size; and S4, adjusting quantity online correction: performing online correction on the torque adjusting quantity and the rigidity adjusting quantity through a nonlinear compliant model. Two-parameter parallel decoupling control over the joint torque and rigidity is achieved, the joint positioning precision and the smooth impact resistance are improved, meanwhile, through multiple compensation, protection and self-learning optimization mechanisms, it is ensured that a joint system operates safely and stably under different working conditions, the service life is prolonged, and the requirement for multi-joint collaborative operation is met.

Description

A robot adjustable joint system Technical Field

[0001] This invention provides an adjustable joint system for robots, belonging to the field of adjustable joint systems for robots. Background Technology

[0002] As the core component for robot motion execution, the adjustment precision, compliance, and safety of robot joints directly determine the robot's operational capabilities and application range. With the increasing demands for human-robot collaboration, precision assembly, and complex working conditions, traditional robot joint systems have gradually revealed many shortcomings.

[0003] Currently, most robot joints employ a single-parameter adjustment mode, independently adjusting only the joint torque or stiffness. This fails to achieve coordinated control and decoupling between the two, leading to mutual interference during adjustment and significantly reducing joint positioning accuracy. Simultaneously, joint signal acquisition is susceptible to environmental interference and sensor inherent errors, resulting in insufficient signal accuracy and further impacting adjustment effectiveness. Furthermore, existing joint systems lack robust mechanisms for load surge protection, temperature compensation, and transmission backlash compensation. Under conditions of sudden load changes, temperature variations, and long-term operation, problems such as joint jamming, impact damage, and accuracy degradation can easily occur, severely affecting the lifespan and operational stability of the joint system.

[0004] In multi-joint collaborative operation scenarios, traditional joint systems mostly adopt a centralized control strategy. Each joint's adjustment is highly independent, but coordination is poor, easily leading to interference between joints and causing excessive deviations in the robot's end effector pose. Furthermore, most joint systems only support a single adjustment mode, unable to flexibly switch according to different working conditions such as manual operation and automated work, resulting in poor adaptability. At the same time, the adjustment parameters of existing joint systems are mostly fixed settings, unable to self-learn and optimize based on the decay of mechanical characteristics during long-term operation, making it difficult to achieve continuous improvement in adjustment performance.

[0005] While some flexible joint designs have emerged in existing technologies to address the conflict between rigidity and flexibility, most flexible joints suffer from structural complexity, high cost, and insufficient adjustment precision. This makes it difficult to achieve an adaptive match between high-precision positioning and compliant impact resistance, thus failing to meet the practical needs of complex working conditions. Therefore, developing an adjustable robot joint system capable of decoupled torque and stiffness control, possessing multiple compensation and protection mechanisms, and adaptable to various working conditions and multi-joint collaborative operations has become a pressing technical challenge in the field of robot joint technology. Summary of the Invention

[0006] To address the shortcomings of existing robot joint systems, such as low adjustment accuracy, poor compliance, lack of comprehensive compensation and protection mechanisms, and insufficient adaptability, this invention provides an adjustable robot joint system. This system achieves parallel decoupled control of joint torque and stiffness, improving joint positioning accuracy and compliant impact resistance. Simultaneously, through multiple compensation, protection, and self-learning optimization mechanisms, it ensures the safe and stable operation of the joint system under different working conditions, extends its service life, and adapts to the needs of multi-joint collaborative operations.

[0007] To address the aforementioned problems, the proposed technical solution is as follows: an adjustable joint system for robots, specifically comprising the following steps: S1, signal acquisition: real-time acquisition of the joint target angle command, real-time rotation angle of the joint output end, external load signal at the joint connection, and internal environmental temperature signal through a sensor module built into the joint. The external load signal includes axial load, radial load, and torque load. The sampling frequency is set to 100Hz-500Hz to ensure the real-time performance and integrity of the signal acquisition; S2, signal processing: filtering and noise reduction of the acquired signals. A Kalman filter algorithm is used to eliminate environmental interference and sensor errors, obtaining accurate angle deviation values, load detection values, and temperature detection values. The measured value, where the angle deviation value is the difference between the target angle command and the real-time rotation angle; S3, dual-parameter parallel adjustment: based on the precise angle deviation and load size, a dual-parameter parallel adjustment mechanism is adopted to simultaneously adjust the joint output torque and joint stiffness, realizing decoupled control of torque and stiffness, and avoiding mutual interference between the two leading to a decrease in adjustment accuracy; S4, online correction of adjustment amount: the torque adjustment amount and stiffness adjustment amount are corrected online through a nonlinear compliance model. The nonlinear compliance model is pre-calibrated based on the mechanical characteristics of the joint and can dynamically adjust the correction coefficient according to the real-time angle deviation, load change and temperature change, so that the joint can achieve adaptive matching between high-precision positioning and compliant impact resistance performance, meeting the usage requirements under different working conditions.

[0008] Furthermore, the dual-parameter parallel adjustment mechanism specifically includes: calculating and outputting a torque reference value using an angle closed-loop control strategy, with the proportional coefficient of the angle closed-loop control being 0.5-2.0, the integral coefficient being 0.01-0.1, and the derivative coefficient being 0.05-0.5, and achieving precise adjustment of the torque reference value through a PID algorithm; simultaneously, using the load change rate as the core control parameter, calculating the stiffness adjustment coefficient, which is obtained by the ratio of the difference between two adjacent sampled load detection values ​​to the sampling period; when the load change rate increases, the stiffness adjustment coefficient increases accordingly, and when the load change rate decreases, the stiffness adjustment coefficient decreases accordingly; torque adjustment and stiffness adjustment are calculated independently and output synchronously, with the adjustment response time difference between the two not exceeding 10ms, ensuring the smoothness of joint movement.

[0009] Furthermore, the system also includes a load mutation adaptive protection step: a preset load mutation threshold is set according to the joint's rated load, which is 1.2-1.5 times the rated load; the load change rate is monitored in real time, and when the load mutation exceeds the preset threshold, a compliant buffer mode is automatically triggered to quickly reduce the joint stiffness to 30%-50% of the rated stiffness, while limiting the upper limit of torque output, which does not exceed 80% of the rated torque, to avoid impact damage to the joint's mechanical structure caused by the load mutation; when the load detection value tends to stabilize, and the load change rate of three consecutive samples is lower than 50% of the preset threshold, the joint stiffness and torque output are gradually restored. The restoration process adopts a linear incremental method, and the incremental growth rate can be adjusted according to the working condition requirements to achieve a smooth switch between the buffer mode and the normal adjustment mode without obvious impact.

[0010] Furthermore, the establishment and correction process of the nonlinear compliance model specifically includes: firstly, performing full-condition calibration on the robot's adjustable joint system, collecting joint torque, stiffness, and adjustment error data under different angles, loads, and temperatures, establishing a four-dimensional mapping table of angle-torque-stiffness-temperature, and storing it in the joint controller's storage module; during joint adjustment, calling the corresponding data in the mapping table in real time, and calculating the temperature compensation coefficient based on the current temperature detection value. The temperature compensation coefficient is pre-calibrated according to the correlation between temperature and joint mechanical characteristics, with a range of 0.8-1.2; the temperature compensation coefficient is superimposed on the torque adjustment amount and stiffness adjustment amount to complete the online correction of the adjustment amount, ensuring that the joint adjustment accuracy is always kept within ±0.01° under different temperature environments.

[0011] Furthermore, the system also includes an adaptive compensation step for joint transmission clearance: First, through offline calibration, transmission clearance data of the joint under different rotation directions and different angle change rates are collected to establish a transmission clearance compensation table. The transmission clearance compensation table contains clearance compensation angles corresponding to different rotation directions (clockwise and counterclockwise) and different angle change rates (0.1 rad / s-10 rad / s). During joint adjustment, the joint rotation direction and angle change rate are collected in real time. Based on the collected parameters, the corresponding clearance compensation angle is output from the table and superimposed on the target angle command to eliminate the angle deviation caused by transmission clearance, further improve the positioning accuracy of the joint, and ensure that the repeatability error of the joint does not exceed ±0.005°.

[0012] Furthermore, the system employs a dual-encoder feedback mechanism to improve angle detection accuracy: high-precision absolute encoders are installed at both the drive and output ends of the joint, with each encoder having a resolution of at least 16 bits; the drive-end encoder is used to acquire the rotation angle of the joint drive motor, while the output-end encoder is used to acquire the actual output rotation angle of the joint; the difference between the angles acquired by the two encoders is calculated in real time to obtain the transmission error of the joint, and the transmission error is used as the feedback correction amount for angle closed-loop control to adjust the torque reference value in real time, compensate for the mechanical error of the transmission mechanism, and solve the problem of transmission error that cannot be corrected and insufficient adjustment accuracy caused by single encoder feedback, thereby improving the positioning accuracy of the joint by more than 20%.

[0013] Furthermore, the system also includes a parameter self-learning optimization step: during normal joint operation, multiple sets of adjustment data under various working conditions are recorded in real time. The adjustment data includes the target angle, real-time angle, angle deviation, load detection value, temperature detection value, torque adjustment amount, stiffness adjustment amount, and adjustment error. Every 1000-5000 sets of valid data are accumulated, a parameter optimization program is initiated. The gradient descent algorithm is used to iteratively optimize the correction coefficients of the nonlinear compliance model and the PID parameters of the angle closed-loop control. The optimization objective is to minimize the adjustment error and joint motion impact. The optimized parameters are automatically updated to the joint controller, making the subsequent adjustment process more consistent with the actual operating state and load characteristics of the joint, achieving continuous improvement in adjustment performance and adapting to the decay of joint mechanical characteristics during long-term use.

[0014] Furthermore, the system also includes multi-parameter real-time monitoring and load reduction protection steps: During joint adjustment, the operating current of the joint drive motor, the internal temperature of the joint, and the joint output torque are monitored in real time, with preset current thresholds, temperature thresholds, and torque thresholds. The current threshold is 1.1-1.3 times the rated current of the drive motor, the temperature threshold is 60-70℃, and the torque threshold is 1.0-1.2 times the rated torque of the joint. When any parameter exceeds the corresponding preset threshold, the load reduction protection mode is immediately activated, prioritizing the reduction of the joint output torque and stiffness by 20%-40% of the current value, while maintaining the joint's compliant characteristics and avoiding rigid impacts. The system monitors changes in each parameter in real time, and when all parameters return to below the preset thresholds, the normal adjustment mode is gradually restored to ensure the safe and stable operation of the joint system and extend the joint's service life.

[0015] Furthermore, the system supports manual / automatic dual-mode switching, and the two modes can be switched seamlessly: In manual mode, the user can directly specify the stiffness level and torque output range through the robot control system's operating interface. The stiffness level is divided into 5-10 levels, each corresponding to a fixed stiffness value. The torque output range can be adjusted between 20% and 100% of the rated torque. The system adjusts the fixed value according to the parameters given by the user, which is suitable for manual operation and special working conditions. In automatic mode, the system automatically performs adaptive adjustment according to the above method without user intervention. It automatically adjusts the torque and stiffness dynamically according to parameters such as target angle, load changes, and temperature changes, which is suitable for automated operation conditions. When switching between the two modes, a parameter smoothing transition algorithm is used to ensure that the changes in joint torque and stiffness are not abrupt. The angle deviation of the joint during the switching process does not exceed ±0.02°, avoiding the impact of switching shock on the joint and the work object.

[0016] Furthermore, when the robot operates in a multi-joint collaborative manner, the system employs a distributed coordination control strategy to achieve synchronous adjustment of multiple joints: taking the target pose of the robot's end effector as the core optimization objective, the system calculates the target angle commands for each joint using an inverse kinematics algorithm to ensure consistent motion coordination among the joints; it collects the angle deviation, load detection values, and adjustment status of each joint in real time, establishes a multi-joint collaborative optimization model, and distributes the angle adjustment, torque adjustment, and stiffness adjustment amounts of each joint in a distributed and coordinated manner to avoid mutual interference between joints that could lead to end effector pose deviation; when a joint experiences a sudden load change or parameter exceedance, the system dynamically adjusts the adjustment parameters of other joints to ensure that the pose accuracy of the end effector is not affected, while protecting the safety of each joint system and improving the operational stability and accuracy of the multi-joint robot.

[0017] Due to the adoption of the above technical solutions, the beneficial effects of the adjustable joint system of the robot of the present invention are as follows: 1. The present invention adopts a dual-parameter parallel adjustment mechanism to achieve decoupled control of joint torque and stiffness. The two are calculated independently and output synchronously to avoid mutual interference. Combined with signal noise reduction processing of Kalman filter algorithm, the joint adjustment accuracy is significantly improved. At the same time, through online correction of nonlinear compliance model, the adaptive matching of high-precision positioning and compliant impact resistance performance is achieved to adapt to different working conditions.

[0018] 2. This invention sets up multiple compensation mechanisms, including temperature compensation, transmission gap compensation, and dual encoder transmission error compensation. Temperature compensation ensures that the joint adjustment accuracy is stable within ±0.01° under different temperature environments. Transmission gap compensation ensures that the joint repeatability error does not exceed ±0.005°. The dual encoder feedback mechanism improves the joint positioning accuracy by more than 20%, thus comprehensively solving the problem of insufficient accuracy in the prior art.

[0019] 3. The present invention has a complete protection mechanism, including load change adaptive protection and multi-parameter real-time monitoring load reduction protection, which can effectively avoid impact damage to the joint mechanical structure caused by load change and parameter over-limit, ensure the safe and stable operation of the joint system, and extend the service life of the joint.

[0020] 4. This invention sets up a self-learning optimization step for adjustment parameters, and iteratively optimizes the model correction coefficients and PID parameters through the gradient descent algorithm to achieve continuous improvement in adjustment performance, adapt to the decay of mechanical characteristics during long-term operation of the joint, without the need for manual intervention to adjust parameters, thus reducing maintenance costs.

[0021] 5. This invention supports seamless switching between manual and automatic modes, adapting to different working conditions such as manual operation and automated work, with a smooth and shock-free switching process; for multi-joint collaborative operation scenarios, a distributed coordination control strategy is adopted to ensure consistent coordination among joints, avoid mutual interference, and improve the operational stability and end-effector pose accuracy of multi-joint robots.

[0022] 6. The present invention has a reasonable overall structural design, requires no complex additional hardware, has controllable cost, is compatible with various robot equipment such as industrial robots, service robots, and humanoid robots, has a wide range of applications, is easy to promote and apply on a large scale, and solves the industry pain points of existing flexible joint structures being complex, costly, and poorly adaptable. Attached Figure Description

[0023] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 is a core control flowchart of a robot adjustable joint system according to the present invention.

[0024] Figure 2 is a flowchart of the adaptive protection and compliant buffering of load change in a robot adjustable joint system according to the present invention.

[0025] Figure 3 is a flowchart of the transmission backlash compensation based on dual encoders for an adjustable joint system of a robot according to the present invention.

[0026] Figure 4 is a flowchart of the core control of the adjustable joint system of the robot according to the present invention. Detailed Implementation

[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Embodiments

[0028] This embodiment provides an adjustable joint system for a robot, specifically including the following steps: S1, signal acquisition: The joint's target angle command, the joint's output rotation angle, the external load signal at the joint connection, and the internal ambient temperature signal are acquired in real time through the sensor module 1 built into the joint. The external load signal includes axial load, radial load, and torque load. The sampling frequency is set to 300Hz to ensure the real-time performance and integrity of the signal acquisition. The sensor module 1 adopts an integrated design, integrating an angle sensor, a load sensor, and a temperature sensor. It is compact, fits the internal installation space of the joint, and has anti-interference capabilities, which can adapt to the environmental requirements of different scenarios such as industry and service.

[0029] S2. Signal Processing: The signal filtering and noise reduction module 2 filters and reduces noise in the acquired signals. The Kalman filter algorithm is used to eliminate environmental interference and sensor errors, and obtain accurate angle deviation, load detection, and temperature detection values. The angle deviation is the difference between the target angle command and the real-time rotation angle. The state equation and observation equation of the Kalman filter algorithm are pre-configured according to the characteristics of joint signal acquisition. The filter coefficient can be dynamically adjusted to ensure the filtering effect while taking into account the signal processing speed and meeting the real-time adjustment requirements.

[0030] S3. Dual-Parameter Parallel Adjustment: Based on precise angle deviation and load magnitude, the dual-parameter parallel adjustment module 3 employs a dual-parameter parallel adjustment mechanism to synchronously adjust the joint output torque and joint stiffness, achieving decoupled control of torque and stiffness and avoiding mutual interference that could lead to a decrease in adjustment accuracy. As shown in Figure 2, the dual-parameter parallel adjustment mechanism specifically includes: calculating and outputting the torque reference value using an angle closed-loop control strategy. The proportional coefficient of the angle closed-loop control is 1.2, the integral coefficient is 0.05, and the derivative coefficient is 0.2. The torque reference value is precisely adjusted using a PID algorithm. Simultaneously, the stiffness adjustment coefficient is calculated using the load change rate as the core control parameter. The load change rate is obtained by the ratio of the difference between two adjacent sampled load detection values ​​to the sampling period. When the load change rate increases, the stiffness adjustment coefficient increases accordingly; when the load change rate decreases, the stiffness adjustment coefficient decreases accordingly. Torque adjustment and stiffness adjustment are calculated independently and output synchronously, with the adjustment response time difference controlled within 8ms to ensure the smoothness of joint movement.

[0031] S4. Online Correction of Adjustment Amount: The torque and stiffness adjustment amounts are corrected online through the nonlinear compliance model correction module 4. The nonlinear compliance model is pre-calibrated based on the mechanical characteristics of the joint and can dynamically adjust the correction coefficient according to real-time angle deviation, load changes, and temperature changes. This enables the joint to achieve adaptive matching between high-precision positioning and compliant impact resistance, meeting the usage requirements under different working conditions. As shown in Figure 4, the establishment and correction process of the nonlinear compliance model specifically includes: firstly, full-condition calibration of the robot's adjustable joint system is performed, collecting data on different angles (0°-360°), different loads (0-rated load), and different temperatures (-10℃-60℃). The joint torque, stiffness, and adjustment error data are used to establish a four-dimensional mapping table of angle-torque-stiffness-temperature, which is stored in the storage module of the joint controller 12. During joint adjustment, the corresponding data in the mapping table is called in real time, and the temperature compensation coefficient is calculated in combination with the current temperature detection value. The temperature compensation coefficient is pre-calibrated according to the correlation between temperature and joint mechanical characteristics, and the range is 0.8-1.2. In this embodiment, the temperature compensation coefficient is calculated by linear interpolation based on real-time temperature. The temperature compensation coefficient is superimposed on the torque adjustment amount and stiffness adjustment amount to complete the online correction of the adjustment amount, ensuring that the joint adjustment accuracy is always kept within ±0.01° under different temperature environments.

[0032] In this embodiment, the system also includes an adaptive compensation step for joint transmission clearance: the transmission clearance compensation module 6 first uses offline calibration to collect transmission clearance data of the joint under different rotation directions and different angle change rates, and establishes a transmission clearance compensation table. The transmission clearance compensation table contains clearance compensation angles corresponding to different rotation directions (clockwise and counterclockwise) and different angle change rates (0.1 rad / s-10 rad / s). During joint adjustment, the joint rotation direction and angle change rate are collected in real time. The corresponding clearance compensation angle is output according to the collected parameters by looking up the table. The clearance compensation angle is superimposed on the target angle command to eliminate the angle deviation caused by transmission clearance, further improve the positioning accuracy of the joint, and make the repeatability error of the joint not exceed ±0.005°.

[0033] Meanwhile, the system employs a dual-encoder feedback mechanism to improve angle detection accuracy: high-precision absolute encoders are installed at both the drive and output ends of the joint via a dual-encoder feedback module 7, with both encoders having an 18-bit resolution. The drive-end encoder is used to acquire the rotation angle of the joint drive motor, while the output-end encoder is used to acquire the actual output rotation angle of the joint. The difference between the angles acquired by the two encoders is calculated in real time to obtain the joint's transmission error. This transmission error is used as the feedback correction for angle closed-loop control, adjusting the torque reference value in real time to compensate for the mechanical error of the transmission mechanism. This solves the problem of insufficient correction and adjustment accuracy caused by single encoder feedback, improving the joint's positioning accuracy by more than 25%. Example

[0034] This embodiment, based on embodiment 1, adds adaptive protection against sudden load changes, real-time monitoring and load reduction protection for multiple parameters, self-learning optimization of adjustment parameters, manual / automatic dual-mode switching, and multi-joint collaborative control functions, as follows: Adaptive protection against sudden load changes: This is achieved through the load change protection module 5, and the specific steps are shown in Figure 3: A load change threshold is preset, which is set according to the rated load of the joint and is 1.3 times the rated load; The load change rate is monitored in real time. When the load change exceeds the preset threshold, a compliant buffer mode is automatically triggered, quickly reducing the joint stiffness to 40% of the rated stiffness, while limiting the upper limit of torque output, which does not exceed 80% of the rated torque, to avoid impact damage to the joint's mechanical structure caused by the load change; When the load detection value tends to stabilize, and the load change rate of three consecutive samples is less than 50% of the preset threshold, the joint stiffness and torque output are gradually restored. The restoration process adopts a linear increment method, with the increment rate set to 5% / ms of the rated stiffness / torque, to achieve a smooth switch between the buffer mode and the normal adjustment mode without significant impact.

[0035] Multi-parameter real-time monitoring and load reduction protection: This is achieved through the multi-parameter monitoring and protection module 9. During joint adjustment, the module monitors the operating current of the joint drive motor, the internal temperature of the joint, and the joint output torque in real time. Preset current, temperature, and torque thresholds are provided. The current threshold is 1.2 times the rated current of the drive motor, the temperature threshold is 65℃, and the torque threshold is 1.1 times the rated torque of the joint. When any parameter exceeds the corresponding preset threshold, the load reduction protection mode is immediately activated, prioritizing a reduction in the joint output torque and stiffness by 30% of the current value, while maintaining the joint's compliant characteristics to avoid rigid impacts. The module monitors changes in each parameter in real time. When all parameters return to below the preset thresholds, the normal adjustment mode is gradually restored to ensure the safe and stable operation of the joint system and extend the joint's service life.

[0036] Parameter self-learning optimization: This is achieved through parameter self-learning optimization module 8. During normal joint operation, it records adjustment data under multiple working conditions in real time. The adjustment data includes target angle, real-time angle, angle deviation, load detection value, temperature detection value, torque adjustment amount, stiffness adjustment amount, and adjustment error. Every 3000 sets of valid data are accumulated, a parameter optimization program is started. The gradient descent algorithm is used to iteratively optimize the correction coefficient of the nonlinear compliance model and the PID parameters of the angle closed-loop control. The optimization objective is to minimize the adjustment error and joint motion impact. The optimized parameters are automatically updated to the joint controller 12, making the subsequent adjustment process more consistent with the actual operating state and load characteristics of the joint, achieving continuous improvement in adjustment performance and adapting to the decay of joint mechanical characteristics during long-term use. The learning rate of the gradient descent algorithm is set to 0.01, and the number of iterations is set to 100 to ensure the optimization effect while avoiding over-optimization that could lead to adjustment instability.

[0037] Manual / Automatic Dual-Mode Switching: Achieved through the dual-mode switching module 10, supporting seamless switching between manual and automatic modes. In manual mode, the user can directly specify the stiffness level and torque output range through the robot control system's operating interface. The stiffness level is divided into 8 levels, each corresponding to a fixed stiffness value. The torque output range can be adjusted between 20% and 100% of the rated torque. The system adjusts the fixed value according to the parameters given by the user, suitable for manual operation and special working conditions. In automatic mode, the system automatically performs adaptive adjustment according to the method in Example 1, without user intervention. It automatically adjusts the torque and stiffness dynamically based on parameters such as target angle, load changes, and temperature changes, suitable for automated operation conditions. During dual-mode switching, a parameter smoothing transition algorithm is used to ensure that the changes in joint torque and stiffness are not abrupt, and the angle deviation of the joint during the switching process does not exceed ±0.02°, avoiding the impact of switching shock on the joint and the work object.

[0038] Multi-joint cooperative control: When the robot operates in a multi-joint cooperative manner, the multi-joint cooperative control module 11 employs a distributed coordination control strategy to achieve synchronous adjustment of multiple joints. With the target pose of the robot's end effector as the core optimization objective, the target angle commands for each joint are calculated using an inverse kinematics algorithm to ensure consistent motion coordination among the joints. The module also collects real-time data on angle deviations, load detection values, and adjustment states of each joint to establish a multi-joint cooperative optimization model. This model distributes the angle, torque, and stiffness adjustments of each joint in a distributed and coordinated manner, preventing mutual interference between joints that could lead to end effector pose deviations. When a joint experiences a sudden load change or parameter exceedance, the adjustment parameters of other joints are dynamically adjusted to ensure the pose accuracy of the end effector remains unaffected, while simultaneously protecting the safety of each joint system and improving the operational stability and accuracy of the multi-joint robot. In this embodiment, the multi-joint cooperative optimization model is constructed using a neural network algorithm, and after offline training and online fine-tuning, the accuracy and response speed of the cooperative control are ensured. Example

[0039] The difference between this embodiment and Embodiment 2 is as follows: the sampling frequency is set to 100Hz; the proportional coefficient, integral coefficient, and derivative coefficient of the angle closed-loop control are 0.5, 0.01, and 0.05, respectively; the torque and stiffness adjustment response time difference is 10ms; the load mutation threshold is 1.2 times the rated load; after triggering the compliant buffer mode, the joint stiffness is reduced to 30% of the rated stiffness, and the upper limit of torque output is 80% of the rated torque; the temperature compensation coefficient range is 0.8-1.2; the joint adjustment accuracy is maintained within ±0.01°; and the angle change rate range in the transmission clearance compensation table is 0.1rad. / s-10rad / s, repeatability error not exceeding ±0.005°; dual encoders with 16-bit resolution, improving positioning accuracy by over 20%; parameter optimization program initiated every 1000 sets of valid data; in multi-parameter monitoring, the current threshold is 1.1 times the rated current of the drive motor, the temperature threshold is 60℃, the torque threshold is 1.0 times the rated torque of the joint, and the load reduction is 20% of the current value; in manual mode, stiffness levels are divided into 5 levels; in multi-joint collaborative control, the inverse kinematics algorithm adopts the Jacobian matrix pseudo-inverse algorithm to ensure the accuracy of the target angle command calculation for each joint. Example

[0040] The difference between this embodiment and Embodiment 2 is as follows: the sampling frequency is set to 500Hz; the proportional coefficient of the angle closed-loop control is 2.0, the integral coefficient is 0.1, and the derivative coefficient is 0.5; the torque and stiffness adjustment response time difference is 5ms; the load mutation threshold is 1.5 times the rated load; after triggering the compliant buffer mode, the joint stiffness is reduced to 50% of the rated stiffness, and the upper limit of torque output is 80% of the rated torque; the temperature compensation coefficient range is 0.8-1.2; the joint adjustment accuracy is maintained within ±0.01°; and the angle change rate range in the transmission clearance compensation table is 0.1rad / s-1. 0 rad / s, repeatability error not exceeding ±0.005°; dual encoder resolution of 20 bits, improving positioning accuracy by more than 30%; parameter optimization program is launched every 5000 sets of valid data; in multi-parameter monitoring, the current threshold is 1.3 times the rated current of the drive motor, the temperature threshold is 70℃, the torque threshold is 1.2 times the rated torque of the joint, and the load reduction is 40% of the current value; in manual mode, the stiffness level is divided into 10 levels; in multi-joint collaborative control, the inverse kinematics algorithm adopts the Newton-Raphson algorithm to improve the calculation efficiency of target angle commands for each joint under complex working conditions.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0042] The present invention and its embodiments have been described above. This description is not restrictive. In short, if a person skilled in the art is inspired by this description and designs a similar structure and embodiment without departing from the spirit of the present invention, such design should fall within the protection scope of the present invention.

Claims

1. An adjustable joint system for a robot, characterized in that, Specifically, the following steps are included: S1. Signal Acquisition: The joint's built-in sensor module acquires the target angle command, real-time rotation angle at the joint output end, external load signal at the joint connection, and internal ambient temperature signal in real time. The external load signal includes axial load, radial load, and torque load. The sampling frequency is set to 100Hz-500Hz to ensure the real-time performance and completeness of the signal acquisition. S2. Signal Processing: The acquired signals are filtered and noise-reduced using a Kalman filter algorithm to eliminate environmental interference and sensor errors, obtaining accurate angle deviation, load detection, and temperature detection values. The angle deviation is the difference between the target angle command and the real-time rotation angle. S3. Dual-Parameter Parallel Adjustment: Based on the accurate angle deviation and load magnitude, a dual-parameter parallel adjustment mechanism is used to simultaneously adjust the joint output torque and joint stiffness, achieving decoupled control of torque and stiffness and avoiding mutual interference that could reduce adjustment accuracy. S4. Online Correction of Adjustment Amount: The torque adjustment amount and stiffness adjustment amount are corrected online through a nonlinear compliance model. The nonlinear compliance model is pre-calibrated based on the mechanical characteristics of the joint and can dynamically adjust the correction coefficient according to real-time angle deviation, load change and temperature change, so that the joint can achieve adaptive matching between high-precision positioning and compliant impact resistance performance, and meet the usage requirements under different working conditions.

2. The adjustable joint system for a robot according to claim 1, characterized in that: The dual-parameter parallel adjustment mechanism specifically includes: calculating and outputting a torque reference value using an angle closed-loop control strategy, with a proportional coefficient of 0.5-2.0, an integral coefficient of 0.01-0.1, and a derivative coefficient of 0.05-0.5 for the angle closed-loop control, and achieving precise adjustment of the torque reference value through a PID algorithm; simultaneously, using the load change rate as the core control parameter, calculating the stiffness adjustment coefficient, which is obtained by the ratio of the difference between two adjacent sampled load detection values ​​to the sampling period; when the load change rate increases, the stiffness adjustment coefficient increases accordingly, and when the load change rate decreases, the stiffness adjustment coefficient decreases accordingly; torque adjustment and stiffness adjustment are calculated independently and output synchronously, with a response time difference of no more than 10ms between the two, ensuring the smoothness of joint movement.

3. The adjustable joint system for a robot according to claim 2, characterized in that: It also includes a load mutation adaptive protection step: a preset load mutation threshold is set according to the rated load of the joint, which is 1.2-1.5 times the rated load; the load change rate is monitored in real time, and when the load mutation exceeds the preset threshold, a compliant buffer mode is automatically triggered to quickly reduce the joint stiffness to 30%-50% of the rated stiffness, while limiting the upper limit of torque output, which does not exceed 80% of the rated torque, to avoid impact damage to the joint's mechanical structure caused by the load mutation; when the load detection value tends to stabilize, and the load change rate of three consecutive samples is less than 50% of the preset threshold, the joint stiffness and torque output are gradually restored. The restoration process adopts a linear incremental method, and the incremental growth rate can be adjusted according to the working condition requirements to achieve a smooth switch between the buffer mode and the normal adjustment mode without obvious impact.

4. The adjustable joint system for a robot according to claim 1, characterized in that: The establishment and correction process of the nonlinear compliant model specifically includes: First, the robot's adjustable joint system is calibrated under all working conditions, and joint torque, stiffness, and adjustment error data are collected under different angles, loads, and temperatures. A four-dimensional mapping table of angle-torque-stiffness-temperature is established and stored in the storage module of the joint controller. During joint adjustment, the corresponding data in the mapping table is called in real time, and the temperature compensation coefficient is calculated in combination with the current temperature detection value. The temperature compensation coefficient is pre-calibrated based on the correlation between temperature and joint mechanical characteristics, with a range of 0.8-1.

2. The temperature compensation coefficient is superimposed on the torque adjustment amount and stiffness adjustment amount to complete the online correction of the adjustment amount, ensuring that the joint adjustment accuracy is always kept within ±0.01° under different temperature environments.

5. The adjustable joint system for a robot according to claim 1, characterized in that: It also includes an adaptive compensation step for joint transmission clearance: First, through offline calibration, transmission clearance data of the joint under different rotation directions and different angle change rates are collected to establish a transmission clearance compensation table. The transmission clearance compensation table contains the clearance compensation angle corresponding to different rotation directions (clockwise and counterclockwise) and different angle change rates (0.1 rad / s-10 rad / s). During joint adjustment, the joint rotation direction and angle change rate are collected in real time. The corresponding clearance compensation angle is output according to the collected parameters by looking up the table. The clearance compensation angle is superimposed on the target angle command to eliminate the angle deviation caused by transmission clearance, further improve the positioning accuracy of the joint, and make the repeatability error of the joint not exceed ±0.005°.

6. The adjustable joint system for a robot according to claim 1, characterized in that: This method employs a dual-encoder feedback mechanism to improve angle detection accuracy: high-precision absolute encoders are installed at the drive end and output end of the joint, with both encoders having a resolution of no less than 16 bits; the drive end encoder is used to acquire the rotation angle of the joint drive motor, and the output end encoder is used to acquire the actual output rotation angle of the joint. The difference between the angles collected by the two encoders is calculated in real time to obtain the transmission error of the joint. The transmission error is used as the feedback correction amount for angle closed-loop control. The torque reference value is adjusted in real time to compensate for the mechanical error of the transmission mechanism. This solves the problem that the transmission error caused by the feedback of a single encoder cannot be corrected and the adjustment accuracy is insufficient, thus improving the positioning accuracy of the joint by more than 20%.

7. The adjustable joint system for a robot according to claim 1, characterized in that: It also includes a parameter self-learning optimization step: during normal joint operation, multiple sets of adjustment data under various working conditions are recorded in real time. The adjustment data includes target angle, real-time angle, angle deviation, load detection value, temperature detection value, torque adjustment amount, stiffness adjustment amount, and adjustment error. Every 1000-5000 sets of valid data are accumulated, a parameter optimization program is initiated. The gradient descent algorithm is used to iteratively optimize the correction coefficients of the nonlinear compliance model and the PID parameters of the angle closed-loop control. The optimization objective is to minimize the adjustment error and joint motion impact. The optimized parameters are automatically updated to the joint controller, making the subsequent adjustment process more consistent with the actual operating state and load characteristics of the joint, achieving continuous improvement in adjustment performance and adapting to the decay of joint mechanical characteristics during long-term use.

8. The adjustable joint system for a robot according to claim 1, characterized in that: It also includes multi-parameter real-time monitoring and load reduction protection steps: During joint adjustment, the operating current of the joint drive motor, the internal temperature of the joint, and the joint output torque are monitored in real time. Preset current threshold, temperature threshold, and torque threshold are set, where the current threshold is 1.1-1.3 times the rated current of the drive motor, the temperature threshold is 60-70℃, and the torque threshold is 1.0-1.2 times the rated torque of the joint. When any parameter exceeds the corresponding preset threshold, the load reduction protection mode is immediately activated, prioritizing the reduction of the joint output torque and stiffness by 20%-40% of the current value, while maintaining the joint's compliant characteristics and avoiding rigid impacts. The changes of each parameter are monitored in real time, and when all parameters return to below the preset threshold, the normal adjustment mode is gradually restored to ensure the safe and stable operation of the joint system and extend the service life of the joint.

9. The adjustable joint system for a robot according to claim 1, characterized in that: This method supports manual / automatic dual-mode switching, and the two modes can be switched without disturbance: In manual mode, the user can directly give the stiffness level and torque output range through the operation interface of the robot control system. The stiffness level is divided into 5-10 levels, each corresponding to a fixed stiffness value. The torque output range can be adjusted between 20% and 100% of the rated torque. The system adjusts the fixed value according to the parameters given by the user, which is suitable for manual operation and special working conditions. In automatic mode, the system automatically performs adaptive adjustment according to the method described in any of claims 1-8 without user intervention. It automatically adjusts the torque and stiffness dynamically according to parameters such as target angle, load change, and temperature change, which is suitable for automated operation conditions. When switching between the two modes, a parameter smoothing transition algorithm is adopted to ensure that the changes in joint torque and stiffness are not abrupt. The angle deviation of the joint during the switching process does not exceed ±0.02°, avoiding the impact of switching shock on the joint and the working object.

10. The adjustable joint system for a robot according to claim 1, characterized in that: When the robot works in multi-joint collaborative mode, a distributed coordination control strategy is adopted to achieve synchronous adjustment of multiple joints: taking the target pose of the robot's end effector as the core optimization target, the target angle command of each joint is calculated through the inverse kinematics algorithm to ensure that the motion of each joint is coordinated and consistent. The system collects the angle deviation, load detection value, and adjustment status of each joint in real time, establishes a multi-joint collaborative optimization model, and distributes the angle adjustment, torque adjustment, and stiffness adjustment of each joint in a distributed and coordinated manner to avoid mutual interference between joints that could lead to end effector pose deviation. When a joint experiences a sudden load change or parameter exceedance, the system dynamically adjusts the adjustment parameters of other joints to ensure that the pose accuracy of the end effector is not affected, while protecting the safety of each joint system and improving the operational stability and accuracy of the multi-joint robot.

Images