Unmanned aerial vehicle system based on flexible mechanical arm and control method

Abstract

The invention relates to an unmanned aerial vehicle system based on a flexible mechanical arm. The unmanned aerial vehicle system comprises the flexible mechanical arm and an unmanned aerial vehicle. The flexible mechanical arm comprises a plurality of body section modules and clamping jaw modules, and the adjacent body section modules are connected in series through supporting pipe columns. The body section module comprises a paper folding structure, a pneumatic module and a first shape memory alloy spring; each paper folding structure comprises a plurality of joints, and the multiple first shape memory alloy springs are arranged on the outer side of each paper folding structure; the clamping jaw module is arranged on the body section module at the tail end of the flexible mechanical arm; the unmanned aerial vehicle comprises a flight control system, and the flight control system is used for controlling the flight of the unmanned aerial vehicle and adjusting the working states of the segment module and the clamping jaw module. According to the method, real-time response can be made to key structure parameters such as load, rigidity, position and force control at the same time, the limitation of a traditional single task target is broken through, and the motion state of each joint can be automatically adjusted according to the actual working condition.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) control, and in particular to a UAV system and control method based on a flexible robotic arm. Background Art

[0002] With the continuous development of drone technology, its application in many fields such as industry, agriculture, logistics and rescue is becoming more and more extensive. However, when faced with complex operational tasks, existing drone systems and their control methods still have some limitations.

[0003] UAV platforms typically consist of flight control systems, power systems, communications systems, and payload systems, enabling aerial operations. A flexible robotic arm is a mechanical structure designed to mimic biological muscles or natural structures, enabling multi-degree-of-freedom motion to perform precise manipulation tasks. It can better adapt to environmental changes and adjust its shape when encountering objects of varying shapes and sizes for better grasping and manipulation. Furthermore, when colliding with the environment or other objects, the flexible robotic arm can absorb impact forces, reducing damage to the drone and its surroundings and improving operational safety. These technologies lay the foundation for the integration of drones with flexible robotic arms, enabling drones to operate in confined or complex environments while improving operational safety and adaptability.

[0004] Existing technical solutions have explored the integration of multi-rotor drones with robotic arms. For example, the Chinese invention patent "A UAV Equipped with a Robotic Arm" (Publication No. CN105314102a) discloses a drone equipped with a hyper-redundant robotic arm. This robotic arm comprises an upper arm, a middle arm, a lower arm, an electronic control module, and a robotic gripper. The end gripper is replaceable to meet the needs of different industries for drones equipped with robotic arms with different functions. However, the robotic arm in this technical solution has low degrees of freedom and a limited workspace. The requirement to replace the robotic gripper to adapt to different mission requirements also indicates that the gripper structure has poor environmental adaptability.

[0005] Another existing technology solution, "A Robotic Arm with a Multi-rotor UAV" (Publication No. CN105014687a), discloses a robotic arm attached to a multi-rotor UAV. The robotic arm is connected to the bottom of the multi-rotor UAV and consists of a main arm, a lower arm, and a gripper. Although the UAV platform has sufficient load capacity, the limited degrees of freedom of the robotic arm prevents it from reaching any position and posture within the operating space, limiting its ability to flexibly avoid obstacles.

[0006] In summary, existing technical solutions have the following major shortcomings: First, the robot arm's degrees of freedom and environmental adaptability are insufficient. Existing robot arms generally have low degrees of freedom, which limits their workspace and operational flexibility. Second, the robot arm has low rigidity. During movement or when subject to external interference, it is easy to cause vibration, reduce control accuracy, deteriorate system stability, and even cause structural fatigue damage. These problems limit the application capabilities of UAV systems in complex environments, especially in tasks that require precise operation and environmental interaction. Therefore, existing technical solutions have obvious deficiencies in load capacity, environmental adaptability, operational flexibility, safety protection performance, and manufacturing costs, and are in urgent need of further technical improvement and innovation. Summary of the Invention

[0007] To this end, the technical problem to be solved by the present invention is to overcome the deficiencies in the prior art and provide a UAV system and control method based on a flexible robotic arm.

[0008] In a first aspect, to solve the above technical problems, the present invention provides a UAV system based on a flexible robotic arm, comprising: a flexible robotic arm and a UAV, wherein the UAV is connected to the flexible robotic arm; The flexible robotic arm includes several segment modules and gripper modules, and adjacent segment modules are connected in series via support columns. The segment modules include an origami structure, a pneumatic module, and a first shape memory alloy spring. The pneumatic module is embedded in the origami structure to control the length of the origami structure. The origami structure includes multiple joints, and the first shape memory alloy spring includes multiple springs, each of which is arranged on the outside of the origami structure to adjust the bending angle of the joint. A gripper module is provided on the segment module at the end of the flexible robotic arm, and includes an origami gripper, a sliding assembly, a traction device, and a second shape memory alloy spring; the sliding assembly is provided on the inner side of the origami gripper, the traction device is provided between the sliding assembly and the gripping end of the origami gripper, and the second shape memory alloy spring is provided on the outer side of the origami gripper for controlling the opening and closing of the origami gripper; The UAV includes a flight control system, which is used to control the flight of the UAV and adjust the working state of the flexible robotic arm.

[0009] In one embodiment of the present invention, the flight control system includes a flight control board, an inertial measurement unit (IMU), a global positioning system (GPS), a barometer, and a communication module. The inertial measurement unit (IMU), the global positioning system (GPS), and the barometer are all connected to the flight control board to monitor the attitude angle, position, and pressure altitude of the UAV; the communication module is connected to the flight control board to realize data communication between the UAV and a ground station or an operator. The flight control board includes a microcontroller, a human-computer interaction module, a visual recognition module, a flight control data acquisition module, a path planning module, a pneumatic control module, a shape memory alloy control module, a gripper control module, a Kalman filter fusion module, a compensation algorithm module, and a safety monitoring module. Human-computer interaction module, used for information exchange between operators and the system; Visual recognition module, which is mainly used to identify and locate target objects; Flight control data acquisition module, used to obtain the flight control data of the drone; A pneumatic control module, used to control the extension and contraction of the pneumatic module of the segment module; a shape memory alloy control module, configured to control the shape of the first shape memory alloy spring and the second shape memory alloy spring; The gripper control module is used to control the opening and closing of the gripper module and the gripping force of the gripper module; Kalman filter fusion module, used to fuse flight control data; The compensation algorithm module is used to correct the deviation of the robotic arm caused by the change of the drone's posture based on the fused flight control data; Safety monitoring module, which monitors the drone's air pressure, the spring temperature of the flexible robotic arm, and the gripping pressure of the gripper module in real time; The microcontroller is connected to the human-computer interaction module, path planning module, pneumatic control module, shape memory alloy control module, gripper control module, Kalman filter fusion module, compensation algorithm module and safety monitoring module, receives data from each module and sends control instructions.

[0010] In a second aspect, in order to solve the above technical problems, the present invention provides a control method for a UAV system based on a flexible robotic arm as described in the first aspect, which specifically comprises the following steps: S1: obtaining the coordinate position of the target object; the specific method of obtaining the coordinate position of the target object includes the user inputting the target coordinates through the graphical user interface GUI and obtaining the three-dimensional coordinates of the object through the visual recognition module; S2: generating movement paths of the segment module and the gripper module of the flexible robotic arm based on the acquired coordinate position of the target object; S3: Based on the generated movement path, the drone driving body module and the gripper module are controlled to move according to the movement path; at the same time, the attitude angle sent by the flight control system of the drone is obtained, and the attitude angle includes the roll angle and the pitch angle; S4: Fusing the roll angle and pitch angle through Kalman filtering to obtain an estimated attitude angle; S5: Based on the data fused by the Kalman filter, the bending angle of the flexible robotic arm is compensated through the compensation algorithm module; S6: After the gripper module grasps the target object, it estimates the mass of the target object and adjusts the stiffness of the segment module and the clamping force of the gripper module according to the weight of the target object.

[0011] In one embodiment of the present invention, in step S2, the specific steps of generating the moving path are as follows: S21: Solve the inverse kinematics equations based on the gradient projection method to obtain the bending angles of each joint of the flexible robotic arm; S22: The RRT* algorithm is used and Gaussian sampling strategy is introduced to generate collision-free paths in three-dimensional space.

[0012] In one embodiment of the invention, in step S21, the specific steps of the gradient projection method include: S211: Establishing a Jacobian matrix of the flexible manipulator and determining iteration parameters and constraint parameters, constructing a manipulator motion model based on the Jacobian matrix, and converting the displacement constraint of the gripper module of the flexible manipulator into the joint velocity constraint of the segment module; S212: constructing a hierarchical dynamic weighting factor matrix, and optimizing joint angle parameters through the hierarchical dynamic weighting factor matrix; S213: Calculating the position error between the target object and the gripper module of the current flexible robotic arm based on the robotic arm motion model, and converting the position error into the velocity of the gripper module by using a gradient descent method; S214: Calculate a weighted pseudo-inverse matrix based on the Jacobian matrix, and use the calculated weighted pseudo-inverse matrix to solve the joint velocity that satisfies the end constraint; and construct a weighted null space projection matrix, define the null space optimization target, generate an optimization direction, and guide the joint away from the extreme position; S215: Based on the joint velocity and zero space optimization that satisfy the end constraint, generate the joint velocity adjustment of the body segment module, convert the optimized joint velocity adjustment into a joint position increment; apply the joint position increment to the current joint vector, and iteratively update the joint vector; S216: Perform constraint projection on the updated joint vector to ensure that the joint state meets the constraint conditions; and calculate the updated position error to determine whether the termination condition is met or the maximum number of iterations is reached. If so, terminate the iteration and output the final joint angle vector; otherwise, return to step S214 to continue the iterative optimization process.

[0013] In one embodiment of the present invention, the expression of the robot arm motion model in step S211 is as follows: ; in, is the Jacobian matrix; Joint velocity; is the joint angle vector of the robot arm, where k is the degree of freedom of the robot arm; is the speed of the end effector, is the three-dimensional linear velocity, is the three-dimensional angular velocity.

[0014] In one embodiment of the present invention, in step S213, based on the robot arm motion model, the expression for calculating the position error between the target object and the current gripper module of the flexible robot arm is as follows: ; is the position error vector of the gripper module; is the position of the target, is the current position of the gripper module; By using the gradient descent method, the expression for converting the position error into the terminal velocity is as follows: ; in, is the error gain coefficient, which controls the convergence speed.

[0015] In one embodiment of the present invention, in step S212, the expression of the hierarchical dynamic weighting factor matrix W is as follows: ; in, is the weight of the first joint to the nth joint of the flexible manipulator, where n is the degree of freedom of the flexible manipulator, that is, the sum of the total number of joints of the flexible manipulator and the number of gripper modules, and The values decrease in sequence.

[0016] In one embodiment of the present invention, in step S5, the bending angle of the flexible manipulator is compensated by a compensation algorithm module based on the data fused by the Kalman filter; wherein the relationship between the compensation angle of the flexible manipulator and the posture angle is expressed as follows: ; in, is the compensation angle of the robot arm, and are the roll angle and pitch angle of the drone, is the feedforward compensation coefficient.

[0017] In one embodiment of the present invention, in step S6, the specific method of identifying the mass of the target object includes: Get the motor current of the flexible robotic arm when it is unloaded; Obtain the motor current when the flexible robotic arm performs a grasping action; Based on the two current data obtained, the estimated mass of the target object is obtained. The expression of the estimated mass of the target object is as follows: ; Where m is the estimated mass of the target object, is the motor current when the robot arm performs the grasping action, is the motor current when there is no load, k is the proportional coefficient of the motor torque and current, and g is the acceleration due to gravity.

[0018] The above technical solution of the present invention has the following beneficial effects compared with the prior art: (1) The UAV system based on the flexible robotic arm described in the present invention is integrated with the UAV. The axial deformation of the flexible robotic arm relies on the pneumatic module to promote the axial extension of the flexible robotic arm, so that the axial bending stiffness of the robotic arm is large and the elongation is considerable. The length can be adjusted as needed. The bending deformation adopts the shape memory alloy spring and pneumatic auxiliary drive. The shape memory alloy wire is heated to shrink it, and the pneumatic module is inflated to achieve large-angle bending. The shape memory alloy wire and the pneumatic module respond quickly, allowing the robotic arm to quickly switch configurations and efficiently complete tasks in scenes such as grasping objects and avoiding obstacles. A multi-dimensional dynamic coupling mechanism is realized, which can simultaneously make real-time responses to key structural parameters such as load, stiffness, position and force control, breaking through the limitations of traditional single task objectives. This multi-dimensional collaborative response characteristic enables the robotic arm to automatically adjust the motion state of each joint according to the actual working conditions when facing complex and changeable working scenes, ensuring that the robotic arm maintains stable stiffness under different load conditions, accurately locates the position of the target object, and realizes precise control of the grasping force, thereby significantly improving the comprehensive performance and environmental adaptability of the robotic arm.

[0019] (2) The control method of the flexible manipulator-based UAV system described in the present invention realizes the rapid path planning of the flexible manipulator in complex environments through the coordinated optimization of the gradient projection method and the improved RRT* algorithm, shortens the path planning time, and effectively solves the real-time planning problem of the super-redundant manipulator in narrow spaces and dynamic obstacle scenes; adopts the dynamic weighting factor matrix W to optimize the joint control torque distribution, so that the load capacity of the manipulator is significantly improved; Kalman filtering integrates the attitude angle to realize feedforward compensation when the UAV attitude deviation is greater than 5°, effectively offsets the influence of flight disturbance on the positioning of the manipulator end, and ensures the relative stability of the target object; and also realizes the adaptive adjustment of the manipulator stiffness during the grasping process through load mass identification combined with aerodynamic stiffness adjustment, ensuring the stable grasping of objects of different masses. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings.

[0021] Figure 1Schematic diagram of the structure of the UAV system based on the flexible robotic arm in the present invention; Figure 2 Schematic diagram of the structure of the flexible robotic arm in the present invention; Figure 3 Schematic diagram of the structure of the body segment module in the present invention; Figure 4 for Figure 3 Schematic diagram of the internal structure of the midsegment module; Figure 5 Schematic diagram of the structure of the gripper module in the present invention; Figure 6 Schematic diagram of the internal structure of the gripper module of the present invention; Figure 7 This is a flow chart of a control method for a UAV system based on a flexible robotic arm according to the present invention; Explanation of the reference numerals in the specification: 1. UAV; 2. Segment module; 21. Origami structure; 22. Pneumatic module; 23. First shape memory alloy spring; 3. Gripper module; 31. Origami gripper; 32. Sliding assembly; 33. Traction device; 34. Second shape memory alloy spring; 321. Slide rail; 322. Slide plate. DETAILED DESCRIPTION

[0022] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0023] Reference Figure 1-2 As shown, the present invention provides a UAV system based on a flexible robotic arm, comprising: a flexible robotic arm and a UAV 1, wherein the UAV 1 is connected to the flexible robotic arm; the flexible robotic arm comprises three segment modules 2 and a gripper module 3, wherein adjacent segment modules 2 are arranged in series via a support column; the UAV 1 and the segment modules 2 are connected via inter-module connecting bolts, and the remaining segment modules 2 and gripper modules 3 are also connected in the same manner. The inter-module connecting bolts pass through the corresponding connection holes of adjacent modules, fixing the multiple modules together to form a complete structure, ensuring the relative position stability between the modules and maintaining the structural integrity of the entire device.

[0024] The body segment module 2 includes an origami structure 21, a pneumatic module 22, and a first shape memory alloy spring 23. The pneumatic module 22 is embedded in the origami structure 21 to control the length of the origami structure 21. The origami structure 21 includes multiple joints. The first shape memory alloy spring 23 includes multiple springs, each of which is disposed on the outside of each origami structure 21 to adjust the bending angle of the joint.

[0025] In this embodiment, four first shape memory alloy springs 23 are connected at one end to the lower PCB and at the other end to the upper PCB. When heated, the first shape memory alloy springs 23 return to their memorized shape, causing the distance and angle between the lower and upper PCBs to change.

[0026] The specific design of the body segment module 2 is as follows Figure 3 and 4 As shown, the pneumatic module 22 includes a convergence plate 221, a main air supply pipe 222, a first branch air supply pipe 223, a second branch air supply pipe 224 and an air bag 225. One end of the main air intake pipe 222 is connected to the convergence plate 221, and the other end is connected to the air pump. The convergence plate 221 is connected to the air bag 225 through the first branch air supply pipe 223 and the second branch air supply pipe 224.

[0027] The gripper module 3 is provided on the segment module 2 at the end of the flexible robotic arm. The gripper module 3 includes an origami gripper 31, a sliding assembly 32, a traction device 33, and a second shape memory alloy spring 34. The sliding assembly 32 is provided on the inner side of the origami gripper 31, the traction device 33 is provided between the sliding assembly 32 and the gripping end of the origami gripper 31, and the second shape memory alloy spring 34 is provided on the outer side of the origami gripper 31 for controlling the opening and closing of the origami gripper 31. The sliding assembly 32 includes a slide rail 321 and a slide plate 322. The slide plate 322 is arranged on the outside of the slide rail 321 and can move along the slide rail 321. The second shape memory alloy spring 34 is arranged between the slide plate 322 and the PCB circuit board connected to the top of the origami gripper 31. The traction device 33 is arranged between the slide plate 322 and the gripping end of the origami gripper 31. The specific structure is as follows Figure 5 and 6 shown.

[0028] UAV 1 includes a flight control system for controlling the flight of UAV 1 and adjusting the working state of the flexible robotic arm. In this embodiment, the flight control system includes a flight control board, an inertial measurement unit (IMU), a global positioning system (GPS), a barometer, and a communication module. The IMU, GPS, and barometer are all connected to the flight control board to monitor the attitude angle, position, and barometric altitude of UAV 1. The communication module is connected to the flight control board to enable data communication between UAV 1 and a ground station or operator. The flight control board includes a microcontroller, a human-computer interaction module, a visual recognition module, a flight control data acquisition module, a path planning module, a pneumatic control module, a shape memory alloy control module, a gripper control module, a Kalman filter fusion module, a compensation algorithm module, and a safety monitoring module. Human-computer interaction module, used for information exchange between operators and the system; Visual recognition module, which is mainly used to identify and locate target objects; A flight control data acquisition module, used to obtain flight control data of the UAV 1; A pneumatic control module, used to control the extension and contraction of the pneumatic module 22 of the body segment module 2; a shape memory alloy control module, configured to control the shape of the first shape memory alloy spring and the second shape memory alloy spring; The clamping jaw control module is used to control the opening and closing of the clamping jaw module 3 and the clamping force of the clamping jaw module 3; Kalman filter fusion module, used to fuse flight control data; A compensation algorithm module is used to correct the deviation of the robotic arm caused by the posture change of the drone 1 based on the fused flight control data; A safety monitoring module monitors the air pressure of the drone 1, the spring temperature of the flexible robotic arm, and the clamping pressure of the gripper module 3 in real time; The microcontroller is connected to the human-computer interaction module, path planning module, pneumatic control module, shape memory alloy control module, gripper control module, Kalman filter fusion module, compensation algorithm module and safety monitoring module, receives data from each module and sends control instructions.

[0029] like Figure 7 As shown, the present invention also provides a control method for a UAV system based on a flexible robotic arm, which specifically includes the following steps: S1: obtaining the coordinate position of the target object; the specific method of obtaining the coordinate position of the target object includes the user inputting the target coordinates through the graphical user interface GUI and obtaining the three-dimensional coordinates of the object through the visual recognition module; S2: generating movement paths of the segment module 2 and the gripper module 3 of the flexible robotic arm based on the acquired coordinate position of the target object; S3: Based on the generated movement path, the drone 1 is controlled to drive the body segment module 2 and the gripper module 3 to move according to the movement path; at the same time, the attitude angle sent by the flight control system of the drone 1 is obtained, and the attitude angle includes the roll angle and the pitch angle; S4: Fusing the roll angle and pitch angle through Kalman filtering to obtain an estimated attitude angle; S5: Based on the data fused by the Kalman filter, the bending angle of the flexible robotic arm is compensated through the compensation algorithm module; S6: After the gripper module 3 grasps the target object, it estimates the mass of the target object and adjusts the stiffness of the segment module 2 and the clamping force of the gripper module 3 according to the weight of the target object.

[0030] In step S2, the specific steps of generating the moving path are as follows: S21: Solve the inverse kinematics equations based on the gradient projection method to obtain the bending angles of each joint of the flexible robotic arm; S22: The RRT* algorithm is used and Gaussian sampling strategy is introduced to generate collision-free paths in three-dimensional space.

[0031] In step S21, the specific steps of the gradient projection method include: S211: Establishing the Jacobian matrix of the flexible manipulator and determining the iteration parameters and constraint parameters, constructing the manipulator motion model based on the Jacobian matrix, and converting the displacement constraint of the gripper module 3 of the flexible manipulator into the joint velocity constraint of the segment module 2; The expression of the robot arm motion model is as follows: ; in, is the Jacobian matrix; is the joint velocity; is the joint angle vector of the robot arm, where k is the degree of freedom of the robot arm; is the speed of the end effector, is the three-dimensional linear velocity, is the three-dimensional angular velocity.

[0032] S212: constructing a hierarchical dynamic weighting factor matrix, and optimizing joint angle parameters through the hierarchical dynamic weighting factor matrix; S213: Based on the robot arm motion model, calculate the position error between the target object and the current gripper module 3 of the flexible robot arm, and convert the position error into the velocity of the gripper module 3 by gradient descent method; based on the robot arm motion model, the expression for calculating the position error between the target object and the current gripper module 3 of the flexible robot arm is as follows: ; is the position error vector of the gripper module 3; is the position of the target, is the current position of the gripper module 3; By using the gradient descent method, the expression for converting the position error into the terminal velocity is as follows: ; in, is the error gain coefficient, which controls the convergence speed.

[0033] S214: Calculate the weighted pseudo-inverse matrix based on the Jacobian matrix, and use the calculated weighted pseudo-inverse matrix to solve the joint velocity that satisfies the end constraint; and construct a weighted null space projection matrix, define the null space optimization target, generate the optimization direction, and guide the joint away from the extreme position. The specific process is as follows: First calculate the weighted pseudo-inverse matrix: ,in represents the Jacobian matrix, The specific value at the kth iteration, is the identity matrix, is the regularization parameter, and In order to suppress the numerical instability of singular points, It can avoid matrix singularity and ensure the existence of inverse matrix; when When , smaller singular values will be amplified, reducing the impact of noise on the solution and improving numerical stability. It can be regarded as a damping coefficient, balancing the end pose error and the joint motion amplitude; Then solve the joint velocity that satisfies the end constraint (special solution) .

[0034] Then construct the weighted null space projection matrix: ;

[0035] Define the null space optimization goal and generate the optimization direction , ,in, ; is the angle of the i-th joint; is the maximum value of the i-th joint angle; is the minimum value of the i-th joint angle; is the middle value of the i-th joint angle, guiding the joint away from the extreme position, is the objective function, is the objective function The local value at the kth iteration.

[0036] S215: Based on the joint velocity and zero space optimization that satisfy the end constraint, generate the joint velocity adjustment value of the body segment module 2, and convert the optimized joint velocity adjustment value into a joint position increment; Apply the joint position increment to the current joint vector and iteratively update the joint vector. The specific calculation process is as follows: Calculate the joint velocity delta:

[0037] is the joint velocity increment at the kth iteration, is the previous joint velocity; is the projection matrix from joint space to task space.

[0038] Convert joint velocity increments to joint position increments, , is the step size, which can be set as a constant or dynamically adjusted, such as Take the iterative time step.

[0039] Iteratively update the joint vector: ; in is the updated joint angle vector, is the current joint angle vector; is the joint position increment.

[0040] S216: Perform constraint projection on the updated joint vector to ensure that the joint state meets the constraint conditions; and calculate the updated position error to determine whether the termination condition is met or the maximum number of iterations is reached. If so, terminate the iteration and output the final joint angle vector; otherwise, return to step S214 to continue the iterative optimization process; The specific process is as follows: The updated joint vector is projected into the constraint space:

[0041] in, is the projection function, ; Calculate the updated position error ,like Or the maximum number of iterations is reached, the iteration is terminated, and the joint angle vector is output .

[0042] In step S212, the expression of the hierarchical dynamic weighting factor matrix W is as follows: The expression of the hierarchical dynamic weighting factor matrix W is as follows: ; in, is the weight of the first joint to the nth joint of the flexible manipulator, where n is the degree of freedom of the flexible manipulator, that is, the sum of the total number of joints of the flexible manipulator and the number of gripper modules 3, and The values decrease in sequence.

[0043] The origami structure 21 in the segment modules 2 of this embodiment supports bending in four directions (forward, backward, left, and right) and axial elongation. Each segment module 2 includes five joints, each with five degrees of freedom. The flexible robotic arm described in this embodiment includes three segment modules 2 and one gripper module 3. The three segment modules 2 provide a total of 15 joints, which, together with the origami gripper 31, provide a total of 16 joints, meaning a total of 16 degrees of freedom.

[0044] The 16 joints of the flexible manipulator can be divided into a core stabilization layer, a flexible adjustment layer, and an end-execution layer. The core stabilization layer includes the joints connecting the first segment module 2 and the drone 1, which are the key parts for carrying the main load and maintaining the stability of the manipulator. The flexible adjustment layer covers the joints of the second and third segment modules 2, which are mainly responsible for the posture adjustment and path planning of the manipulator in space. The end-execution layer, namely the joints of the gripper module 3, focuses on achieving precise grasping operations, and the corresponding layered dynamic weighting factor matrix W is For the core stabilization layer joint weights, To adjust the joint weights of the flexible layer, is the weight of the terminal execution layer.

[0045] The first segment module 2 connected to the drone 1 includes one pneumatic extension joint and four bending joints. The weight coefficient is set to a high value and is dynamically adjusted in real time according to the load and air pressure: The expression of the core stabilization layer joint weight is as follows: ; The basic weight is , the load sensitive item is , the stiffness sensitive term is Therefore, the dynamic adjustment formula is: ; m is the current load mass, is the maximum load of the flexible robotic arm; p is the current pneumatic muscle pressure; p is the current pneumatic muscle pressure, ; .

[0046] when When the "rack support mode" is triggered, Automatically increased to 2.5, prioritizing joint torque distribution to suppress platform vibration of UAV 1; When the air pressure (High stiffness state), The coefficient increases to 1.5, offsetting the deformation delay of the origami structure 21.

[0047] The flexible adjustment layer consists of 10 joints in the middle. The joint weights are lower than those in the core stability layer, but they are dynamically adjusted based on the distance between the joints and obstacles and the movement speed of the robot arm to achieve efficient path planning and obstacle avoidance. The dynamic weight formula is: ; The basic weight is , the obstacle adjustment item is , the motion speed adjustment item is , so the dynamic adjustment formula is:

[0048] is the distance from the i-th joint to the nearest obstacle, Safety distance; v is the current movement speed of the robot arm, is the maximum movement speed of the robot arm.

[0049] The end effector layer only contains the last joint, which is the opening and closing joint of the gripper module 3. Its weight is dynamically adjusted according to the shape, weight, and surface material of the grasped object to ensure stable grasping. Its dynamic weight formula is: ; The basic weight is , the gripping force adjustment item is , the friction coefficient adjustment term is , the object mass adjustment item is , so the dynamic adjustment formula is:

[0050] in Current grasping force, Maximum gripping force; is the coefficient of friction on the surface of the grasped object, is the maximum friction coefficient of common objects; is the mass of the grasped object; .

[0051] The smaller the surface friction coefficient of an object, the more unstable the grasping. Appropriately increasing the weight can further optimize the control torque and improve grasping stability. The greater the grasping force and the larger the object mass, the higher the grasping requirements. Increasing the weight can prevent the gripper from shaking during grasping and causing the object to fall, thereby achieving stable grasping of objects with different characteristics.

[0052] when (approaching maximum gripping force), the weight automatically increases to 2.0, locking the gripper joint movement to prevent overloading.

[0053] In step S23, the Gaussian sampling strategy is expressed as follows: ; is the distance between the flexible robotic arm and the obstacle, is the distance from the flexible robotic arm to the target object; and are the weight coefficients of obstacle distance and target distance respectively.

[0054] When the robotic arm approaches an obstacle, Reduce, The value of decreases, because the item is preceded by a positive sign, so the sampling probability This means that the probability of sampling near obstacles becomes smaller, thus guiding the robot arm to avoid dangerous areas.

[0055] For target distance , when the robot arm is far away from the target object, Increase, The value of increases, which also increases the sampling probability reduce.

[0056] But when the robot arm approaches the target object, As the value decreases, the sampling probability will increase accordingly, guiding the robotic arm to the location of the target object for sampling and path planning.

[0057] In the expression of Gaussian sampling strategy, when Beyond When the weight coefficient is low, the sampling probability decreases to a limited extent, and the robot arm is prone to deviate from the target point. To this end, two solutions are proposed: The first solution is to introduce a penalty factor , further reducing the sampling probability of the area beyond the threshold, the adjusted sampling probability formula is: ; when hour, , the formula remains unchanged; when hour, ,at this time The value of will be smaller than the original one, which will make the value of the entire exponential part smaller and the sampling probability Further reduce. Then dynamically adjust according to the excess ratio The value of , let the excess ratio be ,but The more the threshold is exceeded, the larger the penalty factor is, and the more obvious the sampling probability is reduced.

[0058] The second solution is to add a distance correction term to enhance the effect of the super-threshold distance on the sampling probability. The adjusted formula is: in The value is between (0.6, 1.0), and the middle value is generally taken. As the weight coefficient of the correction term, n is a positive integer (usually n=2 to enhance the nonlinear effect).

[0059] when hour, , the correction term has no effect; when When the correction A positive value will further reduce the exponential part, thereby reducing the sampling probability.

[0060] Furthermore, the Kalman filter in S4 is specifically implemented as follows: The IMU sensor provides accelerometer and gyroscope data. Through Kalman filtering, these two types of data are fused to obtain a more stable attitude angle estimate, thereby improving the compensation algorithm's response accuracy to attitude deviations.

[0061] The state variables of the Kalman filter are defined as the attitude angles of UAV 1, including roll angle and pitch angle, which are expressed as the state vector:

[0062] in, is the state vector at time k, is the roll angle, is the pitch angle. The state transfer equation calculates the attitude angle at the current moment based on the gyroscope angular velocity integral and introduces system noise Corrected, the expression is: ;

[0063] in, is the state transition matrix, is the control input matrix, is the angular velocity measured by the gyroscope, is the system noise term, Indicates that based on The data at the moment The predicted value of the state at the moment, Indicates that based on The data at the moment The optimal estimate of the state at the moment. Update the uncertainty matrix of the predicted state through covariance prediction , Reflects the credibility of attitude angle estimation, and its expression is as follows: ; in is the forecast covariance matrix, reflecting the uncertainty of the forecast state, for The covariance of .

[0064] Combined prediction covariance and the measurement noise covariance , calculate the Kalman gain , used to balance the weights of predicted values and measured values: ; in is the observation matrix, To measure the noise covariance, the measurement equation is provided by the accelerometer data and is expressed as:

[0065] in, is the measured value, is the observation matrix, is the measurement noise. Using the Kalman gain and measured values , combined with the prior state estimate Calculate the optimal state estimate at the current moment , the formula is: ; The residuals between the measured and predicted values Adjust the attitude angle estimation to suppress the influence of gyroscope drift and accelerometer noise. This step corrects the predicted value with the measured value to obtain a more accurate state estimate and update the covariance matrix: ; in The covariance matrix is updated to reflect the uncertainty of the current estimate, preparing for the next iteration. The Kalman filter calculates the optimal state estimate using a weighted fusion of prior estimates and observations. The update process consists of two steps: prediction and correction. By calculating the covariance matrix and optimizing the weight distribution, the final attitude angle value remains stable and accurate during dynamic changes.

[0066] After the attitude angle estimation is completed, the system detects the deviation. When the deviation exceeds 5°, the compensation algorithm module is triggered to perform feedforward compensation to adjust the body segment bending angle, so that the gripper module 3 at the end of the flexible manipulator maintains a stable state relative to the target object. The efficient fusion capability of the Kalman filter ensures high accuracy of attitude angle estimation and provides a reliable data foundation for subsequent dynamic compensation. Feedforward compensation is used to correct the manipulator arm deviation caused by the attitude change of the drone 1. The calculation formula is: ; is the compensation angle of the robot arm, is the feedforward compensation coefficient, with a value of 0.5° / ° (degrees per degree). and They are the roll angle and pitch angle of UAV 1, that is, the sum of the pitch angle and the roll angle increases by 1°, and the feedforward compensation angle will increase by 0.5°.

[0067] The compensation algorithm calculates the required feedforward adjustment amount based on the real-time attitude angle measured by the IMU sensor, so that the robotic arm can automatically correct the angle when the drone 1 tilts, maintaining the stability of the end relative to the target object.

[0068] After grasping the target object, in order to improve the stability of the flexible robotic arm, it is necessary to accurately identify the mass of the target object and adjust the stiffness of the pneumatic muscles based on the mass so that the flexibility and stiffness of the robotic arm match the characteristics of the target object and improve the grasping stability.

[0069] In step S6, the specific method of identifying the mass of the target object includes: Get the motor current of the flexible robotic arm when it is unloaded; Obtain the motor current when the flexible robotic arm performs a grasping action; Based on the two current data obtained, the estimated mass of the target object is obtained. The expression of the estimated mass of the target object is as follows: ; in, is the estimated mass of the target object, is the motor current when the robot arm performs the grasping action, is the motor current when there is no load, is the proportional coefficient of motor torque and current, is the acceleration due to gravity.

[0070] By calculating the load mass, it can determine whether the robotic arm has successfully grasped the target object, and adjust the flight attitude and robotic arm control strategy accordingly to ensure that the drone maintains balance during the grasping process.

[0071] After the load mass is obtained, another core variable related to pneumatic regulation is the historical air pressure data. The air pressure range is limited to This range is determined by the safe working range of the 0.125mm PET film origami structure used in the airbag of the body segment module 2. If the air pressure exceeds this range, the airbag may rupture or the origami structure 21 may be damaged, affecting the normal operation of the robot arm.

[0072] To minimize the matching error between the pneumatic stiffness of the robot arm and the load of the grasped object, the robot arm adjusts the internal air pressure according to the desired stiffness, and through precise step size and rate control, prevents overshoot or undershoot during the stiffness adjustment process, ensuring that the robot arm always maintains a suitable working state in a dynamic environment.

[0073] In order to match the aerodynamic stiffness of the robot arm with the load of the grasped object, the objective function is defined: ; in, is the current stiffness, is the expected stiffness; The desired stiffness and load mass The relationship is: ; This formula takes into account that as the load increases, the change in the required stiffness of the robot arm is not completely linear. By continuously adjusting the air pressure p, the current stiffness As close to the desired stiffness as possible , thereby minimizing the objective function .

[0074] In order to better describe the relationship between the current aerodynamic stiffness and the load mass, the observation equation is introduced: ; The regression vector , which includes different power terms of load mass and air pressure to reflect their complex effects on aerodynamic stiffness.

[0075] is the parameter to be estimated, , determined by the mechanical structure material properties and experimental data. In order to observe the noise, in actual measurement, due to the existence of various interference factors, the measured aerodynamic stiffness will have certain errors. Used to indicate the aerodynamic stiffness measurement error caused by various interference factors in actual measurement.

[0076] The observation equation describes the relationship between the current aerodynamic stiffness and the load mass, through the regression vector and the parameters to be estimated Load mass , air pressure Factors such as the load and the aerodynamic stiffness are linked together to provide a model basis for calculating the current aerodynamic stiffness.

[0077] Introducing data trade-off gain factors and data confidence covariance Iterative: ; ; in Forgetting Factor , According to the previous data confidence covariance iteration and the current regression vector To calculate, it is used to weigh the impact of new data and old data on parameter estimation. Data confidence covariance iterative iterative The forgetting factor gradually reduces the weight of old data, allowing the algorithm to better adapt to dynamic changes in the load while suppressing the impact of historical noise. As the robotic arm grasps different objects, the load mass will constantly change. The forgetting factor allows the algorithm to respond to these changes more quickly and adjust parameter estimates.

[0078] The calculation of the air pressure regulation is based on the objective function , data trade-off gain factor and data confidence covariance iteration The actual amount and method of adjusting the air pressure are determined by the results.

[0079] Target air pressure , by solving the objective function about The minimum value of is the ideal air pressure value that makes the aerodynamic stiffness close to the desired stiffness.

[0080] The actual air pressure regulation needs to consider the limitations of the mechanical structure, so a constrained recursive regulation formula is adopted:

[0081] in, Function means limiting the result to and To ensure that the air pressure is always within a safe range. The single adjustment step size is limited to: ; here, The function is used to determine the relationship between the target air pressure and the current air pressure. When the difference between the target air pressure and the current air pressure is greater than the maximum adjustment step When the pressure is less than the target pressure, the adjustment is performed according to the maximum adjustment step; otherwise, the adjustment is performed directly according to the difference between the target pressure and the current pressure; at the same time, the rate limit compensation needs to be considered. like , For the control period, the maximum rate Adjust to prevent damage to the mechanical structure caused by excessive pressure adjustment.

[0082] The specific steps of controlling the drone driving segment module 2 and the gripper module 3 to move according to the generated movement path in step S3 are as follows: S31: The pneumatic control module analyzes the movement path. According to the target extension of the body segment module 2, the Bang-Bang controller pre-inflates the air, and then the PID controller fine-tunes the air pressure. S32: The shape memory alloy control module analyzes the movement path, performs constant current heating based on the path planning results, and combines the temperature compensation algorithm to adjust the current through PWM to achieve bending angle control of the body segment module 2 and the gripper module 3; S33: The gripper control module analyzes the moving path and calculates the opening and closing degree and the clamping force required by the gripper module 3 according to the path planning result and the characteristics of the target object.

[0083] Among them, the hardware configuration of the pneumatic control module in step 31 ensures the precise driving and stable air supply of the segment module 2, and supports the efficient motion control of the flexible robotic arm.

[0084] The micro air pump is responsible for providing a stable airflow, which can meet the pneumatic requirements of the body segment module 2 under different working conditions.

[0085] The airflow is distributed by a 3D printed manifold. The internal channel design of the manifold can evenly distribute the airflow and support the simultaneous air supply of 4 sets of airbags, ensuring that multiple segment modules 2 maintain consistent air pressure changes during operation.

[0086] The air pressure sensor is connected to the body segment module 2 via a PU air tube. The real-time monitoring data of the air pressure sensor is used in the closed-loop control system to make the expansion and contraction movement of the pneumatic muscle more precise and controllable.

[0087] The solenoid valve group adopts 24VPWM control mode, which can quickly adjust the air flow cut-off state according to the control instructions to achieve dynamic control of the body segment module 2.

[0088] The specific implementation of the shape memory alloy temperature compensation algorithm in S32 is: Shape memory alloy springs are driven by electric current, which heats them up and causes a phase change, thereby controlling their bending. Changes in ambient temperature can affect the heating rate and ultimate deformation of the shape memory alloy, necessitating a temperature compensation algorithm to ensure consistent control under varying ambient conditions.

[0089] Temperature compensation algorithm based on ambient temperature The driving current is adjusted dynamically according to the change of the current. The calculation formula is: ; in, is the current increment after compensation, A setting of 0.05A / °C means that for every 1°C increase in ambient temperature, the drive current decreases by 0.05A to prevent overheating and ensure precise and controllable deformation of the shape memory alloy. As the ambient temperature decreases, the compensation current increases to allow the spring to reach the desired temperature and maintain a consistent bending angle. The heat dissipation coefficient factor reflects the degree of heat dissipation's influence on the shape memory alloy's heating process. A larger heat dissipation coefficient, h, means faster heat dissipation, requiring more drive current to compensate for heat loss and maintain effective heating of the shape memory alloy. It is the influencing factor of shape memory alloy aging. As the shape memory alloy is used more and more and time passes, its material properties will change, that is, aging will occur.

[0090] The specific method of the gripper control module in S33 is: When powered, the shape-memory alloy spring heats and contracts, causing the slide 322 to move along the rail 321. This generates tension in the attached traction rope, which in turn forces the fingertips of the origami gripper 31 to close. When the current is removed, the spring cools and returns to its original shape, the traction rope relaxes, and the fingertips of the origami gripper 31 unfold. By controlling the current input, the degree of closure of the origami gripper 31 can be adjusted in real time to accommodate objects of varying sizes and shapes.

[0091] The fingertips of the origami gripper 31 of the gripper module 3 are integrated with a four-channel FSR sensor matrix for real-time monitoring of gripping force. Each FSR sensor has a range of 0-10N and can respond to pressure changes in less than 10ms, providing rapid feedback on gripping force. The sensors are distributed in different areas of the fingertips of the origami gripper 31, enabling it to sense the force acting on the target object and provide accurate tactile feedback data. Once this data is input into the control system, the gripping force can be dynamically adjusted to prevent damage to the target object due to excessive gripping force, or grasping failure due to insufficient gripping force.

[0092] During the grasping process, the current input is adjusted in real time based on the data fed back by the FSR sensor to ensure that the origami gripper 31 can firmly clamp the object, and adjustments are made when the force is abnormal, thereby improving the operating accuracy and safety of the flexible robotic arm.

[0093] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A UAV system based on a flexible robotic arm, characterized by: include: A flexible robotic arm and a drone, wherein the drone is connected to the flexible robotic arm; The flexible robotic arm includes several segment modules and gripper modules, and adjacent segment modules are connected in series via support columns. The segment modules include an origami structure, a pneumatic module, and a first shape memory alloy spring. The pneumatic module is embedded in the origami structure to control the length of the origami structure. The origami structure includes multiple joints, and the first shape memory alloy spring includes multiple springs, each of which is arranged on the outside of the origami structure to adjust the bending angle of the joint. A gripper module is provided on the segment module at the end of the flexible robotic arm, and includes an origami gripper, a sliding assembly, a traction device, and a second shape memory alloy spring; the sliding assembly is provided on the inner side of the origami gripper, the traction device is provided between the sliding assembly and the gripping end of the origami gripper, and the second shape memory alloy spring is provided on the outer side of the origami gripper for controlling the opening and closing of the origami gripper; The UAV includes a flight control system, which is used to control the flight of the UAV and adjust the working states of the body module and the gripper module.

2. The UAV system based on a flexible robotic arm according to claim 1, characterized in that: The flight control system includes a flight control board, an inertial measurement unit (IMU), a global positioning system (GPS), a barometer, and a communication module. The inertial measurement unit (IMU), the global positioning system (GPS), and the barometer are all connected to the flight control board and are used to monitor the attitude angle, position, and pressure altitude of the UAV, respectively. The communication module is connected to the flight control board and is used to realize data communication between the UAV and a ground station or an operator. The flight control board includes a microcontroller, a human-computer interaction module, a visual recognition module, a flight control data acquisition module, a path planning module, a pneumatic control module, a shape memory alloy control module, a gripper control module, a Kalman filter fusion module, a compensation algorithm module, and a safety monitoring module. Human-computer interaction module, used for information exchange between operators and UAV systems; Visual recognition module, which is mainly used to identify and locate target objects; Flight control data acquisition module, used to obtain the flight control data of the drone; A pneumatic control module, used to control the extension and contraction of the pneumatic module of the segment module; a shape memory alloy control module, configured to control the shape of the first shape memory alloy spring and the second shape memory alloy spring; A gripper control module, used to control the opening and closing of the gripper module and the gripping force of the gripper module; Kalman filter fusion module, used to fuse flight control data; The compensation algorithm module is used to correct the deviation of the robotic arm caused by the change of the drone's posture based on the fused flight control data; Safety monitoring module, which monitors the drone's air pressure, the spring temperature of the flexible robotic arm, and the gripping pressure of the gripper module in real time; The microcontroller is connected to the human-computer interaction module, path planning module, pneumatic control module, shape memory alloy control module, gripper control module, Kalman filter fusion module, compensation algorithm module and safety monitoring module, receives data from each module and sends control instructions.

3. A control method for a UAV system based on a flexible robotic arm according to claim 2, characterized in that: The specific steps include: S1: obtaining the coordinate position of the target object; the specific method of obtaining the coordinate position of the target object includes the user inputting the target coordinates through the graphical user interface GUI and obtaining the three-dimensional coordinates of the object through the visual recognition module; S2: generating movement paths of the segment module and the gripper module of the flexible robotic arm based on the acquired coordinate position of the target object; S3: Based on the generated movement path, the drone driving body module and the gripper module are controlled to move according to the movement path; at the same time, the attitude angle sent by the flight control system of the drone is obtained, and the attitude angle includes the roll angle and the pitch angle; S4: Fusing the roll angle and pitch angle through Kalman filtering to obtain an estimated attitude angle; S5: Based on the data fused by the Kalman filter, the bending angle of the flexible robotic arm is compensated through the compensation algorithm module; S6: After the gripper module grasps the target object, it estimates the mass of the target object and adjusts the stiffness of the segment module and the clamping force of the gripper module according to the weight of the target object.

4. The control method of a UAV system based on a flexible robotic arm according to claim 3, characterized in that: In step S2, the specific steps of generating the moving path are as follows: S21: Solve the inverse kinematics equations based on the gradient projection method to obtain the bending angles of each joint of the flexible robotic arm; S22: The RRT* algorithm is used and Gaussian sampling strategy is introduced to generate collision-free paths in three-dimensional space.

5. The control method of a UAV system based on a flexible robotic arm according to claim 4, characterized in that: In step S21, the specific steps of the gradient projection method include: S211: Establishing a Jacobian matrix of the flexible manipulator and determining iteration parameters and constraint parameters, constructing a manipulator motion model based on the Jacobian matrix, and converting the displacement constraint of the gripper module of the flexible manipulator into the joint velocity constraint of the segment module; S212: constructing a hierarchical dynamic weighting factor matrix, and optimizing joint angle parameters through the hierarchical dynamic weighting factor matrix; S213: Calculating the position error between the target object and the gripper module of the current flexible robotic arm based on the robotic arm motion model, and converting the position error into the velocity of the gripper module by using a gradient descent method; S214: Calculate a weighted pseudo-inverse matrix based on the Jacobian matrix, and use the calculated weighted pseudo-inverse matrix to solve the joint velocity that satisfies the end constraint; and construct a weighted null space projection matrix, define the null space optimization target, generate an optimization direction, and guide the joint away from the extreme position; S215: Based on the joint velocity and zero space optimization that satisfy the end constraint, generate the joint velocity adjustment of the body segment module, convert the optimized joint velocity adjustment into a joint position increment; apply the joint position increment to the current joint vector, and iteratively update the joint vector; S216: Perform constraint projection on the updated joint vector to ensure that the joint state meets the constraint conditions, and calculate the updated position error to determine whether the termination condition is met or the maximum number of iterations is reached. If so, terminate the iteration and output the final joint angle vector; otherwise, return to step S214 to continue the iterative optimization process.

6. The control method of a UAV system based on a flexible robotic arm according to claim 5, characterized in that: The expression of the robot arm motion model in step S211 is as follows: ； in, is the Jacobian matrix; Joint velocity; is the joint angle vector of the robot arm, where k is the degree of freedom of the robot arm; is the speed of the end effector, is the three-dimensional linear velocity, is the three-dimensional angular velocity.

7. The control method of a UAV system based on a flexible robotic arm according to claim 5, characterized in that: In step S213, based on the robot arm motion model, the expression for calculating the position error between the target object and the gripper module of the current flexible robot arm is as follows: ； is the position error vector of the gripper module; is the position of the target, is the current position of the gripper module; By using the gradient descent method, the expression for converting the position error into the terminal velocity is as follows: ； in, is the error gain coefficient, which controls the convergence speed.

8. The control method of a UAV system based on a flexible robotic arm according to claim 5, characterized in that: In step S212, the expression of the hierarchical dynamic weighting factor matrix W is as follows: ； in, is the weight of the first joint to the nth joint of the flexible manipulator, where n is the degree of freedom of the flexible manipulator, that is, the sum of the total number of joints of the flexible manipulator and the number of gripper modules, and The values decrease in sequence.

9. The control method of a UAV system based on a flexible robotic arm according to claim 4, characterized in that: In step S5, based on the data fused by the Kalman filter, the bending angle of the flexible manipulator is compensated by the compensation algorithm module; wherein, the relationship between the compensation angle of the flexible manipulator and the posture angle is expressed as follows: ; in, is the compensation angle of the robot arm, and are the roll angle and pitch angle of the drone, is the feedforward compensation coefficient.

10. The control method of a UAV system based on a flexible robotic arm according to claim 4, characterized in that: In step S6, the specific method of identifying the mass of the target object includes: Get the motor current of the flexible robotic arm when it is unloaded; Obtain the motor current when the flexible robotic arm performs a grasping action; Based on the two current data obtained, the estimated mass of the target object is obtained. The expression of the estimated mass of the target object is as follows: ； Where m is the estimated mass of the target object, is the motor current when the robot arm performs the grasping action, is the motor current when there is no load, k is the proportional coefficient of the motor torque and current, and g is the acceleration due to gravity.

Images