A Modeling Method for Dexterous Arm Robots
By using a modeling method that simulates the structure of the human arm and hand, and designing a controller using a state transfer strategy, the problems of accuracy and collaborative ability in modeling dexterous arm robots were solved. This enabled synchronous and coordinated control of the robotic arm and the multi-finger system, and reduced computational costs.
Patent Information
- Application Number
- CN202410953810.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2044-07-16
AI Technical Summary
Existing dexterous arm robot modeling methods are not precise enough, lack simplicity and universality, have poor collaboration capabilities between robotic arms and multi-finger systems, and have a high computational burden and are time-consuming for model training.
By adopting a modeling method that simulates the structure of the human arm and hand, and designing the controller of the robotic arm and multi-finger system through a state transfer strategy, we can achieve synchronous and coordinated control of the arm and hand, and decouple the robotic arm and multi-finger system from dynamics to reduce computational costs.
It achieves efficient coordinated control of the dexterous arm robot system, reduces computational burden, and improves system flexibility and simplicity.
Smart Images

Figure CN118682765B_ABST
Abstract
Description
Technical Field
[0001] This invention primarily relates to the field of robotics, particularly modeling methods and control techniques for dexterous arm robots. This technical field encompasses mechanical design, control engineering, and theoretical mathematics. The modeling method proposed in this invention combines knowledge from these fields to describe the kinematics, dynamics, perception, and control aspects of dexterous arm robots using simplified mathematical expressions, facilitating simulation and control algorithm design. Background Technology
[0002] In recent years, dexterous arm robot systems have been widely used in various fields, including manufacturing, healthcare, logistics and warehousing, aerospace, and military applications, attracting widespread attention from researchers and scholars worldwide. Currently, the design of dexterous arm robots has become a research hotspot and has been the subject of in-depth study.
[0003] 1. Currently, there are two main modeling methods for dexterous arm robots: 1) theoretical mathematical modeling, which analyzes the structural characteristics of the dexterous arm system and analyzes its kinematic and dynamic models; 2) model training methods, such as deep learning, reinforcement learning, neural networks, etc., which train the system model by sampling and testing data, and design and operate the corresponding controller.
[0004] 2. Currently, there are three major problems and shortcomings in the modeling of dexterous arm robots:
[0005] First, the modeling methods proposed in existing literature are not precise enough, and lack simplicity and universality. This is because these modeling methods are only applicable to specific system structures and do not differentiate between robotic arms and multi-finger systems.
[0006] Secondly, existing modeling methods for dexterous hand robots are insufficient to support the robotic arm and multi-finger system in completing the same objectives and some complex operational tasks. This is because there are no separate controllers designed for the robotic arm and multi-finger system, resulting in poor collaboration between the robotic arm and multi-finger system and an inability to perform synchronous and coordinated control.
[0007] Third, model training methods require high computing power and incur a heavy computational burden. Methods such as deep learning, reinforcement learning, and neural networks require significant computational resources for training, and the accuracy of these models is limited by training time and intensity, making them unsuitable for timely engineering applications. Furthermore, model training necessitates the collection of large amounts of data samples, a time-consuming process.
[0008] The existing technical problems are as follows:
[0009] 1. Existing modeling methods are not accurate enough, lack simplicity and universality, and fail to distinguish between robotic arms and multi-finger systems;
[0010] 2. In existing dexterous arm robot systems, the robotic arm and multi-finger system have poor collaboration capabilities and cannot perform synchronous and coordinated control, which makes it difficult for the robotic arm and multi-finger system to complete the same goal and some complex operation tasks.
[0011] 3. The model training method in the modeling method of dexterous arm robot has the problem of high computing power requirements, time-consuming process and large computational burden. Summary of the Invention
[0012] The technical objective of this invention is to propose a modeling method for dexterous arm robots. Addressing the problems mentioned in the background and existing technologies, this invention proposes a modeling method suitable for high-degree-of-freedom dexterous arm systems by simulating the natural movements and functions of the human arm and hand. The dynamic model obtained by this method has a mathematical expression similar to that of the human arm and hand. Simultaneously, it employs an arm-hand state transfer strategy. By designing state transfer equations between the robotic arm system and the multi-finger system, the transfer of states between the two systems can be achieved. Controllers for the robotic arm and multi-finger system are designed independently according to control requirements. On one hand, this strategy enables synchronized operation between the arm and hand, making control more coordinated. On the other hand, compared with traditional integrated arm-hand systems, this invention, through the state transfer strategy, achieves dynamic decoupling of the robotic arm and multi-finger system, significantly reducing computational costs, making the system more flexible, and reducing dependence on high-performance computing resources.
[0013] The technical solution adopted in this invention is a modeling method for a dexterous arm robot, which specifically includes the following steps:
[0014] Inspired by the structure of the human arm and fingers, the following simplified system model is proposed: The dexterous arm robot includes a robotic arm system and a multi-finger system, with the multi-finger system mounted on the robotic arm system; Both the robotic arm system and the multi-finger system consist of links and rotary joints, with the links mounted on the rotary joints to form dexterous arm robot units; Multiple dexterous arm robot units are connected in series and assembled using a reference arm-hand structure; The rotary joints are driven by motors.
[0015] First, the robotic arm system is modeled. From the Lagrange equations, we know that when the generalized coordinates q of the n-DOF rigid robotic arm system... i Once i = 1, ..., n is determined, the dynamic equations of the robotic arm system are defined as functions of the generalized coordinates:
[0016]
[0017] in and q represents the kinetic and potential energy of the robotic arm system, respectively. i It is a generalized coordinate system. It is a generalized velocity, where t is time and τ is velocity. i It is a generalized force.
[0018] The dynamic equations from coordinate system i to coordinate system i+1 are established using the Denavit-Hartenberg method and the following coordinate transformation:
[0019]
[0020] Where θ i α i a i d i Denotes the Denavit-Hartenberg (DH) parameters of the i-th joint, in equation (2) This represents the position vector of coordinate system i+1 relative to coordinate system i. This represents the rotation matrix from coordinate system i to coordinate system i+1. Therefore, the position and orientation vectors of coordinate system n relative to the base coordinate system are obtained by multiplying the coordinate transformation matrix from rotation joint 1 to rotation joint n, i.e.
[0021] According to the dynamic equation (2), the kinetic energy of the i-th link and the motor installed on the corresponding rotary joint are expressed as follows:
[0022]
[0023] Where m li and m mi Let the masses of the i-th link and the motor be represented respectively. Let represent the inertia tensor of the i-th link relative to coordinate system i. Let R represent the inertia tensor of the i-th motor relative to coordinate system i-1. i R represents the rotation matrix of coordinate system i relative to the base coordinate system. mi This represents the rotation matrix of coordinate system i-1 relative to the base coordinate system, and the velocity vector of link i relative to the rotary joint. The Jacobian matrix is represented as
[0024]
[0025] in
[0026]
[0027] z j-1 p represents the z-axis vector of coordinate system j-1. li and p j-1These represent the position vectors of coordinate system i and coordinate system j-1 relative to the base coordinates, respectively.
[0028] Similarly, the velocity vector of motor i relative to the rotary joint The Jacobian matrix is represented as
[0029]
[0030] in
[0031]
[0032] p mi k represents the position vector of motor i relative to the base coordinate system. i Indicates the reduction ratio, z mi Let p represent the z-axis vector of coordinate system i-1. A motor is installed at the center of each rotary joint. Since the position of motor i coincides with the origin of coordinate system i-1, we can further obtain p. mi =p i-1 R mi =R i-1 ,
[0033] The potential energy of link i and motor i is expressed as:
[0034]
[0035] Where g0 represents the gravitational acceleration vector of the i-th coordinate system relative to the base coordinate system, and if the y-axis is the perpendicular coordinate axis of the base coordinate system, then...
[0036] Based on the Jacobian matrix above, the dynamic model of the robotic arm system can be expressed as follows:
[0037]
[0038] in
[0039]
[0040] B(q)=(b ij ) n×n , It is the control input signal of the robotic arm system.
[0041] Based on the simplified model above, namely equation (3), the multi-finger system is regarded as a linkage structure similar to a robotic arm system. Following this modeling approach, the modeling method based on the Lagrange equation and DH parameters can be applied to the multi-finger system. Assuming that the degree of freedom of each finger is m, the following dynamic model of the multi-finger system is obtained by following the same modeling process:
[0042]
[0043] in Let be the inertia matrix, Coriolis force matrix, and gravity vector of the i-th finger system, respectively. and It is the set of real matrix.
[0044] It is the angular displacement vector of the i-th finger system. It is the control input of the i-th finger system.
[0045] For the finger system, the controller controls q. i1 ,q i2 ,…,q im The rotational joint state of the finger system is composed of the known state of the robotic arm system and the rotational joint states of each finger in the multi-finger system. This is achieved by transmitting the angular displacement, angular velocity, and angular acceleration of the robotic arm system under control to the multi-finger system in real time. Then, based on the output state of the robotic arm system, a controller for the multi-finger system is designed to ultimately control the dexterous arm system to complete the corresponding operation.
[0046] Compared with existing technologies, this invention is inspired by the actual human arm and hand skeletal structure and movement mechanism, and proposes a general modeling method for general high-degree-of-freedom dexterity arm and hand systems. The obtained dynamic model is simple and practical, and has a similar mathematical expression.
[0047] Based on the established dynamic model, an arm-hand state transfer strategy is adopted, and controllers for the robotic arm and multi-finger system are designed according to the control requirements to achieve synchronized and more coordinated control of the arm and hand.
[0048] Compared to integrated arm-hand systems, this invention decouples the robotic arm and multi-finger system dynamically, reducing computational costs. Attached Figure Description
[0049] Figure 1 This is a simplified structural model diagram of a five-finger dexterity arm and hand system.
[0050] Figure 2 This is the control logic diagram of the dexterous arm system.
[0051] Figure 3 This is a simplified structural diagram of the five-finger dexterity arm and hand system.
[0052] Figure 4 It is a diagram of the trajectory of the joint in motion.
[0053] Figure 5It is a control signal trajectory diagram. Detailed Implementation
[0054] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0055] Figure 1 This is a simplified structural model diagram of a five-finger dexterity arm and hand system.
[0056] Figure 2 This is the control logic diagram of the dexterous arm system.
[0057] Figure 3 This is a simplified structural diagram of the five-finger dexterity arm and hand system.
[0058] like Figure 1 and Figure 3 The robotic arm system consists of links and joints. The joints are divided into rotary joints and translating joints. For a robotic arm, it only includes rotary joints. The motor that drives the rotary joints is mounted at the center of the joint. A multi-finger system can be regarded as multiple robotic arm systems.
[0059] like Figure 2 In the control logic of the dexterous arm system, the controller τ of the robotic arm is designed first. Then, the state transfer strategy of the arm-hand system is adopted to design the state transfer equation between the robotic arm system and the multi-finger system. Finally, the controller of the multi-finger system is designed.
[0060] This method includes the following steps:
[0061] Step 1: Design a simplified structure for the dexterous arm system. First, based on the given control requirements, determine the required degrees of freedom, number of fingers, and other parameters of the robotic arm in the dexterous arm system, and obtain its simplified structural model:
[0062] Step 2: Based on the simplified model described above, obtain the DH parameters θ for each joint. i α i a i d i As shown below:
[0063] Table 1. DH parameters of the five-finger dexterity arm system
[0064]
[0065] Step 3: Based on the DH parameters of the robotic arm and each finger, calculate the inertia matrix, Coriolis force matrix, and gravity vector of the robotic arm and multi-finger system using equation (4).
[0066] Robotic arm dynamics model:
[0067]
[0068] The dynamic model of the i-th finger:
[0069]
[0070] in
[0071] Step 4: Design the controller τ for the robotic arm;
[0072] Step 5: Transmit the status of the robotic arm to the multi-finger system.
[0073] Step 6: Design the controller τ for each finger based on the output state of the robotic arm. fi .
[0074] like Figure 4 , Figure 5 The diagram illustrates the changes in joints and control signals over time in the model established by this invention. Its advantages can be summarized as follows:
[0075] (1) Fast response: As can be seen from the figure, within the first few seconds, all the generalized velocity components quickly change from the initial state to the new state, indicating that the control system has a fast response capability.
[0076] (2) Robustness: The fluctuation of the joint state over time is large at the beginning, but then it decreases rapidly and stabilizes at a level close to zero, showing that the system has good robustness and can effectively reduce oscillation.
[0077] (3) Stability: The joint state and control signal tend to stabilize in about 20 seconds, indicating that the control system can control the robotic arm and maintain it in the desired state.
Claims
1. A method for modeling a dexterous arm robot, characterized in that, Specifically, the steps include the following: First, the robotic arm system is modeled. From the Lagrange equations, we know that when the generalized coordinates of the n-DOF rigid robotic arm system... Once determined, the dynamic equations of the robotic arm system are defined as functions of the generalized coordinates: in , These represent the kinetic and potential energy of the robotic arm system, respectively. It is a generalized coordinate system. It is a generalized velocity, where t is time. It is a generalized force; The dynamic equations from coordinate system i to coordinate system i+1 are established using the Denavit-Hartenberg method and the following coordinate transformation: in Denotes the Denavit-Hartenberg (DH) parameters of the i-th joint, in equation (2) , , , ; This represents the position vector of coordinate system i+1 relative to coordinate system i. Represents the rotation matrix from coordinate system i to coordinate system i+1; Therefore, the position and orientation vectors of coordinate system n relative to the base coordinate system are obtained by multiplying the coordinate transformation matrix from rotation joint 1 to rotation joint n, i.e. ; According to the dynamic equation (2), the kinetic energy of the i-th link and the motor installed on the corresponding rotary joint are expressed as follows: in and Let the masses of the i-th link and the motor be represented respectively. Let represent the inertia tensor of the i-th link relative to coordinate system i. Let represent the inertia tensor of the i-th motor relative to coordinate system i-1. Let represent the rotation matrix of coordinate system i relative to the base coordinate system. This represents the rotation matrix of coordinate system i-1 relative to the base coordinate system, and the velocity vector of link i relative to the rotary joint. The Jacobian matrix is represented as in This represents the z-axis vector of coordinate system j-1. and These represent the position vectors of coordinate systems i and j-1 relative to the base coordinates, respectively. Similarly, the velocity vector of motor i relative to the rotary joint The Jacobian matrix is represented as in This represents the position vector of motor i relative to the base coordinate system. Indicates the reduction ratio. Let i represent the z-axis vector of coordinate system i-1; a motor is installed at the center of each rotary joint. Since the position of motor i coincides with the origin of coordinate system i-1, we can further obtain... , ; The potential energy of link i and motor i is expressed as: in Let represent the gravitational acceleration vector of the i-th coordinate system relative to the base coordinate system. If the y-axis is perpendicular to the base coordinate system, then... ; Based on the Jacobian matrix above, the dynamic model of the robotic arm system can be expressed as follows: in , , , It is the control input signal of the robotic arm system; According to equation (3), the multi-finger system is regarded as a linkage structure similar to a robotic arm system. Based on the Lagrange equation and the DH parameter modeling method, the degree of freedom of each finger is m, and the following dynamic model of the multi-finger system is obtained: in , , Let be the inertia matrix, Coriolis force matrix, and gravity vector of the i-th finger system, respectively. It is the set of real number matrices; It is the angular displacement vector of the i-th finger system. It is the control input of the i-th finger system; (1) For the finger system, the controller controls The rotational joint state of the finger system is composed of the known state of the robotic arm system and the rotational joint states of each finger in the multi-finger system. This is achieved by transmitting the angular displacement, angular velocity, and angular acceleration states of the robotic arm system under control to the multi-finger system in real time. Then, based on the output state of the robotic arm system, a controller for the multi-finger system is designed to ultimately control the dexterous arm system to complete the corresponding operation.
2. The method for modeling a dexterous arm robot according to claim 1, characterized in that, The dexterous arm robot includes a robotic arm system and a multi-finger system, with the multi-finger system mounted on the robotic arm system. Both the robotic arm system and the multi-finger system consist of links and rotary joints. The links are mounted on the rotary joints to form dexterous arm robot units. Multiple dexterous arm robot units are connected in series and assembled into a dexterous arm robot through a reference arm-hand structure. The rotary joints are driven by motors.
Citation Information
Patent Citations
Fusion system of mechanical arm and dexterous hand and motion control method
CN114714327A
Dexterous hand and mechanical arm reinforcement learning cooperative control method based on heuristic trajectory
CN117733850A
