A complex underactuated robot full-dynamics decoupling servo control system and method

Abstract

The application relates to the technical field of mobile robots and discloses a complex under-actuated robot full-dynamics decoupling servo control system and method, wherein the control system comprises a calibration target module, a feature sensing module, a data processing module, a position control module and a posture control module; the control method comprises the following steps: energy equations and potential energy equations are established by analyzing the force conditions of each moving component of a robot body; a robot dynamics model is established by substituting a Lagrange function into a Lagrange equation; a conversion equation representing the conversion relationship between robot control variables and self-state variables is established by analyzing the mapping relationship between the robot control variables and the self-state variables; and a control model is obtained based on the conversion equation, a control algorithm is designed, and a control signal is generated. On the basis of ensuring that the robot self-model can fully reflect the dynamics characteristics of the robot, the posture and position dynamics of the robot are decoupled, the accuracy and stability of the body control are improved, and the maneuverability and sensitivity of the robot are effectively exerted.

Description

Technical Field

[0001] This invention relates to the field of mobile robot technology, specifically to a fully dynamic decoupled servo control system and method for a complex underactuated robot. Background Technology

[0002] With the continuous development of social productivity, people's demand for robots to quickly, autonomously, and accurately complete various set tasks in complex scenarios is constantly increasing, and robots are playing an increasingly important role in daily production and life. To design autonomous robots, it is necessary to develop automated algorithms in four modules: perception, planning, decision-making, and control. This enables the robot to actively acquire information about its surrounding environment, calculate its position information based on its onboard sensors, complete task decisions and path planning based on the above perception information, and finally, the control algorithm changes the robot's posture and position to complete the control task. Therefore, it can be seen that the design of the robot's control algorithm is the foundation and fundamental guarantee for achieving autonomous tasks. Furthermore, the accurate establishment and decoupling of complex robot models will greatly facilitate the design of control algorithms and improve their efficiency and stability.

[0003] Since complex robot dynamics models often have nonlinear, strongly coupled, and underactuated characteristics, current robot control algorithm design mainly falls into two categories: considering only the kinematic model and directly decoupling translation and rotation.

[0004] Control methods that only consider the robot's kinematics model typically assume that the robot's internal posture control module can track the control signals generated by the external position control module at a sufficiently fast speed, which limits their application in real-world scenarios. Furthermore, because kinematic models cannot characterize the force relationship between the robot and its external environment, while they can achieve simple position control, they also lead to problems such as low control efficiency and target loss due to actuator saturation. Therefore, robot control methods that only consider kinematic models are insufficient to meet today's societal demands for robots to safely and autonomously complete pre-defined tasks in complex environments.

[0005] Meanwhile, in the design of robot control algorithms, the method of directly decoupling translational and rotational motions and then designing control algorithms separately based on the robot's posture and position dynamics models often omits coupling terms and higher-order terms in the complete robot dynamic equations. While this method yields a relatively simple robot dynamic model and simplifies the design process of the corresponding control algorithm, the simplified dynamic equations fail to fully reflect the robot's dynamic characteristics. Control algorithms designed based on this model cannot guarantee the full utilization of the robot's maneuverability, inevitably reducing the robot's operating speed and thus affecting the system's execution efficiency.

[0006] Because complex robot dynamics models often exhibit nonlinearity, strong coupling, and underactuation, existing technologies for designing robot control algorithms generally only consider their kinematic models or directly omit coupling terms in the robot's dynamic equations, decoupling translational and rotational components. Directly using kinematic equations to design robot control algorithms can only achieve position and velocity control, making it difficult to achieve precise control of the robot's actuators. Furthermore, the method of directly decoupling by omitting coupling and higher-order terms in the robot's dynamic equations fails to fully reflect the robot's dynamic characteristics. In summary, existing technologies cannot fully utilize the robot's maneuverability, resulting in slow movement speeds and low overall system efficiency. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a fully dynamic decoupled servo control system and method for complex underactuated robots. While ensuring that its own model fully reflects the dynamic characteristics of complex robots, it decouples the robot's posture and position dynamics. This will greatly facilitate the design of robot control algorithms based on dynamic models, improve the accuracy and stability of robot control, effectively leverage the mobility and sensitivity of various types of robots, expand their working range, and improve their task execution efficiency.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0009] A fully dynamic decoupled servo control system for a complex underactuated robot includes: a calibration target module, a feature perception module, a data processing module, a position control module, and an attitude control module;

[0010] The calibration target module, which serves as the calibration target signal source for the robot's onboard sensors, can provide perception information, as well as its own absolute position and attitude information.

[0011] The feature perception module acquires the perception information of the calibrated target and the surrounding environment information through its environmental perception unit, which includes various airborne sensors that enable the robot to acquire environmental perception capabilities.

[0012] The data processing module accurately identifies and acquires the perception information P of the calibration target from the surrounding environment information transmitted by the feature perception module by running data processing algorithms corresponding to the robot's onboard sensors. At the same time, the data processing module combines the state information S of the position control module and the attitude control module, and calculates the relationship matrix R between the calibration target perception information P and the robot control module state information S by running a dynamic decoupling algorithm. This enables the decoupled control of translation and rotation state variables in the robot's dynamic model and generates control signals.

[0013] The position control module receives control signals transmitted by the data processing module, converts the control signals into corresponding position control signals, and transmits them to the attitude control module.

[0014] The attitude control module receives the position control signal transmitted by the position control module, converts the position control signal into a corresponding attitude control signal, and controls the robot actuator to achieve the corresponding attitude and position changes through the attitude control signal.

[0015] A control method for a complex underactuated robot's fully dynamic decoupled servo control system includes the following steps:

[0016] Step 1: By analyzing the forces acting on each moving part of the robot, establish the kinetic energy equation T for each moving part of the robot. i and potential energy equation V i , where i = 1, 2, 3...n, and n represents the number of moving parts of the robot;

[0017] Step 2: By using the Lagrange function L = ∑(T) i -V i Substituting i = 1, 2, 3...n into the Lagrange equation, a robot dynamics model is established;

[0018] Step 3: By analyzing the mapping relationship between the robot's control variables and its own state variables, establish a transformation equation representing this transformation relationship; where the robot's control variables include the perception information P of the calibration target, and the robot's own state variables include the state information S of the position control module and the attitude control module;

[0019] Step 4: Based on the transformation equation, obtain the control model for robot rotation and translation decoupled, and generate the control signal.

[0020] Furthermore, step three specifically includes:

[0021] S31. Select the robot's onboard sensors according to the robot's specific target task. Onboard sensors include vision sensors, ultrasonic sensors, or lidar.

[0022] S32. Select the appropriate sensing and calibration target according to the type of airborne sensor;

[0023] S33. The robot position controller and attitude controller control the robot to change its own attitude and position, while measuring the perception information of the corresponding calibrated target based on the airborne sensors.

[0024] S34. Based on the robot's position information, posture information, and the perception information of the calibration target, derive the relationship matrix R between the rate of change of the perception information of the calibration target dP / dt and the rate of change of the robot's own state information dS / dt, and obtain the transformation equation dP / dt=RdS / dt.

[0025] in x, y, and z represent the robot's displacement along the x-axis, y-axis, and z-axis of the inertial coordinate system, respectively. θ and ψ represent the robot's roll, pitch, and yaw angles, respectively, and T represents transpose.

[0026] Furthermore, step four specifically includes:

[0027] S41. Design the position and attitude control algorithm for the control model based on the transformation equation:

[0028] f = K f (m, g)Re P τ=K τ (J, ω)Re ω ;

[0029] Where f is the robot's position control signal, τ is the robot's attitude control signal, and K f K τ This is an adjustable parameter matrix, where m is the robot mass, g is the gravitational acceleration, J is the moment of inertia, ω is the robot angular velocity, and e is the angular velocity. P e ω All are state tracking errors;

[0030] S42. Based on the position and attitude control algorithm, decouple the translation and rotation state variables in the robot dynamics model and generate the control signal.

[0031] Compared with the prior art, the beneficial technical effects of the present invention are:

[0032] This invention changes the existing method of directly using kinematic models or omitting coupling and higher-order terms in the dynamic model in the design of complex robot control algorithms. Instead, it adopts a complete dynamic model derived using the Lagrange method. The complete robot dynamic model often has strong coupling characteristics. The position control module and attitude control module control the robot to perform corresponding position and attitude changes. The onboard perception module transmits the changes in features in the external calibration target module to the data processing module. Then, based on the change information of the robot's position, attitude and calibration target, the relationship matrix between the rate of change of the calibration target and the rate of change of the robot's own state is derived, and a transformation equation is established. Finally, the robot control algorithm is designed through the transformation equation, thereby obtaining a decoupled robot dynamic control model and control method.

[0033] The method and system provided by this invention decouple the robot's posture and position dynamics while ensuring that the robot's own model fully reflects its dynamic characteristics. This greatly facilitates the design of robot control algorithms based on dynamic models, improves the accuracy and stability of robot control, effectively leverages the mobility and sensitivity of various types of robots, expands their working range, and improves their task execution efficiency. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the composition of the control system of the present invention;

[0035] Figure 2 This is a block diagram illustrating the principle of the control method of the present invention;

[0036] Figure 3 This is a flowchart illustrating the derivation of the robot's state transition equations in this invention. Detailed Implementation

[0037] A preferred embodiment of the present invention will now be described in detail with reference to the accompanying drawings.

[0038] Terminology Explanation:

[0039] Underactuated robots: generally refers to robots whose control dimension is smaller than the dimension of the robot's controllable space degrees of freedom.

[0040] Strong coupling in robot dynamics models: Robot dynamics models generally consist of equations representing translational and rotational kinematics and dynamics. Because complex robots often change their own displacement state (translational motion) and this also affects their own posture (rotational motion), or the robot may rely entirely on changes in its own posture to generate displacement changes, the translational kinematics equations mentioned above often include rotational state variables. Similarly, rotational dynamics equations may also include translational state variables. Therefore, the dynamics models of complex robots exhibit strong coupling characteristics.

[0041] like Figure 1 As shown, a complex underactuated robot fully dynamic decoupled servo control system includes: a calibration target module, a feature perception module, a data processing module, a position control module, and an attitude control module;

[0042] The calibration target module, which serves as the calibration target signal source for the robot's onboard sensors, can provide perception information, as well as its own absolute position and attitude information.

[0043] The feature perception module acquires the perception information of the calibrated target and the surrounding environment information through its environmental perception unit, which includes various airborne sensors that enable the robot to acquire environmental perception capabilities.

[0044] The data processing module accurately identifies and acquires the perception information P of the calibration target from the surrounding environment information transmitted by the feature perception module by running data processing algorithms corresponding to the robot's onboard sensors. At the same time, the data processing module combines the state information S of the position control module and the attitude control module, and calculates the relationship matrix R between the calibration target perception information P and the robot control module state information S by running a dynamic decoupling algorithm. This enables the decoupled control of translation and rotation state variables in the robot's dynamic model and generates control signals.

[0045] The position control module receives control signals transmitted by the data processing module, converts the control signals into corresponding position control signals, and transmits them to the attitude control module.

[0046] The attitude control module receives the position control signal transmitted by the position control module, converts the position control signal into a corresponding attitude control signal, and controls the robot actuator to achieve the corresponding attitude and position changes through the attitude control signal.

[0047] like Figure 2 As shown, the servo control method of the complex underactuated robot fully dynamic decoupled servo control system of the present invention includes the following steps:

[0048] S1. By analyzing the forces acting on each moving part of the robot, the kinetic energy equations T for each moving part of the robot are established. i and potential energy equation V i , where i = 1, 2, 3…n, and n represents the number of moving parts of the robot;

[0049] S2, By using the Lagrange function L=Σ(T) i –V i Substituting i = 1, 2, 3…n into the Lagrange equation, a robot dynamics model is established;

[0050] S3. By analyzing the mapping relationship between the robot's control variables and its own state variables, establish a transformation equation that characterizes this transformation relationship;

[0051] S4. Based on the transformation equation, obtain the control model for robot rotation and translation decoupled, and generate the control signal.

[0052] like Figure 3 As shown, the process of deriving the transformation equations and performing decoupling control in S3 and S4 specifically includes:

[0053] S31. Select the robot's onboard sensors according to the robot's specific target task. The onboard sensors include vision sensors, ultrasonic sensors, and lidar.

[0054] S32. Select the appropriate sensing and calibration target according to the type of airborne sensor;

[0055] S33. The robot position controller and attitude controller control the robot to change its own attitude and position, while measuring the perception information of the corresponding calibrated target based on the airborne sensors.

[0056] S34. Based on the robot's position information, posture information, and the perception information of the calibration target, derive the relationship matrix R between the rate of change of the perception information of the calibration target dP / dt and the rate of change of the robot's own state information dS / dt, and obtain the transformation equation dP / dt=RdS / dt.

[0057] in x, y, and z represent the robot's displacement along the x-axis, y-axis, and z-axis of the inertial coordinate system, respectively. θ and ψ represent the robot's roll, pitch, and yaw angles, respectively, and T represents transpose.

[0058] S41. Design the position and attitude control algorithm for the control model based on the transformation equation:

[0059] f = K f (m, g)Re P τ=K τ (J, ω)Re ω ;

[0060] Where f is the robot's position control signal, τ is the robot's attitude control signal, and K f K τ This is an adjustable parameter matrix, where m is the robot mass, g is the gravitational acceleration, J is the moment of inertia, ω is the robot angular velocity, and e is the angular velocity. P e ω All are state tracking errors;

[0061] S42. Based on the position and attitude control algorithm, decouple the translation and rotation state variables in the robot dynamics model and generate the control signal.

[0062] Example

[0063] Taking a robot equipped with a vision sensor as an example, this invention describes a complex underactuated robot's fully dynamic decoupled servo control system and method.

[0064] In the fully dynamic decoupled servo control system, the calibration target module is a visual feature point, the feature perception module is a visual sensor, the data processing module is a robot onboard processor, the position control module is a robot position control algorithm, and the attitude control module is a robot actuator control algorithm; the robot position control algorithm and the robot actuator control algorithm are collectively referred to as position and attitude control algorithms.

[0065] Establish the kinetic energy equations T for each component of the robot i =m i v i / 2 and the potential energy equation V i =m i gl i , where m i For the mass of each moving part, v i For the velocity of motion, l i Let g be the altitude, and g represent the acceleration due to gravity.

[0066] The Lagrange function L = ∑(T) i -V i Substituting the Lagrange equations, the robot's dynamic model is derived and established, which takes the following form:

[0067]

[0068] Where q represents the robot state, u represents the control input, and M, C, N, and A are parameter matrices.

[0069] Establish the transformation equation dP / dt=RdS / dt to characterize the mapping relationship between the robot's control variables and its own state variables, where P=[p u p v ] T The coordinates of the visual feature points on the image plane of the visual sensor;

[0070]

[0071] Where f is a fixed parameter of the vision sensor, and z is the spatial height of the robot.

[0072] The position and attitude control algorithms used in the design of the control model are based on the aforementioned transformation equations.

[0073] f = K f (m, g)Re P τ=K τ (J, ω)Re ω ;

[0074] Where f is the robot's position control signal, τ is the robot's attitude control signal, and K f K τThis is an adjustable parameter matrix, where m is the robot mass, g is the gravitational acceleration, J is the moment of inertia, ω is the robot angular velocity, and e is the angular velocity. P e ω All are state tracking errors;

[0075] Based on position and attitude control algorithms, decoupled control of translational and rotational state variables in the robot dynamics model is realized, control signals are generated, and control tasks are completed.

[0076] The present invention provides a Lagrange method for deriving a complex robot dynamics model to achieve decoupled control; other dynamics model establishment methods can also be used here.

[0077] In a specific embodiment of the technical solution of the present invention, a visual sensor is selected as the feature perception module of the robot dynamics decoupling system. This sensor can be replaced with any other sensing device, including but not limited to ultrasonic sensors, infrared sensors, lidar, etc., depending on the specific task type of the robot.

[0078] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention, and no reference numerals in the claims should be construed as limiting the scope of the claims.

[0079] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A fully dynamic decoupled servo control system for a complex underactuated robot, characterized in that, include: The system includes a target calibration module, a feature perception module, a data processing module, a position control module, and an attitude control module. The calibration target module, which serves as the calibration target signal source for the robot's onboard sensors, can provide perception information, as well as its own absolute position and attitude information. The feature perception module acquires the perception information of the calibrated target and the surrounding environment information through its environmental perception unit, which includes various airborne sensors that enable the robot to acquire environmental perception capabilities. The data processing module accurately identifies and acquires the perception information P of the calibration target from the surrounding environment information transmitted by the feature perception module by running data processing algorithms corresponding to the robot's onboard sensors. At the same time, the data processing module combines the state information S of the position control module and the attitude control module, and calculates the relationship matrix R between the calibration target perception information P and the robot control module state information S by running a dynamic decoupling algorithm. This enables the decoupled control of translation and rotation state variables in the robot's dynamic model and generates control signals. The position control module receives control signals transmitted by the data processing module, converts the control signals into corresponding position control signals, and transmits them to the attitude control module. The attitude control module receives the position control signal transmitted by the position control module, converts the position control signal into a corresponding attitude control signal, and controls the robot actuator to achieve the corresponding attitude and position changes through the attitude control signal. Control methods for control systems include: Step 1: By analyzing the forces acting on each moving part of the robot, establish the kinetic energy equation T for each moving part of the robot. i and potential energy equation V i , where i = 1, 2, 3…n, and n represents the number of moving parts of the robot; Step 2: By using the Lagrange function L=Σ(T) i –V i Substituting i=1,2,3…n into the Lagrange equation, a robot dynamics model is established; Step 3: By analyzing the mapping relationship between the robot's control variables and its own state variables, establish a transformation equation representing this transformation relationship; where the robot's control variables include the perception information P of the calibration target, and the robot's own state variables include the state information S of the position control module and the attitude control module; Step 4: Based on the transformation equation, obtain the control model for robot rotation and translation decoupled, and generate the control signal; Step three specifically includes: S31. Select the robot's onboard sensors according to the robot's specific target task. Onboard sensors include vision sensors, ultrasonic sensors, or lidar. S32. Select the appropriate sensing and calibration target according to the type of airborne sensor; S33. The robot position controller and attitude controller control the robot to change its own attitude and position, while measuring the perception information of the corresponding calibrated target based on the airborne sensors. S34. Based on the robot's position information, posture information, and the perception information of the calibration target, derive the relationship matrix R between the rate of change of the perception information of the calibration target dP / dt and the rate of change of the robot's own state information dS / dt, and obtain the transformation equation dP / dt=RdS / dt. in x, y, and z represent the robot's displacement along the x-axis, y-axis, and z-axis of the inertial coordinate system, respectively; φ, θ, and ψ represent the robot's roll, pitch, and yaw attitude angles, respectively; and T represents transpose.

2. The control method for the fully dynamic decoupled servo control system of a complex underactuated robot according to claim 1, characterized in that, Step four specifically includes: S41. Design the position and attitude control algorithm for the control model based on the transformation equation: ； Where f is the robot's position control signal, and τ is the robot's attitude control signal. , Here is an adjustable parameter matrix, where m is the robot's mass, g is the acceleration due to gravity, J is the moment of inertia, and ω is the robot's angular velocity. , All are state tracking errors; S42. Based on the position and attitude control algorithm, decouple the translation and rotation state variables in the robot dynamics model and generate the control signal.

Images