Adhesion type ducted flight operation robot and control method thereof

Abstract

The invention discloses an adhesion type ducted flight operation robot and a control method thereof, and relates to the field of flight operation robot control. The operation task planning module identifies an operation task target according to a ground station instruction, and plans an expected task trajectory of an operation mechanism in a three-dimensional space according to the position of the operation task target; the tail end track integrated control module determines the position of a flight platform, a yaw angle reference value and all joint angle reference values of a mechanical arm through a nonlinear model. The flight platform pose control module estimates and compensates the coupling effect of the operation mechanism acting on the flight platform in real time, and the operation mechanism pose control module compensates the nonlinear coupling effect between the joints of the mechanical arm and between the mechanical arm and the flight platform in real time; and the tail end physical contact control module performs admittance control in an adhesion performance space of the operation mechanism according to a ground station instruction after the operation mechanism reaches a task target position. The invention aims to improve the stability control precision of the robot.

Description

Technical Field

[0001] The present application relates to the field of flight operation robot control, and in particular to an adhesion-type ducted flight operation robot and a control method thereof. Background Art

[0002] Unmanned aerial vehicles (UAVs), such as multi-rotors, have attracted widespread attention due to their simple structure, vertical takeoff and landing capabilities, and high maneuverability. They have played an important role in aerial photography, power inspections, agricultural plant protection, and other fields. However, the current application of UAVs is primarily limited to non-contact monitoring tasks. For contact-based physical interaction tasks in the airspace, such as remote object retrieval and delivery, dense jungle canopy sampling, wide-area geological exploration and sampling, remote valve manipulation in nuclear leak plants, and crack repair on cross-sea bridges, traditional manual operations are characterized by high risk, low efficiency, and high cost. Therefore, replacing manual labor with flying robots—combining a flight platform and operating mechanism, capable of both aerial flight and manipulation—has broad application value. Compared with non-contact operation tasks, airspace contact operation tasks have two significant characteristics: first, the contact operation task environment is usually a non-open complex environment with various types of environmental interference including task targets and obstacles, which has higher requirements on the robot's load capacity, constrained space passability, and operation process safety; second, during the execution of the operation task, there is a strong coupling effect between the movement of the flight platform and the operation mechanism, which brings huge challenges to the stability and precise control of the flying operation robot. At the same time, the physical interaction between the robot and the operation target further increases the control difficulty. Summary of the Invention

[0003] The purpose of this application is to provide an adhesion-type ducted flying operation robot and a control method thereof, which can improve the robot's stable control accuracy.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] In a first aspect, the present application provides an adhesive ducted flying operation robot, comprising: a body, a flying platform, an operation mechanism, and an integrated controller;

[0006] The flight platform, operating mechanism, and integrated controller are all mounted on the aircraft body; the flight platform and operating mechanism are both connected to the integrated controller; the integrated controller includes: an operating mission planning module, an integrated terminal trajectory control module, a flight platform posture control module, an operating mechanism posture control module, and a terminal physical contact control module;

[0007] The operation task planning module is used to identify the operation task target according to the ground station instructions and plan the expected task trajectory of the operation mechanism in three-dimensional space according to the operation task target position;

[0008] The terminal trajectory integrated control module is used to determine the flight platform position, yaw angle reference value, and reference values ​​of each joint angle of the manipulator based on the desired mission trajectory through a nonlinear model, so as to integrate the redundant degrees of freedom of the flight platform and the operating mechanism and achieve smooth tracking of the time-varying desired mission trajectory by the operating mechanism. The nonlinear model is a mathematical model determined by a sequential quadratic programming algorithm based on the kinematic and dynamic models corresponding to the flight operating robot to solve constrained nonlinear programming problems. The nonlinear model includes constraint conditions and an optimization objective function.

[0009] The flight platform attitude control module is used to estimate and compensate the coupling effect of the operating mechanism on the flight platform in real time based on the flight platform position and yaw angle reference value, so as to ensure the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process;

[0010] The operating mechanism posture control module is used to compensate for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform in real time based on the reference values ​​of the joint angles of the manipulator, so as to ensure the control of the manipulator joints under the posture changes of the flight platform during the operation process;

[0011] The end physical contact control module is used to perform compliant grasping of the task target through admittance control within the adhesion performance space of the operating mechanism after the operating mechanism reaches the task target position according to the instructions of the ground station to complete the task.

[0012] In a second aspect, the present application provides a control method for an adhesive-type ducted flying operation robot, which is implemented using an adhesive-type ducted flying operation robot; the control method for an adhesive-type ducted flying operation robot comprises:

[0013] Get ground station instructions;

[0014] Based on the task planning module, the task target is identified according to the ground station instructions, and the expected task trajectory of the operation mechanism in three-dimensional space is planned according to the task target position;

[0015] Based on the desired mission trajectory, the terminal trajectory integrated control module determines the flight platform position, yaw angle reference value, and reference values ​​of each joint angle of the manipulator through a nonlinear model, so as to integrate the redundant degrees of freedom of the flight platform and the operating mechanism and achieve smooth tracking of the time-varying desired mission trajectory by the operating mechanism. The nonlinear model is a mathematical model determined by a sequential quadratic programming algorithm based on the kinematic and dynamic models corresponding to the flight operating robot to solve constrained nonlinear programming problems. The nonlinear model includes constraint conditions and optimization objective functions.

[0016] Based on the flight platform's position and yaw angle reference values, the flight platform's posture control module estimates and compensates for the coupling effect of the operating mechanism on the flight platform in real time to ensure the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process.

[0017] The operating mechanism posture control module uses the reference values ​​of the joint angles of the manipulator to perform real-time compensation for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform, ensuring the control of the manipulator joints under the posture changes of the flight platform during the operation.

[0018] The end physical contact control module is used according to the ground station instructions. After the operating mechanism reaches the task target position, the operating task target is compliantly grasped through the admittance control within the adhesion performance space of the operating mechanism to complete the task.

[0019] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0020] The present application provides an adhesion-type ducted flying operation robot and its control method. The operation task planning module identifies the operation task target according to the ground station instruction, and plans the expected task trajectory of the operation mechanism in three-dimensional space according to the operation task target position; the terminal trajectory integration control module determines the flight platform position, yaw angle reference value and reference value of each joint angle of the manipulator through a nonlinear model; the flight platform posture control module estimates and compensates for the coupling effect of the operation mechanism on the flight platform in real time, and the operation mechanism posture control module compensates for the nonlinear coupling effect between each joint of the manipulator and between the manipulator and the flight platform in real time; the terminal physical contact control module performs admittance control within the adhesion performance space of the operation mechanism according to the ground station instruction after the operation mechanism reaches the task target position. The present application takes into account the ducted aerodynamic wall effect and the adhesion performance space to control the rapid response of the operation mechanism, realize the grasping of the operation target, and improve the stable control accuracy of the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 This is a schematic diagram of the structure of the adhesive ducted flying operation robot;

[0023] Figure 2 This is a schematic diagram of the control architecture of the adhesive ducted flying operation robot;

[0024] Figure 3 This is a schematic diagram of the computational fluid dynamics numerical simulation analysis of the duct wall effect;

[0025] Figure 4 is the induced tilting moment coefficient curve;

[0026] Figure 5 The flowchart of the control method of the adhesive ducted flying operation robot. DETAILED DESCRIPTION

[0027] In an exemplary embodiment, Figure 1 As shown, an adhesive ducted flying operation robot is provided, including: a body, a flying platform, an operation mechanism and an integrated controller.

[0028] The flight platform, operating mechanism and integrated controller are all arranged on the fuselage; the flight platform and the operating mechanism are both connected to the integrated controller; wherein, the integrated controller includes: an operation task planning module, an end trajectory integrated control module, a flight platform posture control module, an operating mechanism posture control module and an end physical contact control module.

[0029] The operation task planning module is used to identify the operation task target according to the ground station instructions and plan the expected task trajectory of the operation mechanism in three-dimensional space according to the operation task target position.

[0030] The terminal trajectory integrated control module is used to determine the position of the flight platform, the yaw angle reference value and the reference values ​​of each joint angle of the robotic arm according to the expected task trajectory through a nonlinear model, so as to integrate the redundant degrees of freedom of the flight platform and the operating mechanism, and realize the smooth tracking of the operating mechanism to the time-varying expected task trajectory; the nonlinear model is based on the kinematic model and dynamic model corresponding to the flying operating robot, and is a mathematical model determined by the sequential quadratic programming algorithm to solve the constrained nonlinear programming problem; the nonlinear model includes constraint conditions and optimization objective functions.

[0031] The flight platform attitude control module is used to estimate and compensate the coupling effect of the operating mechanism on the flight platform in real time according to the flight platform position and yaw angle reference value, so as to ensure the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process.

[0032] The operating mechanism posture control module is used to compensate for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform in real time according to the reference values ​​of the joint angles of the manipulator, so as to ensure the control of the manipulator joints under the posture changes of the flight platform during the operation.

[0033] The end physical contact control module is used to perform compliant grasping of the task target through admittance control within the adhesion performance space of the operating mechanism after the operating mechanism reaches the task target position according to the instructions of the ground station to complete the task.

[0034] The operating mechanism consists of a 4-DOF serial manipulator with waist, shoulder, elbow, and wrist joints and an end-adhesive actuator. The end-adhesive actuator is a biomimetic silicone rubber-based adhesive array capable of reversibly adhering and detaching objects on various shapes and surfaces. The flight platform includes four ducted propellers.

[0035] In an exemplary embodiment, a control method for an adhesive type ducted flying operation robot is provided, which is implemented using an adhesive type ducted flying operation robot. Figure 5 As shown, the control method of the adhesion type ducted flying operation robot includes:

[0036] Step 100: Get ground station instructions.

[0037] Step 200: Based on the operation task planning module, the operation task target is identified according to the ground station instruction, and the expected task trajectory of the operation mechanism in three-dimensional space is planned according to the operation task target position.

[0038] Step 300: Based on the desired mission trajectory, the terminal trajectory integrated control module determines the flight platform position, yaw angle reference value, and reference values ​​for each joint angle of the manipulator using a nonlinear model. This allows for the integrated allocation of redundant degrees of freedom between the flight platform and the work mechanism, enabling the work mechanism to smoothly track the time-varying desired mission trajectory. The nonlinear model is a mathematical model for solving constrained nonlinear programming problems, determined using a sequential quadratic programming algorithm based on the kinematic and dynamic models of the flight work robot. The nonlinear model includes constraints and an optimization objective function.

[0039] The constraints include: duct safety distance limit, robot arm self-collision avoidance limit, robot arm joint angle range limit, robot movement speed limit and robot movement acceleration limit.

[0040] The expression corresponding to the duct safety distance limit is:

[0041] |d a,x (k+i|k)|≥d a,min .

[0042] The expression corresponding to the robot's self-collision avoidance limit is:

[0043]

[0044] The expression corresponding to the robot arm joint angle range limit is:

[0045] q j,min ≤q j (k+i|k)≤q j,max .

[0046] The expression corresponding to the robot movement speed limit is:

[0047]

[0048] The expression corresponding to the robot motion acceleration limit is:

[0049]

[0050] The optimization objective function includes the cost function corresponding to the task trajectory tracking error, the cost function corresponding to the change of the center of gravity position of the manipulator, and the cost function corresponding to the duct wall effect.

[0051] The cost function corresponding to the task trajectory tracking error is:

[0052]

[0053] The cost function corresponding to the change in the center of gravity position of the robotic arm is:

[0054]

[0055] The cost function corresponding to the duct wall effect is:

[0056]

[0057] Among them, d a,x is the distance between the center of the flight platform and the target or obstacle; d a,min is the minimum safe distance between the center of the flight platform and the target or obstacle; k is the sampling time; i is the sequence number; d e,min is the minimum safe distance between the end effector and the center of the flight platform; is the position of the end-adhesive actuator relative to the flight platform; x is the independent state variable of the flight operation robot; q j is the j-th robotic arm joint angle; q j,min is the minimum angle limit of the joint angle; q j,max is the maximum angle limit of the joint angle; is the motion speed corresponding to the horizontal coordinate of the flight platform position at the k+i sampling moment; is the minimum motion speed corresponding to the horizontal coordinate of the flight platform position; is the maximum movement speed corresponding to the horizontal coordinate of the flight platform position; is the motion speed corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum motion speed corresponding to the vertical coordinate of the flight platform position; is the maximum movement speed corresponding to the vertical coordinate of the flight platform position; is the motion speed corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum movement speed corresponding to the vertical coordinate of the flight platform position; is the maximum movement speed corresponding to the vertical coordinate of the flight platform position; is the angular velocity corresponding to the yaw attitude angle at the k+i sampling moment; is the minimum angular velocity corresponding to the yaw attitude angle; is the maximum angular velocity corresponding to the yaw attitude angle; is the angular velocity corresponding to the joint angle of the manipulator at the k+i sampling moment; is the minimum angular velocity corresponding to the joint angle of the robotic arm; is the maximum angular velocity corresponding to the joint angle of the robotic arm; is the motion acceleration corresponding to the horizontal coordinate of the flight platform position at the k+i sampling moment; is the minimum motion acceleration corresponding to the horizontal coordinate of the flight platform position; is the maximum motion acceleration corresponding to the horizontal coordinate of the flight platform position; is the motion acceleration corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the maximum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the motion acceleration corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the maximum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the angular acceleration corresponding to the yaw attitude angle at the k+ith sampling moment; is the minimum angular acceleration corresponding to the yaw attitude angle; is the maximum angular acceleration corresponding to the yaw attitude angle; h1 is the cost function corresponding to the task trajectory tracking error; N p is the prediction time domain; p e is the position of the end effector of the flight operation robot; p e,d is the expected task trajectory; W1 is the tracking error weight matrix; W s1 is the tracking error terminal weight matrix; h2 is the cost function corresponding to the change of the center of gravity position of the manipulator; e m,xyis the prediction model function of the center of gravity position change of the manipulator; W2 is the weight matrix of the center of gravity position change of the manipulator; W s2 is the terminal weight matrix of the manipulator's center of gravity position change; h3 is the cost function corresponding to the duct wall effect; C WE is the duct wall effect induced tilting moment coefficient; W3 is the duct wall effect weight value; W s3 is the terminal weight value of the duct wall effect.

[0058] Step 400: Based on the flight platform posture control module and the flight platform position and yaw angle reference value, the coupling effect of the operating mechanism on the flight platform is estimated and compensated in real time to ensure the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process.

[0059] Step 500: Using the operating mechanism posture control module, according to the reference values ​​of the joint angles of the manipulator, the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform is compensated in real time to ensure the control of the manipulator joints under the posture changes of the flight platform during the operation.

[0060] Step 600: The end physical contact control module is used to perform compliant grasping of the task target through admittance control within the adhesion performance space of the working mechanism according to the ground station instruction after the working mechanism reaches the task target position to complete the task.

[0061] In one embodiment, the mathematical expression of the kinematic model corresponding to the flying operation robot is:

[0062]

[0063] The mathematical expression of the dynamic model corresponding to the flying operation robot is:

[0064]

[0065] in, is the speed of the end effector at time t; Φ a (t) is the attitude angle of the flight platform at time t; q(t) is the joint angle of the manipulator at time t; is the first-order derivative of x at time t; J tx,e (Φ a (t),q(t)) is J t,e (Φ a ,q) related to x; J t,e (Φ a ,q) is the Jacobian matrix of the end effector; x is the independent state variable of the flight operation robot; M(ξ) is the inertia matrix; is the Coriolis force and centrifugal force matrix; g(ξ) is the gravity matrix; μu is the generalized input force; μ ext is the generalized external force; ξ is the state variable of the flying operation robot; is the first-order derivative of ξ; is the second-order derivative of ξ.

[0066] The corresponding expression for the constrained nonlinear programming problem is:

[0067]

[0068] Among them, J(x(k),u k ) is the system cost function; h j is the optimization objective; u k is the control input vector to be optimized; x(k) is the system state at sampling time k; j is the optimization target number; m is the total number of optimization targets; x(k+i|k) is the state of the system at sampling time k+i predicted by sampling time k; u(k+i) is the control input at sampling time k+i; x(k+i+1|k) is the state of the system at sampling time k+i+1 predicted by sampling time k; T s is the sampling time; x(k|k) is the intermediate variable corresponding to x(k); g(x(k+i|k),u(k+i)) is the general constraint function related to the system state variables and control variables; g min is the upper bound of the system's general constraints; g max is the lower bound of the system's general constraint; x min is the upper constraint boundary of the system state variable; x max is the lower constraint boundary of the system state variable; u min u is the upper constraint boundary of the system control variable; max is the lower constraint boundary of the system control variable; Δu min is the upper constraint boundary of the system control variable increment; Δu max Set the lower bound for the increment of the system control variable.

[0069] Real-time estimation and compensation of the coupling effect of the operating mechanism on the flight platform is achieved based on the update law of the operating mechanism coupling effect estimation and the control output of the operating mechanism coupling compensation.

[0070] The update law of the estimated coupling effect of the operating mechanism is expressed as follows:

[0071]

[0072] The control output of the working mechanism coupling compensation corresponds to the expression:

[0073]

[0074] in, is the update law for estimating the coupling effect of the operating mechanism; Γ ad is the adaptive gain; Proj(·,·) is the projection operator; are the estimated values ​​of the coupling force and coupling torque exerted by the operating mechanism on the flight platform; is the estimation error of the flight platform motion state variables; P ad is the solution of the system algebraic Lyapunov equation; T is the transpose; u ad (s) is the control output considering the coupling compensation of the operating mechanism in the frequency domain; C(s) is a low-pass filter; K des is the intermediate variable; a,r (s) is the reference input y a,r Laplace form of F dis (s) is the intermediate variable in Laplace form; for Laplace form of ; s is the Laplace operator.

[0075] The operating mechanism posture control module uses the reference values ​​of the joint angles of the manipulator to perform real-time compensation for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform to ensure the control of the manipulator joints under the posture changes of the flight platform during the operation. Specifically, it includes:

[0076] The joint tracking error of the manipulator is determined based on the reference values ​​of each joint angle of the manipulator, and the PD method is used for feedback control. The mathematical expression corresponding to the feedback control is:

[0077]

[0078] According to the dynamic model corresponding to the flying operation robot, the dynamic model of the flying operation robot's manipulator subsystem is obtained, and the inverse dynamics compensation is determined.

[0079] The mathematical expression corresponding to inverse dynamics compensation is:

[0080]

[0081] Based on feedback control and inverse dynamics compensation, the control of the manipulator joint under the change of the flight platform's posture during the operation is determined, and the manipulator joint control input is obtained; the expression corresponding to the manipulator joint control input is:

[0082] u m (t) = u ff (t)+u fb (t).

[0083] Among them, u fb (t) is the feedback control at time t; K p,q is the proportional gain matrix; K d,qis the differential gain matrix; e q (t) is the tracking error of the robot arm joint; for e q The derivative of (t); u ff (t) is the inverse dynamics compensation at time t; M m is the simplified inertia matrix of the manipulator subsystem; ξ(t) is the state variable of the flight operation robot at time t; is the second derivative of the desired reference angle of the manipulator joint; C m It is the name of the Coriolis force and centrifugal force matrix of the manipulator subsystem; is the first-order derivative of ξ(t); is the differential of the desired reference angle of the manipulator joint; g m It is the name of the gravity matrix of the manipulator subsystem, which is equivalent to g m (ξ); M a It is the abbreviation of the coupling force and coupling torque exerted by the flight platform on the robotic arm; is the second-order derivative of the flight platform state variable ζ(t) at time t; u m (t) is the robot arm joint control input at time t.

[0084] By performing admittance control within the adhesion performance space of the operating mechanism, the reference input of the operating mechanism posture control module is updated, and the operating task target is compliantly grasped based on the reference input to complete the operating task.

[0085] The expression for the reference input is:

[0086] q r,d (t) = q r (t)+Δq r (t).

[0087] The expression corresponding to the adhesion performance space is:

[0088]

[0089] Among them, q r,d (t) is the reference input at time t; q r (t) is the desired reference angle of the manipulator joint; Δq r (t) is the equivalent expected increment of the joint angle of the manipulator; Ω is the adhesion performance space; f ext,n f ext The normal component of ext,t f ext The tangential component of ext is the physical contact force of the end-adhesive actuator; m ext is the torque; F sat is the adhesion saturation boundary; F pre is the minimum preload boundary; Fdetach is the desorption force boundary; M detach is the desorption moment boundary; τ is the shear stress at the root of the adhesion microstructure at the contact interface; τ max is the microstructural failure boundary.

[0090] Specifically, an adhesive ducted flying operation robot is composed of a body, a flight platform, an operation mechanism, and an integrated controller, wherein: (1) the flight platform includes four ducted propellers; (2) the operation mechanism includes a 4-DOF serial manipulator and an end-adhesive actuator:

[0091] ① The 4-DOF serial robotic arm includes waist joint, shoulder joint, elbow joint, and wrist joint. ② The end adhesion actuator is a bionic silicone rubber-based adhesion array.

[0092] (3) The integrated controller receives the robot sensor feedback information and ground station control instructions in real time, and comprehensively controls the flight platform and the operating mechanism; among them, the sensor feedback information includes the operating target position information obtained by the depth vision camera, the body position information obtained by GPS or radar, the body posture, linear velocity, angular velocity, linear acceleration, and angular acceleration information measured and calculated by the flight platform IMU, the joint angle, angular velocity, and angular acceleration information obtained by the robot arm joint, and the contact force and torque information measured by the end six-dimensional force sensor.

[0093] The operation task planning module identifies the operation task targets such as the objects to be grasped according to the instructions of the ground station, and plans the expected task trajectory of the robot end effector in three-dimensional space according to the target position; the end trajectory integrated control module obtains the flight platform position, yaw angle reference, and each joint angle reference of the manipulator according to the expected task trajectory through nonlinear model prediction optimization, and realizes the smooth tracking of the end effector to the time-varying expected task trajectory through the integrated optimal allocation of the redundant degrees of freedom of the flight platform and the operation mechanism driven by the expected task. In this process, the task trajectory tracking error, the change of the center of gravity position of the manipulator, the effect of the duct wall, the duct safety distance limit, the manipulator self-collision avoidance limit, the manipulator joint angle range limit, the robot movement speed limit, the robot movement acceleration limit and other constraints are comprehensively considered; the flight platform posture control module obtains the ducted propeller control input according to the flight platform position and yaw angle reference, which includes an inner loop robust control module, an operation mechanism coupling estimation module, The operating mechanism coupling compensation module and the outer loop position control module realize real-time estimation and compensation of the coupling effect of the operating mechanism on the flight platform, ensuring the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process; the operating mechanism posture control module obtains the control input of each joint of the manipulator based on the reference of each joint angle of the manipulator, including the feedback control module and the inverse dynamics compensation module, and realizes real-time compensation of the nonlinear coupling effect between each joint of the manipulator and between the manipulator and the flight platform, ensuring high-precision control of the manipulator joint under the change of the flight platform posture during the operation process; the end physical contact control module realizes the smooth grasping of the operation target through the admittance control in the adhesion performance space of the end effector according to the instructions of the ground station when the end effector of the robot reaches the task target position, completing the operation task, and avoiding the risk of robot instability and damage to the operation target caused by the collision and impact of the conventional flying operation robot and the rigid grasping process of the operation target. See the control architecture diagram of the adhesion-type ducted flying operation robot for details. Figure 2 .

[0094] (1)Job task planning module.

[0095] The task planning module receives the target position instruction p from the ground station. o =[x o y o z o ] T , or receive the target task instruction to be grasped from the ground station, and identify and obtain the target position instruction p through the onboard depth vision camera o The three-dimensional expected task trajectory p of the end effector of the flight operation robot from the current position to the operation target position is planned by the RRT* method. e,d =[x e,d y e,d ze,d ] T , the expected grasping contact force f is obtained through expert knowledge d .

[0096] (2) Terminal trajectory integrated control module.

[0097] The terminal trajectory integrated control module receives the expected task trajectory p output by the task planning module e,d , the expected reference η of the flight platform position and yaw angle is obtained through real-time optimization algorithm a,r and the desired reference q of each joint angle of the robot arm r , and serve as the input of the flight platform posture control module and the operation mechanism posture control module respectively, to realize the end effector to the desired task trajectory p e,d Precise and smooth tracking.

[0098] ① Establish the kinematic and dynamic model of the flying operation robot.

[0099] Define the position p of the end effector of the flying robot e =[x e y e z e ] T , flight platform position p a =[x a y a z a ] T , flight platform roll, pitch, and yaw attitude angles Flight platform speed Flight platform angular velocity Robot arm joint angle q = [q1 q2 q3 q4] T , flight platform state variables Flight platform independent state variable η a =[x a y a z a ψ a ] T , the flight platform underactuated dependent state variables The state variable of the flying operation robot ξ=[ζ T q T ] T , independent state variables of the flight operation robot Based on the multi-rigid body dynamics theory, the kinematic model and dynamic model of the flying operation robot are established.

[0100] Kinematic model:

[0101] The position p of the end effector of the flying operation robot e for:

[0102]

[0103] R a is the rotation matrix from the flight platform coordinate system to the world coordinate system, is the position of the end effector relative to the flight platform.

[0104] Linear speed of the operating mechanism's robotic arm link i and angular velocity for:

[0105]

[0106] is the rotation matrix of the robot arm link i coordinate system and i-1 coordinate system, is the position of the robot arm link i relative to the i-1 coordinate system, is the velocity of the robot arm joint angle i, e zi is the unit vector of the Z axis of the coordinate system of link i.

[0107] The speed of the end effector of the flying robot is obtained by recursive solution for:

[0108]

[0109] J t,e (Φ a ,q) is the Jacobian matrix of the end effector.

[0110] According to the assumption that the body attitude angle is small, the effect of the under-actuated non-independent state variables of the flight platform on the end effector velocity is ignored. The kinematic model of the flight operation robot is obtained as follows:

[0111]

[0112] J tx,e (Φ a (t),q(t)) is J t,e (Φ a ,q) are the items related to x.

[0113] Kinetic model:

[0114] The Newton-Euler equations of the flight platform of the flying operation robot are:

[0115]

[0116] and are the net external force and torque acting on the flight platform, m ais the mass of the flight platform, I a is the inertia matrix of the flight platform.

[0117] The Newton-Euler equation of the link i of the operating mechanism robot arm is:

[0118]

[0119] F i i and are the net external force and torque on the robot arm link i, m li with I li is the mass and inertia matrix of the robot arm link i, and are the linear acceleration and angular acceleration of the center of mass of the robot link i.

[0120] The force balance equation and torque balance equation of the operating mechanism's robotic arm link i are:

[0121]

[0122] R i is the rotation matrix from the robot arm link i coordinate system to the world coordinate system, f i i is the force exerted by the robot link i-1 on the link i, is the torque of the robot link i-1 acting on the link i, τ i is the driving torque on joint i of the robot arm, is the position of the center of mass of the robot arm link i relative to the link i coordinate system.

[0123] The dynamic model of the flying operation robot is obtained by recursive solution:

[0124]

[0125] M(ξ) ​​is the inertia matrix, is the Coriolis force and centrifugal force matrix, g(ξ) is the gravity matrix, μ u is the generalized input force, including the ducted propeller lift and counter torque, and the manipulator joint drive torque, μ ext It is a generalized external force, that is, the physical interaction force between the end effector and the work target.

[0126] ②Construct an integrated terminal trajectory control architecture.

[0127] An integrated terminal trajectory control architecture based on nonlinear model prediction optimization is constructed, which includes three parts: prediction model, rolling optimization, and feedback correction.

[0128] First, define the prediction model of the system Get its discrete form x(k+1)=x(k)+T s u(k), where is the control input, T s is the sampling time, k is the sampling moment, and k+1 is the next sampling moment.

[0129] Secondly, at the k sampling time, based on the current state x(k) of the system and the prediction model, the constrained nonlinear programming problem is solved by the sequential quadratic programming (SQP) algorithm to obtain the control input sequence u that optimizes the system performance index * (k+i), where the constrained nonlinear programming problem is described as:

[0130]

[0131] J(x(k),u k ) is the system cost function, which contains multiple optimization objectives h j .u k is the control input vector to be optimized, denoted as u k =[u T (k)u T (k+1)…u T (k+N m -1)] T , N m is the control time domain. i=0,1,…,N p , N p is the prediction time domain. min with x max are the upper and lower constraint boundaries of the system state variables, u min with u max are the upper and lower constraint boundaries of the system control variables, Δu min and Δu max is the upper and lower constraint boundaries of the system control variable increment, Δu(k+i)=u(k+i)-u(k+i-1) is the control variable increment, g min With g max are the upper and lower bounds of the system’s general constraints, and g(x(k+i|k),u(k+i)) is the general constraint function related to the system’s state variables and control variables.

[0132] Finally, the first element u of the input sequence is optimally controlled * (k) Obtain the output of the terminal trajectory integrated control module, that is, the optimal expected reference of the flight platform position and yaw angle at the next moment and the optimal expected reference of each joint angle of the robotic arm The prediction model is updated according to the actual state of the system, and the above constrained nonlinear programming problem is re-solved at the next moment.

[0133] ③Design optimization objective function.

[0134] The kinematic model of the flight operation robot is discretized to obtain the end effector position prediction model:

[0135] p e (k+i+1|k)=p e (k+i|k)+T s J tx,e (x(k+i|k))u(k+i).

[0136] The cost function of the end effector tracking error (the cost function corresponding to the task trajectory tracking error) is:

[0137]

[0138] W1 is the tracking error weight matrix, W s1 is the tracking error terminal weight matrix.

[0139] According to the kinematic model of the flight operation robot, the projection of the center of gravity of the manipulator relative to the flight platform in the horizontal plane of the flight platform body is obtained as follows:

[0140]

[0141] is the position of the center of mass of the robot arm link i relative to the flight platform, e xy is the planar unit projection matrix.

[0142] The prediction model of the center of gravity position change of the manipulator obtained by discretization is:

[0143]

[0144] is the projection of the center of gravity of the robot arm in the initial position.

[0145] The cost function of the change in the center of gravity position of the robotic arm is:

[0146]

[0147] W2 is the weight matrix of the robot arm’s center of gravity position change, W s2 is the terminal weight matrix of the robot arm's center of gravity position change.

[0148] Flying robots operate in complex, multi-obstacle environments, such as indoor environments. When performing tasks like grasping, walls, obstacles, and objects can affect the aerodynamic flow field of the platform's ducted propellers, generating an induced tilting torque that degrades system performance and can even cause the robot to tip over. This is known as the ducted wall effect. To minimize disturbances to the platform caused by the arm's center of gravity, avoid extreme positions of the arm's joints, and improve target tracking speed, the platform's motion should dominate the operation, with the arm extended as little as possible. Furthermore, due to the wall effect, the robot's performance deteriorates when the platform is too close to the environment or target.

[0149] Define the dimensionless duct wall effect induced tilting moment coefficient Among them, M WE is the tilting moment caused by the wall effect, R r is the radius of the ducted propeller, ω is the speed of the ducted propeller, and ρ is the air density. Figure 3 As shown in the figure, the wall effect of the ducted propeller of the flying robot is numerically simulated and analyzed by the computational fluid dynamics (CFD) method, and the C WE Dimensionless distance d from the wall WE / R r The relationship curve between Figure 4 As shown, where d WE is the distance from the center of the ducted propeller to the wall (target or obstacle). The numerical fitting of the relationship curve yields:

[0150] C WE =f WE (d WE )≈0.002052(d WE / R r ) -1.323 +3.153×10 -5 .

[0151] The discretization prediction model of the duct wall effect is:

[0152]

[0153] The cost function of the duct wall effect is:

[0154]

[0155] W3 is the weight value of the duct wall effect, W s3 is the terminal weight value of the duct wall effect.

[0156] The incremental cost function of the control action is:

[0157]

[0158] W4 is the control action incremental weight matrix.

[0159] ④Design constraint function.

[0160] a.Culvert safety distance limit.

[0161] |d a,x (k+i|k)|≥d a,min .

[0162] d a,x is the distance between the center of the flight platform and the target or obstacle, d a,min It is the minimum safe distance between the center of the flight platform and the target or obstacle.

[0163] b. Robotic arm self-collision avoidance limit.

[0164]

[0165] d e,min It is the minimum safe distance between the end effector and the center of the flight platform.

[0166] c. Limitation of the robot arm joint angle range.

[0167] q j,min ≤q j (k+i|k)≤q j,max .

[0168] q j (j=1,2,3,4) is the joint angle of the robot arm, q j,min With q j,max They are the minimum and maximum angle limits of the joint angle respectively.

[0169] d. Robot movement speed limit.

[0170]

[0171] Δ min With Δ max They are the minimum (the speed is negative) and maximum limits of the corresponding motion variable speed.

[0172] e. Robot movement acceleration limit.

[0173]

[0174] Δ min With Δ max They are the minimum (acceleration is negative) and maximum limits of the corresponding motion variable acceleration.

[0175] (3) Flight platform attitude control module.

[0176] The flight platform attitude control module receives the flight platform position and yaw angle expected reference η output by the terminal trajectory integrated control module a,r , outputs ducted propeller control inputs, drives the flight platform to accurately track the desired reference position and yaw angle, and simultaneously compensates for the coupling effects of the working mechanism in real time. It includes an inner-loop robust control module, a working mechanism coupling estimation module, a working mechanism coupling compensation module, and an outer-loop position control module.

[0177] ① Inner loop robust control module.

[0178] Define the flight platform motion state variables Outer loop output variable η a =[x a y a z a ψ a ] T , the inner loop output variable y a =[u a v a w a ψ a ] T .x a (t) is the flight platform motion state variable at time t; is x a The first derivative of (t); y a (t) is the inner loop output variable at time t.

[0179] By linearizing the dynamic model of the flying robot, the state space model of the flying platform is obtained as follows:

[0180]

[0181] A a is the open-loop state matrix of the flight platform, B a is the open-loop control matrix of the flight platform, u a It is the control input of the flight platform, including altitude channel, lateral channel, longitudinal channel, heading channel, u a (t) is the flight platform control input at time t; f m That is, f m (t) is the coupling force and coupling torque of the operating mechanism acting on the flight platform. a Output matrix for the flight platform.

[0182] Define the inner loop tracking error of the flight platform as e a (t) = y a,r (t)-y a (t), the augmented motion state variable of the augmented tracking error is x aug=[e a x a ] T , inner loop robust controller u a (t) Designed to:

[0183] u a (t) = K st x aug (t).

[0184] K st =[K FF -K FB ].

[0185] The inner loop robust controller consists of two static control matrices K FF and K FB Composition, where the control matrix K FB Used to achieve stabilization and decoupling of the system state, the control matrix K FF Used to realize the system inner loop output variable y a For reference input y a,r Tracking. K st By solving the closed-loop transfer function H that minimizes the system performance ∞ The optimization problem of the norm is obtained. a，r (t) is the reference input at time t; y a (t) is the system inner loop output variable at time t.

[0186] ②Operation mechanism coupling estimation module.

[0187] The inner loop closed-loop system composed of the flight platform and the inner loop robust control module is:

[0188]

[0189] A ac is the state matrix of the inner closed-loop system of the flight platform, B ac It is the control matrix of the inner closed-loop system of the flight platform.

[0190] The designed state estimator is:

[0191]

[0192] They are respectively the estimated values ​​of the flight platform motion state variables, the estimated values ​​of the coupling force and coupling torque of the operating mechanism acting on the flight platform, and the estimated values ​​of the inner loop output variables, u ad (t) is the control output considering the coupling compensation of the operating mechanism.

[0193] The system estimation error dynamics is:

[0194]

[0195] is the estimation error of the flight platform motion state variables, is the estimation error of the coupling force and torque of the operating mechanism. for The first derivative of .

[0196] Define the Lyapunov candidate function

[0197]

[0198] P ad is the solution of the system algebraic Lyapunov equation, Γ ad is the adaptive gain. for The transpose of .

[0199] The derivative is:

[0200]

[0201] The update law for the estimated coupling effect of the designed operating mechanism is:

[0202]

[0203] Proj(·,·) is the projection operator. Substituting it into the above formula, we can get:

[0204]

[0205] Then the system estimation error is Lyapunov uniformly stable.

[0206] ③Operating mechanism coupling compensation module.

[0207] The control output considering the coupling compensation of the operating mechanism is designed as follows:

[0208]

[0209] H ax (s)=(sI-A ac ) -1 .

[0210] H a (s)=C a H ax (s)B ac .

[0211] H ax (s), H a (s) are all intermediate variables.

[0212] C(s) is a low-pass filter that satisfies the following conditions:

[0213]

[0214] u ad The real-time compensation of the coupling force and coupling torque of the operating mechanism on the flight platform is realized, ensuring that the flight platform posture control module is responsive to the reference input y a,r Precise tracking.

[0215] ④Outer ring position control module.

[0216] Define the outer loop tracking error of the flight platform as e ap (t) = η a,r (t)-η a (t), the outer loop position controller adopts the PD method and is designed as:

[0217]

[0218] K p,a is the proportional gain matrix, K d,a is the differential gain matrix.

[0219] (4) Operating mechanism posture control module.

[0220] The operating mechanism posture control module receives the desired reference q of the manipulator joint angle output by the terminal trajectory integrated control module r The system outputs control inputs for each joint of the manipulator, driving them to accurately track the desired reference while simultaneously compensating in real time for nonlinear coupling effects between the joints and between the manipulator and the flight platform. It includes a feedback control module and an inverse dynamics compensation module.

[0221] ①Feedback control module.

[0222] Define the robot arm joint tracking error as e q (t) = q r (t)-q(t), the feedback control adopts the PD method and is designed as:

[0223]

[0224] K p,q is the proportional gain matrix, K d,q is the differential gain matrix.

[0225] ②Inverse dynamics compensation module.

[0226] According to the dynamic model of the flight operation robot, the dynamic model of the flight operation robot's manipulator subsystem is obtained as follows:

[0227]

[0228] M m (ξ) is M m (ξ(t)) is the inertia matrix of the manipulator subsystem, Right now is the Coriolis force and centrifugal force matrix of the robotic arm subsystem, g m (ξ) is g m (ξ(t)) is the gravity matrix of the manipulator subsystem, which is affected by the coupling effect of state variables such as the flight platform attitude and speed. Right now is the coupling force and coupling torque of the flight platform on the robotic arm, u m That is u m (t) is the robot arm joint control input.

[0229] The inverse dynamics compensation control is designed as follows:

[0230]

[0231] and are the differential and quadratic differential of the desired reference joint angle of the robot arm, respectively.

[0232] Combining feedback control and inverse dynamics compensation, the control input of the robot arm joint is obtained as:

[0233] u m (t) = u ff (t)+u fb (t).

[0234] (5) Terminal physical contact control module.

[0235] The end physical contact control module follows the instructions from the ground station. When the robot's end effector reaches the target position, the module achieves a smooth, stable and safe target grasping process through the design of an admittance controller based on the adhesion performance space without changing the original trajectory control module architecture and without switching between contact grasping and trajectory tracking control laws.

[0236] The adhesion performance of the end-adhesion actuator is tested by multifunctional friction and wear testing machine and other equipment to analyze the normal adhesion force f ad,n , tangential adhesion force f ad,t , normal desorption force f de,n , tangential desorption force f de,t , desorption torque m de Physical contact force f with the end-adhesive actuator ext , torque m ext The quantitative relationship between the adhesion saturation boundary F sat , minimum preload boundary F pre, desorption force boundary F detach , desorption torque boundary M detach , microstructural failure boundary τ max , forming the adhesion performance space as follows:

[0237]

[0238] f ext,n f ext The normal component, f ext,t f ext is the tangential component of the contact interface, and τ is the shear stress at the root of the adhesion microstructure.

[0239] The desired compliant contact characteristics between the end-adhesive actuator and the work target are designed to be equivalent to a virtual mass-spring-damper system, that is:

[0240]

[0241] p e,d is the expected task trajectory of the end effector, p e,c For a compliant trajectory that exhibits the desired compliant contact characteristics, f d is the expected grasping contact force, f ext M is the physical contact force on the end effector, measured by the six-axis force / torque sensor on the end effector. d is the inertia matrix of the virtual mass spring damper system, B d is the damping matrix of the virtual mass spring damper system, K d is the stiffness matrix of the virtual mass-spring-damper system.

[0242] Define Δp e =p e,c -p e,d is the virtual position deviation, based on the measured physical contact force f ext , expected grasping contact force f d , expected task trajectory p e,d The virtual position deviation Δp that satisfies the adhesion performance space constraint Ω is calculated by the above formula e,Ω , then the equivalent expected increment of the robot arm joint angle is:

[0243]

[0244] J tm,e (Φ a ,q) is the Jacobian matrix J of the end effector t,e (Φ a ,q) is related to the joints of the robotic arm, For J tm,e (Φ a ,q) is the pseudo-inverse matrix.

[0245] The reference input of the updated working mechanism posture control module is:

[0246] q r,d (t) = q r (t)+Δq r (t).

[0247] The working mechanism posture control module drives the manipulator joint angle to track the new reference input q r,d Make full use of the rapid response of the robot arm joints to achieve a smooth, stable and safe grasping process of the operation target.

[0248] In summary, the present application provides an adhesive ducted flying operation robot, which is composed of a body, a flight platform, an operation mechanism, and an integrated controller, wherein: (1) the flight platform includes four ducted propellers; ① the ducted structure provides greater lift under the same outer dimensions, that is, it has a greater operating load capacity under the premise of ensuring the passability of a narrow space in a complex operating environment; ② the duct can directly contact or even collide with the environment / operation target, thereby improving the operation capability and operation safety in a complex, multi-obstacle and narrow environment; (2) the operation mechanism includes a 4-DOF serial robot arm and an end-adhesive actuator; ① the 4-DOF serial robot arm can directly contact or collide with the environment / operation target, thereby improving the operation capability and operation safety in a complex, multi-obstacle and narrow environment; The robotic arm consists of waist joints, shoulder joints, elbow joints, and wrist joints, and has the ability to operate omnidirectionally in three-dimensional space. ② The end adhesion actuator is a bionic silicone rubber-based adhesion array, which can achieve reversible adhesion / detachment grasping of various shapes / material surfaces. It is lightweight and does not require external energy drive. It has both universality and versatility in operating objects, and can save effective loads, reduce the disturbance torque of the operating mechanism, and ensure the operating load capacity; (3) The integrated controller realizes the integrated control of the flight-operation process of the robot flight platform and the operating mechanism to ensure operating performance and efficiency.

[0249] The present application discloses a control method for an adhesive ducted flying operation robot, comprising an operation task planning module, an end trajectory integrated control module, a flight platform posture control module, an operation mechanism posture control module, and an end physical contact control module, wherein: (1) the operation task planning module identifies the operation task target such as the object to be grasped according to the ground station instruction, and plans the desired task trajectory of the robot end effector in three-dimensional space according to the target position; (2) the end trajectory integrated control module obtains the flight platform position, yaw angle reference, and each joint angle reference of the manipulator according to the desired task trajectory through nonlinear model prediction optimization, and realizes the stable tracking of the end effector to the time-varying desired task trajectory through the integrated optimal allocation of the redundant degrees of freedom of the flight platform and the operation mechanism driven by the desired task. In this process, the task trajectory tracking error, the change of the center of gravity position of the manipulator, the effect of the duct wall, and the constraints such as the duct safety distance limit, the self-collision avoidance limit of the manipulator, the range limit of the joint angle of the manipulator, the robot movement speed limit, and the robot movement acceleration limit are comprehensively considered; (3) the flight platform posture control module obtains the desired task trajectory according to the flight platform The ducted propeller control input is obtained based on the position and yaw angle references, including the inner loop robust control module, the operating mechanism coupling estimation module, the operating mechanism coupling compensation module, and the outer loop position control module, which realizes real-time estimation and compensation of the coupling effect of the operating mechanism on the flight platform, ensuring the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process; (4) The operating mechanism posture control module obtains the control input of each joint of the manipulator according to the reference of each joint angle of the manipulator, including the feedback control module and the inverse dynamics compensation module, and realizes real-time compensation of the nonlinear coupling effect between each joint of the manipulator and between the manipulator and the flight platform, ensuring high-precision control of the manipulator joint under the change of the flight platform posture during the operation process; (5) The end physical contact control module, according to the ground station command, realizes the smooth grasping of the operation target through the admittance control in the adhesion performance space of the end effector after the robot end effector reaches the task target position, completes the operation task, and avoids the risk of robot instability and damage to the operation target caused by the collision and impact of the conventional flying operation robot and the rigid grasping process of the operation target.

[0250] Terminal trajectory control achieves integrated, real-time optimal allocation of the flight platform and operating mechanism, taking into account dynamic coupling. It also considers the influence of duct wall effects for the first time, improving the accuracy and smoothness of terminal trajectory tracking in complex, multi-obstacle constrained environments. Traditional methods either use separate, independent planning of the flight platform and operating mechanism, or only consider integrated allocation at the kinematic level.

[0251] The dynamic coupling between the flight platform and the work mechanism is considered in the posture control, and real-time compensation is achieved. This is the first time that the influence of the flight platform's posture dynamics on the manipulator's joint tracking control has been considered and compensated in real time, improving the robot's control accuracy. Traditional methods either only consider the kinematic coupling between the flight platform and the work mechanism, or treat the work mechanism's motion as an external disturbance to the flight platform, and only perform disturbance compensation control on the flight platform.

[0252] For the first time, the end-point physical contact control establishes and considers the spatial constraints of adhesion performance during the grasping operation, achieving safe and smooth grasping of the work target. This not only avoids the risk of damage and instability to the robot itself caused by collision and impact during conventional rigid grasping, but also avoids the risk of damage to fragile work targets.

[0253] In terms of structure, the core components of the flying operation robot mainly include the flying platform and the operating mechanism.

[0254] (1) In terms of flight platform, in the existing technology, the flight platform configuration of flying operation robots is mainly multi-rotor and helicopter. However, open rotor configurations such as multi-rotor and helicopter cannot directly contact the environment, and the load capacity is extremely sensitive to the structural size. It is difficult to carry high-performance operating mechanisms to perform work tasks in a compact size with narrow space passability. The duct structure adopted in this application has the following beneficial effects: First, the duct structure provides greater lift under the same outer dimensions, that is, it has a greater operating load capacity under the premise of ensuring the passability of narrow spaces in complex working environments; second, the duct can directly contact or even collide with the environment / operation target, thereby improving the operation capability and operation safety in complex, multi-obstacle and narrow environments. Its safety includes both active safety of avoiding collision between the propeller and the surrounding environment or collision with the operating mechanism, resulting in propeller crash, and passive safety of avoiding damage to the environment, especially the operation target, due to the propeller.

[0255] (2) In terms of operating mechanism, in the existing technology, the end effectors of flying operating robots are mainly of three types: mechanical claws (including rigid and soft), suction cups, and magnetic suction. They can only match operating targets of specific shapes / materials. Among them, mechanical claws can only be used for targets such as small balls and round rods. The target size is limited by the opening diameter of the mechanical claws. Suction cups or magnetic suction require the target surface to be smooth or magnetic, which greatly limits its universality and versatility in real environments. The bionic adhesive actuator used in this application has the following beneficial effects: First, the bionic adhesive actuator is based on a μm-level silicone rubber-based microstructure array design, which can achieve reversible adhesion / detachment grasping of operating targets of various shapes / material surfaces through microscopic intermolecular forces (van der Waals forces), thereby improving the universality and versatility of the operating objects; second, it is light in weight, has no complex mechanical structure, and does not require external energy drive. During the operation process, it can reduce the disturbance torque of the operating mechanism to a certain extent, while saving the effective load of the flight platform and improving the operating load capacity.

[0256] In terms of control, the control of the flying operation robot mainly involves terminal trajectory control, underlying posture control, and contact (grasping) control.

[0257] (1) In terms of terminal trajectory control, the flying operation robot is a multi-degree-of-freedom redundant system, including the four degrees of freedom of the flight platform (longitudinal, lateral, vertical, and yaw) and the n degrees of freedom of each joint of the operating mechanism. On the one hand, the coupling effect between the degrees of freedom is strong, and the dynamic characteristics of each degree of freedom of the flying platform are significantly different from the dynamic characteristics of each degree of freedom of the operating mechanism. On the other hand, the operation process of the flying operation robot faces multiple environmental or system constraints such as operating environment limitations, actuator performance limitations, and self-collision of the operating mechanism. For the operating task target in three-dimensional space, the terminal trajectory control module needs to reasonably allocate the redundant degrees of freedom of the flying platform and the operating mechanism to drive the end effector to the desired target position efficiently, quickly, and accurately, while avoiding collisions between the flying platform and the environment / target or between the flying platform and the operating mechanism, and minimizing the mutual influence between the flying platform and the operating mechanism. Existing technologies mainly include independent two-stage control and closed-loop inverse kinematics control. First, independent two-stage control, or "hover first, then maneuver," involves controlling the flying robot to hover in front of or above the target and then controlling the manipulator to perform tasks such as grasping. This approach suffers from four drawbacks: First, the motion ranges of the flying platform and the manipulator must be determined offline. This approach is only suitable for tracking fixed targets and cannot effectively handle time-varying three-dimensional target trajectories. For example, the same control strategy cannot be used for forward and backward flight conditions. Second, the relative position of the target and the flying robot must be known, and the hovering position of the flying platform must be within the working space of the manipulator. Otherwise, the manipulator's joints may be in extreme positions or even suffer from singular or unsolvable inverse kinematics, leading to dangerous situations. Third, for a given three-dimensional target, the flying platform and the manipulator perform their respective trajectory tracking in stages, resulting in slow overall system response time. Fourth, the manipulator controls the flying platform as a fixed base. However, the flying platform inevitably floats during actual hovering, and hovering errors can reduce the tracking accuracy of the end effector. ② Closed-loop inverse kinematics control, which achieves "open-loop allocation and closed-loop tracking" of the end-effector trajectory by solving the inverse Jacobian matrix, has two disadvantages: First, this method cannot consider the actual dynamic response characteristics of the system during the solution process. Since the dynamic characteristics of each degree of freedom of the flight platform are significantly different from those of the operating mechanism, direct inversion may cause oscillation of the actual trajectory, making it difficult to ensure the smoothness and accuracy of the end-effector trajectory response; second, this method cannot consider various constraints such as the environment and actuator performance limitations during the solution process, which may cause dangers such as collisions between the flight platform and the environment / target or between the flight platform and the operating mechanism.The terminal trajectory integrated control module proposed in this application has the following beneficial effects: First, it realizes the integrated real-time optimal allocation of redundant degrees of freedom taking into account the different dynamic characteristics of the flight platform and the operating mechanism, improves the terminal trajectory tracking efficiency, and at the same time ensures the tracking accuracy and smoothness under the dynamic time-varying conditions of the desired task trajectory; Second, it takes into account various constraints such as the environment and actuator performance limitations, and for the first time considers the influence of the duct wall effect in the terminal trajectory control of the flying operating robot, thereby improving the safety of operations in complex multi-obstacle constraint environments.

[0258] (2) In terms of low-level posture control, relative motion is inevitable between the flight platform and the operating mechanism during the operation process. Due to the coupling effect between the two, on the one hand, the static and dynamic force changes caused by the operating mechanism will bring huge interference to the flight platform, resulting in a significant change in the state of the flight platform, causing the flight platform's posture oscillation and a decrease in tracking accuracy; on the other hand, the flight platform's posture changes and speed and acceleration fluctuations will in turn lead to a decrease in the control accuracy of the operating mechanism. In the existing technology, the operating mechanism is regarded as an external disturbance in the flight platform control, and the flight platform control accuracy is improved to a certain extent through the anti-disturbance control algorithm; in the control of the operating mechanism, either the flight platform coupling effect is ignored, or only the kinematic coupling effect between the flight platform and the operating mechanism is considered, that is, only the influence of the flight platform position and posture changes on the operating mechanism control is considered, without considering the influence of the flight platform speed and acceleration oscillation, resulting in limited control accuracy. The flight platform and operating mechanism posture control module proposed in this application has the following beneficial effects: in the posture control of the flight platform and the operating mechanism, the dynamic coupling between them is fully considered and real-time compensation is achieved, including position, attitude, linear velocity, angular velocity, linear acceleration, angular acceleration, as well as coupling force and torque. Among them, the influence of the flight platform posture dynamics on the robotic arm joint tracking control is considered for the first time, thereby improving the control accuracy of the robot.

[0259] (3) In terms of contact (grasping) control, the existing technology realizes grasping operations based on mechanical claws, suction cups, magnetic suction, etc. The robot control process only considers trajectory tracking and posture control, and does not consider the physical interaction dynamic characteristics of the end effector and the work target; In addition, the existing technology discloses the use of adhesion mechanisms to achieve drone perching, but on the one hand, the perching object is a static wall / ceiling, etc., rather than a dynamic target, and on the other hand, it does not consider the physical interaction dynamic characteristics of the adhesion mechanism and the perching object. The adhesion mechanism realizes grasping through intermolecular forces (van der Waals forces). The anisotropic characteristics caused by the microstructure of the adhesive material, and the deformation or even partial collapse of the microstructure cause the macroscopic adhesion / desorption mechanical characteristics of the robot to change non-uniformly with the oscillation of the flight platform and the time-varying pressure load on the contact surface, causing transient characteristics such as insufficient adhesion, saturation, and failure during the physical interaction process, resulting in a decrease in adhesion efficiency during the grasping process and even damage to the end adhesion actuator. The end-point physical contact control module proposed in this application has the following beneficial effects: First, it establishes and considers spatial constraints on adhesion performance during the grasping operation for the first time, achieving safe and smooth grasping of the target. This not only avoids the risk of damage and instability to the robot itself caused by collisions and impacts during conventional rigid grasping, but also avoids the risk of damage to fragile targets. Second, the robot's overall control architecture is simple, eliminating the need to design and switch independent control laws for the flight approach and contact grasping processes. Instead, a single control system is required to control both the flight approach and contact grasping processes. Furthermore, the introduction of the end-point physical contact control module does not compromise the end-point trajectory tracking speed and accuracy.

Claims

1. An adhesive ducted flying robot, characterized in that: include: Airframe, flight platform, operating mechanism and integrated controller; The flight platform, operating mechanism and integrated controller are all installed on the fuselage; The flight platform and the operating mechanism are both connected to an integrated controller; wherein the integrated controller includes: an operating task planning module, an end trajectory integrated control module, a flight platform posture control module, an operating mechanism posture control module, and an end physical contact control module; The operation task planning module is used to identify the operation task target according to the ground station instructions and plan the expected task trajectory of the operation mechanism in three-dimensional space according to the operation task target position; The terminal trajectory integrated control module is used to determine the flight platform position, yaw angle reference value, and reference values ​​of each joint angle of the manipulator based on the desired mission trajectory through a nonlinear model, so as to integrate the redundant degrees of freedom of the flight platform and the operating mechanism and achieve smooth tracking of the time-varying desired mission trajectory by the operating mechanism. The nonlinear model is a mathematical model determined by a sequential quadratic programming algorithm based on the kinematic and dynamic models corresponding to the flight operating robot to solve constrained nonlinear programming problems. The nonlinear model includes constraint conditions and an optimization objective function. The flight platform attitude control module is used to estimate and compensate the coupling effect of the operating mechanism on the flight platform in real time based on the flight platform position and yaw angle reference value, so as to ensure the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process; The operating mechanism posture control module is used to compensate for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform in real time based on the reference values ​​of the joint angles of the manipulator, so as to ensure the control of the manipulator joints under the posture changes of the flight platform during the operation process; The end physical contact control module is used to perform compliant grasping of the task target through admittance control within the adhesion performance space of the operating mechanism after the operating mechanism reaches the task target position according to the instructions of the ground station to complete the task.

2. The adhesive-type ducted flying working robot according to claim 1, characterized in that: The operating mechanism includes a 4-DOF serial robot arm and an end-adhesive actuator; Among them, the 4-DOF serial robot arm includes waist joint, shoulder joint, elbow joint and wrist joint; The end-adhesive actuator is a bionic silicone rubber-based adhesive array; the end-adhesive actuator can reversibly adhere / detach and grasp task targets of various shapes / material surfaces.

3. The adhesive-type ducted flying working robot according to claim 1, characterized in that: The flight platform contains four ducted propellers.

4. A control method for an adhesive ducted flying robot, characterized in that: The method is implemented by using the adhesive type ducted flying working robot according to any one of claims 1 to 3; the control method of the adhesive type ducted flying working robot comprises: Get ground station instructions; Based on the task planning module, the task target is identified according to the ground station instructions, and the expected task trajectory of the operation mechanism in three-dimensional space is planned according to the task target position; Based on the desired mission trajectory, the terminal trajectory integrated control module determines the flight platform position, yaw angle reference value, and reference values ​​of each joint angle of the manipulator through a nonlinear model, so as to integrate the redundant degrees of freedom of the flight platform and the operating mechanism and achieve smooth tracking of the time-varying desired mission trajectory by the operating mechanism. The nonlinear model is a mathematical model determined by a sequential quadratic programming algorithm based on the kinematic and dynamic models corresponding to the flight operating robot to solve constrained nonlinear programming problems. The nonlinear model includes constraint conditions and optimization objective functions. Based on the flight platform's position and yaw angle reference values, the flight platform's posture control module estimates and compensates for the coupling effect of the operating mechanism on the flight platform in real time to ensure the stability and tracking performance of the flight platform under the relative motion of the operating mechanism during the operation process. The operating mechanism posture control module uses the reference values ​​of the joint angles of the manipulator to perform real-time compensation for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform, ensuring the control of the manipulator joints under the posture changes of the flight platform during the operation. The end physical contact control module is used according to the ground station instructions. After the operating mechanism reaches the task target position, the operating task target is compliantly grasped through the admittance control within the adhesion performance space of the operating mechanism to complete the task.

5. The control method of the adhesive ducted flying working robot according to claim 4, characterized in that: Constraints include: duct safety distance limit, robot arm self-collision avoidance limit, robot arm joint angle range limit, robot movement speed limit, and robot movement acceleration limit; The expression corresponding to the duct safety distance limit is: |d a,x (k+i|k)|≥d a,min ; The expression corresponding to the robot's self-collision avoidance limit is: The expression corresponding to the robot arm joint angle range limit is: q j,min ≤q j (k+i|k)≤q j,max ; The expression corresponding to the robot movement speed limit is: The expression corresponding to the robot motion acceleration limit is: The optimization objective function includes the cost function corresponding to the task trajectory tracking error, the cost function corresponding to the change of the center of gravity position of the manipulator, and the cost function corresponding to the duct wall effect; The cost function corresponding to the task trajectory tracking error is: The cost function corresponding to the change in the center of gravity position of the robotic arm is: The cost function corresponding to the duct wall effect is: Among them, d a,x is the distance between the center of the flight platform and the target or obstacle; d a,min is the minimum safe distance between the center of the flight platform and the target or obstacle; k is the sampling time; i is the sequence number; d e,min is the minimum safe distance between the end effector and the center of the flight platform; is the position of the end-adhesive actuator relative to the flight platform; x is the independent state variable of the flight operation robot; q j is the j-th robotic arm joint angle; q j,min is the minimum angle limit of the joint angle; q j,max is the maximum angle limit of the joint angle; is the motion speed corresponding to the horizontal coordinate of the flight platform position at the k+i sampling moment; is the minimum motion speed corresponding to the horizontal coordinate of the flight platform position; is the maximum movement speed corresponding to the horizontal coordinate of the flight platform position; is the motion speed corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum motion speed corresponding to the vertical coordinate of the flight platform position; is the maximum movement speed corresponding to the vertical coordinate of the flight platform position; is the motion speed corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum movement speed corresponding to the vertical coordinate of the flight platform position; is the maximum movement speed corresponding to the vertical coordinate of the flight platform position; is the angular velocity corresponding to the yaw attitude angle at the k+i sampling moment; is the minimum angular velocity corresponding to the yaw attitude angle; is the maximum angular velocity corresponding to the yaw attitude angle; is the angular velocity corresponding to the joint angle of the manipulator at the k+i sampling moment; is the minimum angular velocity corresponding to the joint angle of the robotic arm; is the maximum angular velocity corresponding to the joint angle of the robotic arm; is the motion acceleration corresponding to the horizontal coordinate of the flight platform position at the k+i sampling moment; is the minimum motion acceleration corresponding to the horizontal coordinate of the flight platform position; is the maximum motion acceleration corresponding to the horizontal coordinate of the flight platform position; is the motion acceleration corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the maximum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the motion acceleration corresponding to the vertical coordinate of the flight platform position at the k+i sampling moment; is the minimum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the maximum motion acceleration corresponding to the vertical coordinate of the flight platform position; is the angular acceleration corresponding to the yaw attitude angle at the k+ith sampling moment; is the minimum angular acceleration corresponding to the yaw attitude angle; is the maximum angular acceleration corresponding to the yaw attitude angle; h1 is the cost function corresponding to the task trajectory tracking error; N p is the prediction time domain; p e is the position of the end effector of the flight operation robot; p e,d is the expected task trajectory; W1 is the tracking error weight matrix; W s1 is the tracking error terminal weight matrix; h2 is the cost function corresponding to the change of the center of gravity position of the manipulator; e m,xy is the prediction model function of the center of gravity position change of the manipulator; W2 is the weight matrix of the center of gravity position change of the manipulator; W s2 is the terminal weight matrix of the manipulator's center of gravity position change; h3 is the cost function corresponding to the duct wall effect; C WE is the duct wall effect induced tilting moment coefficient; W3 is the duct wall effect weight value; W s3 is the terminal weight value of the duct wall effect.

6. The control method of the adhesive ducted flying working robot according to claim 4, characterized in that: The mathematical expression of the kinematic model corresponding to the flying operation robot is: The mathematical expression of the dynamic model corresponding to the flying operation robot is: in, is the speed of the end effector at time t; Φ a (t) is the attitude angle of the flight platform at time t; q(t) is the joint angle of the manipulator at time t; is the first-order derivative of x at time t; J tx,e (Φ a (t),q(t)) is J t,e (Φ a ,q) related to x; J t,e (Φ a ,q) is the Jacobian matrix of the end effector; x is the independent state variable of the flight operation robot; M(ξ) is the inertia matrix; is the Coriolis force and centrifugal force matrix; g(ξ) is the gravity matrix; μ u is the generalized input force; μ ext is the generalized external force; ξ is the state variable of the flying operation robot; is the first-order derivative of ξ; is the second-order derivative of ξ.

7. The control method of the adhesive ducted flying working robot according to claim 4, characterized in that: The corresponding expression for the constrained nonlinear programming problem is: Among them, J(x(k),u k ) is the system cost function; h j is the optimization objective; u k is the control input vector to be optimized; x(k) is the system state at sampling time k; j is the optimization target number; m is the total number of optimization targets; x(k+i|k) is the state of the system at sampling time k+i predicted by sampling time k; u(k+i) is the control input at sampling time k+i; x(k+i+1|k) is the state of the system at sampling time k+i+1 predicted by sampling time k; T s is the sampling time; x(k|k) is the intermediate variable corresponding to x(k); g(x(k+i|k),u(k+i)) is the general constraint function related to the system state variables and control variables; g min is the upper bound of the system's general constraints; g max is the lower bound of the system's general constraint; x min is the upper constraint boundary of the system state variable; x max is the lower constraint boundary of the system state variable; u min u is the upper constraint boundary of the system control variable; max is the lower constraint boundary of the system control variable; Δu min is the upper constraint boundary of the system control variable increment; Δu max Set the lower bound for the increment of the system control variable.

8. The control method of the adhesive ducted flying working robot according to claim 4, characterized in that: Real-time estimation and compensation of the coupling effect of the operating mechanism on the flight platform is achieved based on the update law of the operating mechanism coupling effect estimation and the control output of the operating mechanism coupling compensation; The update law of the estimated coupling effect of the operating mechanism is expressed as follows: The control output of the working mechanism coupling compensation corresponds to the expression: in, is the update law for estimating the coupling effect of the operating mechanism; Γ ad is the adaptive gain; Proj(·,·) is the projection operator; are the estimated values ​​of the coupling force and coupling torque exerted by the operating mechanism on the flight platform; is the estimation error of the flight platform motion state variables; P ad is the solution of the system algebraic Lyapunov equation; T is the transpose; u ad (s) is the control output considering the coupling compensation of the operating mechanism in the frequency domain; C(s) is a low-pass filter; K des is the intermediate variable; a,r (s) is the reference input y a,r Laplace form of F dis (s) is the intermediate variable in Laplace form; for Laplace form of ; s is the Laplace operator.

9. The control method of the adhesive ducted flying working robot according to claim 6, characterized in that: The operating mechanism posture control module uses the reference values ​​of the joint angles of the manipulator to perform real-time compensation for the nonlinear coupling between the joints of the manipulator and between the manipulator and the flight platform to ensure the control of the manipulator joints under the posture changes of the flight platform during the operation. Specifically, it includes: The joint tracking error of the manipulator is determined based on the reference values ​​of each joint angle of the manipulator, and the PD method is used for feedback control. The mathematical expression corresponding to the feedback control is: According to the dynamic model corresponding to the flying operation robot, the dynamic model of the flying operation robot's manipulator subsystem is obtained, and the inverse dynamics compensation is determined; The mathematical expression corresponding to inverse dynamics compensation is: Based on feedback control and inverse dynamics compensation, the control of the manipulator joint under the change of the flight platform's posture during the operation is determined, and the manipulator joint control input is obtained; the expression corresponding to the manipulator joint control input is: u m (t)=u ff (t)+u fb (t); Among them, u fb (t) is the feedback control at time t; K p,q is the proportional gain matrix; K d,q is the differential gain matrix; e q (t) is the tracking error of the robot arm joint; for e q The derivative of (t); u ff (t) is the inverse dynamics compensation at time t; M m is the simplified inertia matrix of the manipulator subsystem; ξ(t) is the state variable of the flight operation robot at time t; is the second derivative of the desired reference angle of the manipulator joint; C m It is the name of the Coriolis force and centrifugal force matrix of the manipulator subsystem; is the first-order derivative of ξ(t); is the differential of the desired reference angle of the manipulator joint; g m It is the name of the gravity matrix of the manipulator subsystem, which is equivalent to g m (ξ); M a It is the abbreviation of the coupling force and coupling torque exerted by the flight platform on the robotic arm; is the second-order derivative of the flight platform state variable ζ(t) at time t; u m (t) is the robot arm joint control input at time t.

10. The control method of the adhesive ducted flying working robot according to claim 4, characterized in that: By controlling the admittance within the adhesion performance space of the working mechanism, the reference input of the working mechanism's posture control module is updated, and the working task target is compliantly grasped based on the reference input to complete the working task. The expression for the reference input is: q r,d (t)=q r (t)+Δq r (t); The expression corresponding to the adhesion performance space is: Among them, q r,d (t) is the reference input at time t; q r (t) is the desired reference angle of the manipulator joint; Δq r (t) is the equivalent expected increment of the joint angle of the manipulator; Ω is the adhesion performance space; f ext,n f ext The normal component of ext,t f ext The tangential component of ext is the physical contact force of the end-adhesive actuator; m ext is the torque; F sat is the adhesion saturation boundary; F pre is the minimum preload boundary; F detach is the desorption force boundary; M detach is the desorption moment boundary; τ is the shear stress at the root of the adhesion microstructure at the contact interface; τ max is the microstructural failure boundary.

Images