Airborne work robot with adaptive admittance control and control method and system thereof

Abstract

The application discloses an aerial operation robot with adaptive admittance control and a control method and system thereof, the control method regards a full-drive type aerial operation robot as an integrated aerial actuator, and then an unmanned aerial vehicle-mechanical arm integrated system model is constructed; an adaptive admittance controller is introduced as an outer loop force controller, an adaptive admittance control algorithm based on Lyapunov stability theory is designed; and a nonlinear model predictive control method based on the integrated model is designed, which can constrain the sliding force output of the platform. Therefore, by using the motion control method of the full-drive type aerial robot platform and the adaptive admittance control algorithm, the end sliding operation mechanism and the contact operation surface are ensured to smoothly slide, and the aerial precise sliding control and force interaction operation in a complex unknown environment are realized.

Description

Technical Field

[0001] This invention belongs to the field of robot control technology, specifically relating to an aerial work robot with adaptive admittance control and its control method and system. Background Technology

[0002] As the scale and complexity of large-scale infrastructure projects such as wind turbine blades, bridge cables, and oil and gas pipelines continue to increase, the efficient and precise inspection and maintenance of aerial robots is becoming an important industry requirement. Compared with traditional manual inspection or heavy ground equipment, aerial robots have outstanding spatial mobility, flexible operating range, and rapid response capabilities, making them particularly suitable for inspection tasks of high-altitude and long-span structures.

[0003] Early aerial inspection primarily relied on airborne vision or infrared sensors for non-contact scanning. While efficient to some extent, its accuracy and reliability often fell short of practical engineering requirements when identifying critical defects such as surface microcracks and internal material damage. Therefore, sliding inspection methods, which enable continuous contact between the sensor and the structural surface, have become a more promising technological approach. By maintaining stable contact, these methods can significantly improve the ability to identify hidden defects and enhance measurement accuracy.

[0004] However, achieving stable and continuous sliding contact operations for aerial robots on large, undefined, and time-varying curved surfaces remains a significant challenge. During contact, robots are susceptible to surface topography changes and environmental disturbances. Traditional control methods struggle to ensure continuous sensor contact and maintain stable contact force, impacting the reliability and consistency of detection data. To address force control, mainstream solutions typically employ a "decoupling" approach, mounting a multi-degree-of-freedom robotic arm on the aerial robot. This confines the complex force control task to the robotic arm, while the flight platform only performs positioning control, treating the robotic arm as an external disturbance. However, this approach inevitably increases system complexity, overall mass, and rotational inertia, negatively impacting the flight platform's dynamic performance and energy consumption.

[0005] Therefore, it is necessary to develop an integrated control strategy suitable for fully driven aerial robots to support them in achieving accurate, stable, and coherent sliding contact detection operations on complex curved surfaces. In particular, under conditions where the position and stiffness of the environment are unknown and dynamically changing, it is still possible to maintain a stable and accurate end-effector interaction force while ensuring the stability of the robot's own flight attitude. Summary of the Invention

[0006] To address the challenge of enabling all-wheel-drive aerial robots to maintain a stable and precise end effector force while performing continuous sliding detection on curved surfaces under conditions of unknown and dynamically changing environmental position and stiffness, and to ensure stable flight attitude, thereby completing high-quality contact detection tasks, this invention provides an aerial operation robot with adaptive admittance control, along with its control method and system.

[0007] Therefore, the following technical solution is provided:

[0008] On one hand, the present invention provides an adaptive admittance control method for aerial work robots, applied to all-wheel-drive aerial work robots. The aerial work robot control method includes the following steps:

[0009] Step 1: Treat the all-wheel-drive aerial work robot as an integrated aerial actuator, and then construct an integrated drone-robotic arm system model;

[0010] Step 2: Introduce an adaptive admittance controller as the outer loop force controller in the UAV-robotic arm integrated system model. Based on the adaptive admittance controller, and according to the actual contact force... With Expectation deviation Dynamically adjust admittance parameters This allows for the correction of the pose to obtain an updated reference trajectory;

[0011] Step 3: Input the reference trajectory and cooperative state variables of the aerial work robot into the inner loop motion controller in the UAV-robotic arm integrated system model to obtain the optimal control command for the aerial work robot.

[0012] Optionally, the control relationship of the adaptive admittance controller is as follows:

[0013]

[0014] in: It is a positional error. This is the actual pose. For reference trajectory; It is force tracking error. The actual contact force measured by the end force sensor. For the power of expectation; , , For the desired admittance parameter, Position error The derivative, for The derivative;

[0015] The adaptive admittance controller adjusts the admittance parameters based on the following adaptive law. :

[0016]

[0017] In the formula, It is an adaptive gain, where T is the symbol for the matrix device. Adjusting admittance parameters for adaptive laws The derivative of .

[0018] Optionally, for the adaptive admittance controller, Lyapunov stability analysis is introduced to determine the matrix. and adaptive gain The possible values ​​are as follows:

[0019] matrix It is a diagonal matrix, and the diagonal element b d,ii satisfy:

[0020]

[0021] Among them, parameters , The minimum pose error norm is preset according to the required tracking accuracy. The minimum damping value is greater than 0. For safety margin, This is a positive scalar parameter, representing the desired dynamic performance of the closed-loop system. It is an unknown ideal constant and has stiffness similar to that of the real environment. The relevant parameters are preset values;

[0022] Adaptive gain The value requirements are as follows:

[0023]

[0024] In the formula, for The lower bound margin, for The upper limit margin, The maximum permissible position error squared is preset according to the required tracking accuracy. This indicates finding the smallest eigenvalue of a matrix.

[0025] Optionally, the dynamic model of the integrated UAV-robotic arm system is as follows:

[0026]

[0027] in, For the system's inertia matrix terms, The system's Coriolis force and centrifugal force matrices are given. The gravity vector The environmental force / torque is measured in the end-effector coordinate system, which means that the environment acts on the end of the robot arm and is then transmitted to the entire system. To map the end force to the Jacobian matrix in the generalized force space, For generalized control input vector, , , These represent the driving force and torque of the drone body and the rotational torque of the driving robotic arm, respectively, with q being the position vector. constitute, The location of the drone in the world coordinate system. The attitude angles of the UAV in the UAV body coordinate system. The joint angles of the robotic arm in the drone's body coordinate system; For the system's generalized velocity, This represents the UAV attitude in the world coordinate system, where W is the world coordinate system identifier. Let ω be the angular velocity of the UAV in the UAV's body coordinate system. for The derivative of B, where B is the coordinate system identifier for the UAV body. It is the set of real numbers.

[0028] Optionally, the inner-loop motion controller is a nonlinear model predictive control (NMPC), and the state variable X input to the NMPC controller is defined as:

[0029] ,in, , This represents the UAV's position and attitude in the world coordinate system, where W is the world coordinate system identifier. , The attitude angles and angular velocities of the UAV in the UAV body coordinate system are given. The angles of the robotic arm joints are shown in the drone's body coordinate system. for The derivative of B, where B is the coordinate system identifier for the UAV body. It is the set of real numbers;

[0030] The output of the NMPC controller is the generalized control input vector of the aerial robot: ,in, These are the driving force and torque of the drone body and the rotational torque of the driving robotic arm, respectively.

[0031] The optimization objective function of the NMPC controller is:

[0032]

[0033] Obey:

[0034]

[0035] In the formula, The system dynamics constraints are derived from the system dynamics model (dynamics). Changes in the system state variables must conform to the system dynamics model (dynamics), which is the dynamics model of the integrated UAV-robotic arm system. For state constraints, Input constraints, initial state Set as the current system state , The weight matrices represent the state, input, and terminal state, respectively. The NMPC controller generates the optimal control sequence by solving the optimization objective function in real time, where t is time and T is the predicted time domain endpoint.

[0036] Optionally, the discrete update formula for the reference trajectory in step 2 is as follows:

[0037]

[0038] In the formula, k represents the time or the number of iterations. These are the reference trajectory and the expected trajectory, respectively. , , For the desired admittance parameter, Position error The derivative, for The derivative, The actual contact force measured by the end force sensor. For the power of expectation.

[0039] Secondly, the present invention also provides an aerial work robot based on the above-mentioned aerial work robot control method. The aerial work robot is an all-wheel drive aerial work robot and is equipped with at least a body, a power unit, a rotating mechanism, a robotic arm, an end-effector sliding mechanism, and a control system.

[0040] The power unit consists of six tilting rotors. The central part of the rotating mechanism is fixed to the center of the UAV body, while the outer edge part is rotatable. The robotic arm is connected to the outer edge part and rotates around the center of the body. The robotic arm adopts a single degree of freedom design and is driven by a servo motor. Its proximal end is a rigid arm, and its tail end is connected to an end-effector sliding mechanism for contact with the environment.

[0041] The control system uses the optimal control commands generated in steps 1-3 to control the aerial work robot.

[0042] Optionally, each rotor is installed at a fixed tilt angle of 30°; the plane of motion of the robotic arm is perpendicular to the plane formed by the rotor axes and is offset from the rotors and rotor axes during movement.

[0043] Optionally, the control system includes an onboard computer and a flight controller mounted on the fuselage base;

[0044] The airborne computer controls the servo motor that drives the rotation mechanism via a serial port, and the flight controller generates rotor speed commands and sends them to the electronic speed controller, thereby driving the rotor motor.

[0045] The onboard computer is equipped with a processor and a memory; the processor calls the computer program stored in the memory to obtain the optimal control command in steps 1-3 and transmits it to the flight controller.

[0046] Thirdly, the present invention also provides a control system based on the above-mentioned aerial work robot control method, comprising:

[0047] The dynamics model building module is used to treat the all-wheel drive aerial work robot as an integrated aerial actuator, and then build an integrated drone-robotic arm system model.

[0048] An adaptive admittance control module is used to introduce an adaptive admittance controller as an outer loop force controller. Based on the adaptive admittance controller, the actual contact force is adjusted. With Expectation deviation Dynamically adjust admittance parameters This allows for the correction of the pose to obtain an updated reference trajectory;

[0049] The inner loop motion control module is used to input the reference trajectory and cooperative state variables of the aerial work robot into the inner loop motion controller to obtain the optimal control command for the aerial work robot.

[0050] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that is invoked by a processor to implement the steps of the above-described aerial work robot control method.

[0051] The technical advantage of this invention lies in its innovative design of an integrated UAV-robotic arm all-drive platform, combined with a deep collaborative architecture of predictive control and adaptive admittance control, achieving a significant breakthrough in high-precision force / position coordinated control in complex environments. Compared to traditional rigid robotic arm systems, this invention effectively solves the problems of strong leverage effect and dynamic coupling interference caused by the robotic arm's motion on the UAV platform. Through adaptive admittance law, the system can adjust its admittance stiffness characteristics online, automatically adapting to changes in unknown environmental parameters, significantly improving force tracking accuracy and system stability in continuous sliding detection tasks. This rigid-flexible coupled intelligent control strategy not only ensures the attitude stability of the UAV platform during interaction but also achieves precise and compliant contact of the end effector with curved surfaces, providing a safe and reliable technical solution for high-altitude operations such as infrastructure inspection and wind turbine blade maintenance. Attached Figure Description

[0052] Figure 1 This is a structural schematic diagram of an aerial contact robot in an embodiment of the present invention, wherein 1 is an end-effector sliding operation mechanism, 2 is a robotic arm, 3 is a flight controller, 4 is a rotating mechanism, 5 is a propeller, 6 is an electronic speed controller, 7 is a brushless motor, 8 is a servo motor, and 9 is the body.

[0053] Figure 2 A flowchart illustrating the method in this embodiment of the invention;

[0054] Figure 3 This is a flowchart of the calculation process of the method in the embodiments of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. The technical features involved in the various embodiments of the invention described below can be combined with each other as long as they do not conflict with each other.

[0056] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0058] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0059] This invention abandons the traditional approach of attaching a robotic arm to a drone, directly utilizing the inherent force control potential of the fully driven flight platform itself. The core lies in treating the entire fully driven aerial robot as a single integrated aerial actuator. Predictive control is used to explicitly handle the highly nonlinear dynamic model of the fully driven system, achieving the high-precision trajectory tracking required for slip detection. The core of this invention is the design of adaptive admittance control to handle the interactive force control between the robot and its environment. By adjusting admittance parameters online, it adapts to unknown environments, ensuring the accuracy and stability of force tracking. Furthermore, it preferably incorporates NMPC (Nonlinear Model Predictive Control), which handles the robot's motion control, achieving high-precision trajectory tracking while considering the system's nonlinear dynamics and constraints.

[0060] like Figure 1 As shown, the all-drive aerial work robot is an innovative design based on a traditional six-rotor platform. Its core structure includes a fuselage 9, a power unit, a rotating mechanism 4, a robotic arm 2, an end-effector sliding mechanism 1, and a control system. The power unit consists of six tilting rotors and brushless motors 7, with each rotor mounted at a fixed 30° tilt angle. The central part of the rotating mechanism 4 is fixed to the center of the drone fuselage, while the outer edge is rotatable. The robotic arm 2 is connected to the outer edge and rotates around the center of the fuselage. The plane of motion of the robotic arm is perpendicular to the plane formed by the rotor axes and remains offset from the rotors and rotor axes during movement to avoid interference. The robotic arm 2 adopts a single-degree-of-freedom design and is driven by a servo motor 8. Its proximal end is a rigid arm, and its tail end is connected to the end-effector sliding mechanism 1 for contact with the environment. In terms of communication, the onboard computer 10 installed in the fuselage is responsible for processing high-level control signals and controlling the servo motors driving the rotating structure via a serial port. It also sends control commands to the flight controller 3 via a network port to process low-level flight signals. The rotor speed command generated by the flight controller is sent to the electronic speed controller 6, which in turn drives the rotor motors. The core of the aerial work robot structure described above is the fully driven flight platform and the single-degree-of-freedom rotating robotic arm connected to the center of the fuselage. The system establishes an integrated dynamic model and preferably integrates a nonlinear model predictive control (NMPC) module and an adaptive admittance control module. Its controlled object is the entire robot system, including six rotors and the rotating arm. Specifically, it achieves precise trajectory tracking through NMPC and uses adaptive admittance control to adjust parameters online to ensure accurate tracking of the interaction force.

[0061] From a hardware perspective, the control system of this embodiment includes an onboard computer mounted on a base for processing high-level control signals, and a flight controller for processing low-level flight signals. The command rotor speed signal generated by the flight controller 3 is transmitted to the electronic speed controller 6 to drive the motor and propeller. The onboard computer is equipped with a processor and a memory. The processor calls the computer program stored in the memory to implement the adaptive admittance control aerial robot control method proposed in this invention.

[0062] The following section will focus on a detailed explanation of control methods.

[0063] See Figure 2 and Figure 3 This embodiment provides an adaptive admittance control method for aerial work robots to control the aforementioned all-wheel drive aerial work robot, including the following steps:

[0064] Step 1: Treat the all-wheel-drive aerial work robot as an integrated aerial actuator, and then establish the system dynamic equations based on Lagrange mechanics:

[0065]

[0066] in, For the system's inertia matrix terms, The system's Coriolis force and centrifugal force matrices are given. The gravity vector The above four matrix expressions are all common expressions in this field, and therefore will not be described in detail. They represent the environmental forces / torques measured in the end-effector coordinate system, which are the forces and torques that the environment exerts on the end of the robotic arm and then transmits to the entire system. To map the end force to the Jacobian matrix in the generalized force space, For generalized control input vector, , , These represent the driving force and torque of the drone body and the rotational torque of the driving robotic arm, respectively, with q being the position vector. constitute, The location of the drone in the world coordinate system. The attitude angles of the UAV in the UAV body coordinate system. The joint angles of the robotic arm in the drone's body coordinate system; For the system's generalized velocity, This represents the UAV attitude in the world coordinate system, where W is the world coordinate system identifier. Let ω be the angular velocity of the UAV in the UAV's body coordinate system. for The derivative of B, where B is the coordinate system identifier for the UAV body. It is the set of real numbers.

[0067] The integrated dynamics model considers the robotic arm and the body as a whole, taking into account their interaction. It also actively controls the torque applied by the body to the robotic arm's rotation to handle sliding contact scenarios. It should be understood that the above model has a dimension of 7, unlike the conventional 6-dimensional model, as it additionally considers the single-degree-of-freedom interaction between the drone and the robotic arm. In this embodiment, the system dynamics equations are the mathematical model used for prediction by NMPC, set as a hard constraint in the NMPC optimization problem, and used to calculate the system dynamics constraints.

[0068] Control input interface definition:

[0069]

[0070] The first six components correspond to the six-dimensional force / torque control of the UAV, while the last component controls the rotation mechanism. Depend on constitute, Depend on constitute, Right now .

[0071] For the aforementioned dynamic model, a complete coordinate system is first established, including the world coordinate system, the UAV body coordinate system, the rotating mechanism coordinate system, the robotic arm coordinate system, and the end effector coordinate system. By analyzing the geometric relationships between the components, a complete kinematic chain from the UAV base to the end effector is established. Based on generalized coordinates such as the UAV pose, the rotating mechanism rotation angle, and the robotic arm joint angle, the position and attitude of the robotic arm's end effector in the world coordinate system are calculated. The key to this step is accurately describing the coupling relationship between the UAV motion, the rotating mechanism rotation, and the robotic arm motion, providing accurate pose information for subsequent control algorithms. Specifically:

[0072] First, a complete integrated UAV-robotic arm system model was established, and a world coordinate system was defined. UAV body coordinate system and end effector coordinate system The system's generalized coordinate X is represented as:

[0073]

[0074] The above formula, , This represents the UAV's position and attitude in the world coordinate system, where W is the world coordinate system identifier. , The attitude angles and angular velocities of the UAV in the UAV body coordinate system are given. Here, B represents the joint angle of the robotic arm in the drone's body coordinate system, and B is the marker for the drone's body coordinate system. It is the set of real numbers.

[0075] pose of the end effector in the world coordinate system Calculated using forward kinematics:

[0076]

[0077] in, This represents the position of the end effector in the world coordinate system. The attitude of the end effector in the end effector coordinate system. It is a positive kinematic function.

[0078] Velocity of the end effector in its own coordinate system The relationship with system speed is as follows:

[0079]

[0080] in, Let T be the generalized velocity of the system, and T be the matrix transpose symbol. for The derivative, Let E be the Jacobian matrix in the end-effector coordinate system, and E be the end-effector coordinate system identifier. The actual pose of the end-effector is known in real-time during actual calculations. and actual speed It can determine the interaction with the environment and generate the correct expected trajectory, which is achievable with existing technology, so it will not be elaborated on in detail.

[0081] Step 2: Introduce an adaptive admittance controller as the outer loop force controller. Based on the adaptive admittance controller, and according to the actual contact force... With Expectation deviation Dynamically adjust admittance parameters This leads to the generation of pose correction commands. The pose correction command The trajectory superimposed on the original expected trajectory is used as the updated reference trajectory.

[0082] In this invention, an adaptive admittance controller is designed as the outer loop force controller. This controller is based on the actual contact force measured by the end force sensor. With Expectation deviation Dynamically adjust admittance parameters Then, by integration, we obtain Finally, pose correction commands can be generated based on the admittance model. The updated reference trajectory is obtained by superimposing it with the original desired trajectory. An adaptive law based on Lyapunov stability is used to adjust the admittance stiffness matrix online, enabling the system to automatically adapt to the stiffness of unknown environments. The pose correction amount (pose correction command) output by the admittance controller... After being superimposed with the original desired trajectory, it serves as a new reference input for NMPC, enabling force / position coordinated control. This ensures that the robot maintains stable contact force when interacting with environments of varying stiffness.

[0083] To address the problem of unknown and changing environmental stiffness, the basic idea of ​​the designed adaptive admittance controller is to respond to environmental forces by adjusting the robot's motion. The standard admittance control relationship is as follows:

[0084]

[0085] in: It is position error ( This is the actual pose. (for reference trajectory) It is force tracking error. The actual contact force measured by the end force sensor. For the power of expectation; , , For the desired admittance parameter, Position error The derivative, for The derivative of .

[0086] Environmental stiffness Fixed parameters in unknown situations Cannot guarantee Therefore, the objective of this invention is to design an adaptive law that adjusts online. To cope with unknown environmental stiffness .

[0087] The contact environment is simplified into a spring model. Furthermore, it can accurately determine the distance between the environment and the contact tip of the machine. Then the steady-state force error With position error pass and Related. But Unknown, therefore adjustments are needed. .

[0088] The technical solution of this invention is designed with the following adaptive law:

[0089]

[0090] Based on energy observation and dissipation principles. In adaptive laws... and Essentially, it is an observer of the exchange rate between the system's potential and kinetic energy. Through... and The product of these factors allows for early intervention even when an error trend emerges but the magnitude is still small.

[0091] The discretized representation of the adaptive law is as follows:

[0092]

[0093] The real-time solution for the adaptive admittance controller is expressed as:

[0094]

[0095] In the formula, k represents the time or the number of iterations. These are the reference trajectory and the desired trajectory, respectively. They are obtained through matrix inversion. Real-time calculation of pose corrections to obtain the reference trajectory Input into NMPC. It should be understood that in the above discrete formula... Considered as a pose correction command .

[0096] To ensure system stability, the design is based on stability theory, and the Lyapunov function V is chosen:

[0097]

[0098] The first term is kinetic energy, the second term is potential energy, and the third term is parameter error energy. It is parameter error. It is an unknown ideal constant and has stiffness similar to that of the real environment. Relatedly, the environmental stiffness of the actual working environment corresponds to a range of variation. In actual calculations, it can be set to a value greater than the entire range. It is the adaptive gain, and tr is the trace of the computation matrix.

[0099] right Find the time derivative:

[0100]

[0101] because It is a constant, parameter error derivative Substituting from the admittance equation :

[0102]

[0103] Simplified version:

[0104]

[0105] To ensure the overall stability of the system, it is necessary to ensure Observe the above formula, where... It is already negative semi-definite. In order to eliminate the uncertainty of the remaining terms,

[0106] Substituting the terms into the calculation, the terms generated by the adaptive law are:

[0107]

[0108] because It is a scalar, and Substituting into the above equation, we get:

[0109]

[0110] Adaptive law update term:

[0111]

[0112] Utilizing the cyclic property of traces:

[0113]

[0114] because symmetry:

[0115]

[0116] Summarized as follows:

[0117]

[0118] Expand :

[0119]

[0120] Under the passive assumption of the environment, Damping dissipation term ( (Zhengding), select It is a diagonal matrix, and the diagonal element b d,ii satisfy:

[0121]

[0122] in, , The expected minimum pose error norm, For the minimum damping value, For safety margin, The parameter is a positive scalar, representing the desired dynamic performance of the closed-loop system, and the adaptive gain. Design:

[0123]

[0124] in, for The lower bound margin, for The upper limit margin, The maximum permissible position error squared is preset according to the required tracking accuracy. This indicates finding the minimum eigenvalue of a matrix. Therefore, by choosing an appropriate eigenvalue... , and It can guarantee According to Lyapunov's stability theory, the system is globally asymptotically stable. In summary, this invention has proven the stability of the controller and obtained the range of values ​​for the above parameters under stable conditions. Among them, The range of values ​​is determined based on the actual task scenario.

[0125] Step 3: Input the state variables of the aerial work robot into the inner loop motion controller to obtain the optimal control command for the aerial work robot.

[0126] In this embodiment of the invention, an NMPC is preferably designed as the inner-loop motion controller. This controller takes state variables such as the UAV's position, attitude, rotation angle of the rotating mechanism, and joint angle of the robotic arm as inputs, and generates optimal control commands by solving a finite-time domain optimization problem. The controller explicitly considers the nonlinear dynamic characteristics of the system while handling physical constraints such as actuator saturation and state constraints. Within each control cycle, the NMPC continuously optimizes the control sequence for a future period based on the current state and the updated reference trajectory, and outputs the first term as the actual control quantity, achieving high-precision trajectory tracking and attitude stabilization.

[0127] The design of the NMPC controller defines a state vector, i.e., the system's generalized coordinate X:

[0128]

[0129] Define control input:

[0130]

[0131] Define the optimization objective function:

[0132]

[0133] Obey:

[0134]

[0135] in, These are weight matrices representing the overall system state, system control input, and terminal state, respectively. The terminal state refers to the state in the prediction time domain. The system state at this endpoint With reference terminal status To address the error between the two systems, NMPC solves this optimization problem in real time, generates the optimal control sequence, and directly controls the entire system.

[0136] Specifically, NMPC is discretized using the forward Euler method to discretize the state space:

[0137]

[0138] The period used is The discretized objective function is:

[0139]

[0140] This objective function balances the requirements of tracking performance, control cost, and system smoothness. Real-time optimization employs an efficient sequential quadratic programming algorithm for time-domain prediction. Control time domain N represents the optimization step size. The ACADO toolkit is used to solve the optimization control problem, where a fourth-order implicit Runge-Kutta integrator is used to simulate the system dynamics along the interval in the forward direction. Then, the boundary value problem is solved for each sampling interval. Within each control cycle, the algorithm resolves the optimization problem based on the current state to ensure the real-time performance and adaptability of the control.

[0141] In some embodiments, the present invention also provides a control system based on the above control method, which includes a dynamic model construction module, an adaptive admittance control module, and an inner loop motion control module connected in sequence or interconnected.

[0142] The dynamics model building module is used to treat the all-wheel-drive aerial robot as an integrated aerial actuator, thereby constructing an integrated drone-robotic arm system model.

[0143] An adaptive admittance control module is used to introduce an adaptive admittance controller as an outer loop force controller. Based on the adaptive admittance controller, the actual contact force is adjusted. With Expectation deviation Dynamically adjust admittance parameters This leads to the generation of pose correction commands.

[0144] Specifically, the trajectory obtained by superimposing the pose correction command with the original reference trajectory is used as the updated reference trajectory, which is a type of state variable input to the inner loop motion controller.

[0145] The inner loop motion control module is used to input the state variables of the aerial work robot into the inner loop motion controller to obtain the optimal control command for the aerial work robot.

[0146] Based on the provided process Figure 3 The control process of the all-wheel drive aerial work robot is as follows: The task planner first generates the desired pose trajectory. Interaction with expectations These parameters are then passed to the adaptive admittance controller. The adaptive admittance controller simultaneously receives the actual pose feedback from the all-wheel-drive aerial robot. and actual interaction force By comparing the actual force with the expected force (i.e., force error) And the actual pose and the desired pose, adjust the admittance parameters online. and output the corrected reference pose. and reference power The NMPC controller then performs rolling optimization calculations based on these reference values, the system dynamics model, and the current state to generate the optimal control input. The command is then sent to the all-wheel-drive aerial work robot for execution. After the robot moves according to the control input, it sends back the latest information. and By providing an adaptive admittance controller, a closed-loop control is formed, thereby achieving stable and accurate force / position coordination in unknown environments.

[0147] It should also be understood that the specific implementation process of each module is described in the above method. This invention will not repeat it here. The above division of functional modules is only for illustrative purposes. In some embodiments, some functional modules can be combined and some functional modules can be separated. Each functional module can be implemented in software, hardware, or a combination of software and hardware. The software and hardware devices include, but are not limited to, general-purpose computer equipment, programmable gate arrays, digital signal processors, microprocessors and their corresponding programming or burning software.

[0148] In some embodiments, the present invention also provides a computer-readable storage medium storing a computer program that is invoked by a processor to implement the steps of an aerial work robot control method with adaptive admittance control.

[0149] For details on the implementation of each step, please refer to the description of the control method embodiment above.

[0150] The readable storage medium is a computer-readable storage medium, which can be an internal storage unit of the hardware and software device described in any of the foregoing embodiments, such as the hard drive or memory of the controller. The readable storage medium can also be an external storage device of the controller, such as a plug-in hard drive, Smart MediaCard (SMC), Secure Digital (SD) card, or Flash Card equipped on the controller. Further, the readable storage medium can include both internal storage units and external storage devices of the controller. The readable storage medium is used to store the computer program and other programs and data required by the controller. The readable storage medium can also be used to temporarily store data that has been output or will be output.

[0151] Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0152] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This application refers to flowchart illustrations and / or instructions executed by a processor of a method, apparatus (system), and computer program product according to embodiments of this application to create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams. These computer program instructions may also be stored in a computer-readable storage medium capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowchart illustrations and / or one or more block diagrams. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more blocks of a block diagram.

[0153] It should be emphasized that the examples described in this invention are illustrative rather than limiting. Therefore, this invention is not limited to the examples described in the specific embodiments. Any other embodiments derived by those skilled in the art based on the technical solutions of this invention, without departing from the spirit and scope of this invention, whether modifications or substitutions, are also within the protection scope of this invention.

Claims

1. An aerial work robot control method of adaptive admittance control, characterized by: The application is applied to a full-drive type aerial work robot, and the aerial work robot control method comprises the following steps. Step 1: regarding the full-drive type aerial work robot as an integrated aerial executor, an unmanned aerial vehicle-robot arm integrated system model is constructed; Step 2: Introduce an adaptive admittance controller as the outer loop force controller in the UAV-robotic arm integrated system model. Based on the adaptive admittance controller, and according to the actual contact force... With Expectation deviation Dynamically adjust admittance parameters This allows for the correction of the pose to obtain an updated reference trajectory; Wherein, the Lyapunov stability analysis is introduced to determine the values of the adaptive admittance controller's matrix and adaptive gain , as follows: matrix is a diagonal matrix, and the diagonal elements b d,ii satisfy: ； wherein the parameters , is the expected minimum pose error norm, is a minimum damping value, taking a value greater than 0; is a safety margin, is a positive scalar parameter, is an unknown ideal constant and a parameter related to the real environment stiffness , is a pre-set value; Adaptive gain The value of the adaptive gain is required to be: ； wherein is a lower margin of is an upper margin of is a square of a maximum allowable position error, which is set in advance according to a required tracking accuracy, is an upper margin of is a square of a maximum allowable position error, which is set in advance according to a required tracking accuracy, denotes a minimum eigenvalue of a matrix; is a position error, is an actual pose, is a reference trajectory; Step 3: inputting the reference trajectory of the aerial work robot and the state variable into an inner loop motion controller in the unmanned aerial vehicle-robot arm integrated system model to obtain the optimal control instruction of the aerial work robot; the inner loop motion controller is a nonlinear model predictive control (NMPC), and the state variable X input into the NMPC controller is defined as: , is the position of the UAV in the world coordinate system, is the attitude angle of the UAV in the UAV body coordinate system, is the joint angle of the manipulator in the UAV body coordinate system; is the attitude of the UAV in the world coordinate system, W is the world coordinate system marker, is the angular velocity of the UAV in the UAV body coordinate system, B is the UAV body coordinate system marker, is the set of real numbers; The output of the NMPC controller is a generalized control input vector for the aerial work robot , , , are respectively the driving force and moment of the UAV body and the driving moment of the rotating arm The optimization objective function of the NMPC controller is: ； Subject to: ； In the formula, represents the system dynamics constraint, which is obtained according to a system dynamics model dynamics, and the system state variable change needs to conform to the system dynamics model dynamics, that is, a dynamics model of the unmanned aerial vehicle-mechanical arm integrated system, is a state constraint, is an input constraint, and the initial state is set as the current state of the system , are weight matrices of the state, input and terminal state respectively, the NMPC controller generates an optimal control sequence by solving the optimization objective function in real time, t is time, and T is the time when the prediction time domain ends.

2. The method of claim 1, wherein: The control relationship of the adaptive admittance controller is: ； wherein: is a force tracking error, is an actual contact force measured by the end-effector force sensor, is a desired force; , , is a desired admittance parameter, is a position error derivative, derivative; and derivative. The adaptive admittance controller adjusts the admittance parameters based on an adaptive law : ； wherein T is a matrix of symbols, is the derivative of the adaptive law adjustment admittance parameter .

3. The method of claim 1, wherein: The dynamic model of the unmanned aerial vehicle-robot arm integrated system is as follows: ； where, is the system's inertia matrix term, is the system's Coriolis and centrifugal force matrix, is the gravity vector, is the measured environmental force / torque at the end of the manipulator, i.e., the environment's effect on the system through its action on the manipulator's end; is the Jacobian matrix mapping the end force to the generalized force space, is the generalized control input vector, q is the position vector, composed of , is the system's generalized velocity, is the derivative of .

4. The method of claim 1, wherein: The discrete update formula of the reference trajectory in step 2 is as follows: ； In the formula, k represents the time or number of iterations. These are the reference trajectory and the expected trajectory, respectively. , , For the desired admittance parameter, Position error The derivative, for The derivative, The actual contact force measured by the end force sensor. For the power of expectation.

5. An aerial work robot based on the method of any of claims 1-4, characterized by: The aerial work robot is a full-drive type aerial work robot, and at least comprises a body, a power unit, a rotating mechanism, a robot arm, a terminal sliding work mechanism and a control system. The power unit is composed of six tilting rotors, the central part of the rotating mechanism is fixed to the center of the body, and the outer part is rotatable; the robot arm is connected to the outer part and rotates around the center of the body, the robot arm is designed as single degree of freedom, and is driven by a servo motor, the proximal end is a rigid arm, and the tail end is connected to the terminal sliding work mechanism for contacting the environment; The control system generates the optimal control instruction generated by steps 1-3 to control the aerial work robot.

6. The aerial work robot according to claim 5, characterized in that: The control system comprises an on-board computer and a flight controller installed on the base of the body; The on-board computer controls the servo motor driving the rotating mechanism through a serial port, and the flight controller generates a rotor speed instruction and sends it to an electronic speed regulator to drive the rotor motor; The on-board computer is provided with a processor and a memory; the processor calls the computer program stored in the memory to realize the optimal control instruction obtained by steps 1-3, and transmits the optimal control instruction to the flight controller.

7. A control system based on the method according to any one of claims 1 to 4, characterized in that: There are: A dynamic model construction module for regarding the full-drive type aerial work robot as an integrated aerial executor, and constructing an unmanned aerial vehicle-robot arm integrated system model; an adaptive admittance control module for introducing an adaptive admittance controller as an outer loop force controller, based on which a reference trajectory is updated according to a deviation of an actual contact force from a desired force dynamically adjusting admittance parameters and thereby correcting the pose​ An inner loop motion control module for inputting the reference trajectory of the aerial work robot and the state variable into the inner loop motion controller to obtain the optimal control instruction of the aerial work robot.

8. A computer-readable storage medium, characterized in that: The computer program is stored, and the computer program is called by the processor to realize the steps of the adaptive admittance control aerial work robot control method according to any one of claims 1-4.

Images