Flying robot control method and device based on force feedback and medium

Abstract

The invention provides a flying robot control method and device based on force feedback and a medium, and the method comprises the steps: deploying a six-dimensional force torque sensor at an operation end effector of a flying robot, and obtaining and screening effective torque data in real time to obtain a force error signal; a compensation layer based on the Lyapunov self-adaptive law is utilized, the total disturbance is cooperatively estimated in combination with double observers, and the disturbance serves as a feed-forward quantity to be injected into an attitude controller; and calculating the adjustment amount according to the target impedance model and the force error, and generating an execution mechanism control instruction. According to the technical scheme provided by the invention, stable control of air interaction operation can be realized, and the anti-disturbance capability and interaction force control precision of the flying robot are effectively improved.

Description

Technical Field

[0001] This document relates to the field of flight robot control technology, and in particular to a flight robot control method, device and medium based on force feedback. Background Technology

[0002] With the development of artificial intelligence and robotics, flying robots, thanks to their mobility and flexibility, are widely used in high-altitude operations, covering areas such as power facility maintenance, wind power equipment maintenance, high-altitude structure installation and inspection, and disaster monitoring and emergency response. These applications share common technical challenges: complex operating environments, difficult-to-access operating locations, and extremely high requirements for operational stability and precision. Flying robots must maintain their stable flight attitude while performing precise physical interactions with the environment or manipulated objects, such as contact, grasping, and pushing, and sense and adapt to interactive forces and reactions, posing a severe challenge to stable control technology.

[0003] Current flight robot control largely relies on high-precision positioning and navigation technology to maintain position and track, combined with path planning to avoid static obstacles. However, it has significant shortcomings in aerial interactive operations: First, it cannot effectively handle dynamic interactive forces and has poor robustness to six-dimensional force / torque disturbances generated by the contact between the robotic arm and the manipulated object; second, it is sensitive to external environmental disturbances, such as high wind speeds and turbulence, which can easily disrupt attitude balance and are difficult to compensate for quickly; third, it lacks an effective force feedback mechanism, and force sensor data is not integrated into the attitude control loop in a closed loop; fourth, the integration of control strategies is low, and the combination of advanced force control strategies and attitude adaptive compensation algorithms is not in-depth; fifth, the high-dimensional mechanical information from the six-dimensional force / torque sensors is not fully utilized, making it difficult to build an efficient mechanical state adjustment mechanism.

[0004] To address the aforementioned issues, relevant patents have proposed some solutions, but these solutions have significant limitations. For example, while patent CN111984024B achieves robust position and attitude tracking through geometric control, it does not establish a real-time force feedback mechanism, does not utilize a six-dimensional force / torque sensor to perceive interactive forces, and does not integrate force control strategies. Similarly, while patent CN112527008B simplifies the control structure through decoupling design, it does not involve force feedback and mechanical state adjustment, nor does it incorporate advanced force control strategies.

[0005] In summary, existing technologies have failed to solve the core problem of how to integrate multi-dimensional force perception, attitude adaptive compensation and advanced force control strategies through an efficient real-time force feedback mechanism to achieve real-time and precise adjustment of the mechanical state of the flying robot in aerial interactive operations. It is difficult to simultaneously ensure the stability of flight attitude and the accuracy and safety of interactive operations in complex environments. Summary of the Invention

[0006] This invention provides a force feedback-based control method, device, and medium for flying robots, aiming to solve the above-mentioned problems.

[0007] According to an embodiment of the present invention, a force feedback-based control method for a flight robot is provided, comprising: S1. The torque data of the flying robot is obtained by a six-dimensional force and torque sensor deployed on the end effector of the flying robot. The torque data is filtered and screened to obtain effective torque data. The effective torque data is compared with the preset torque data to obtain the force error signal. S2. Through a compensation layer based on Lyapunov adaptive law, the total disturbance of the flying robot is estimated in real time and injected into the attitude controller as a feedforward quantity. Based on the target impedance model and the measured force error, the adjustment amount data required to achieve the desired force interaction is calculated. The adjustment amount data is input to the attitude controller enhanced by feedforward compensation to generate control commands for driving the actuator of the flying robot. The total disturbance includes effective torque data and external environmental disturbance data.

[0008] According to an embodiment of the present invention, an electronic device is provided, comprising: Processor; and, A memory is configured to store computer-executable instructions, which, when executed, cause the processor to perform the steps of the force feedback-based flight robot control method described above.

[0009] According to an embodiment of the present invention, a storage medium is provided for storing computer-executable instructions, which, when executed, implement the steps of the force feedback-based flight robot control method described above.

[0010] This invention employs a six-dimensional force / torque sensor deployed on the end effector to accurately capture dynamic changes in interactive force / torque, addressing the issues of insufficient force feedback or low data utilization in existing technologies. Based on a Lyapunov adaptive law compensation layer, combined with dual observers for collaborative estimation of total disturbance, and optimized through an adaptive gain dynamic observer, the disturbance estimation accuracy is fed forward to the attitude controller in real time. This overcomes the shortcomings of insufficient disturbance immunity in existing technologies, ensuring hovering accuracy and attitude stability in high wind speeds and turbulent conditions, and preventing operational deviations or collisions. Furthermore, by combining the adaptive disturbance compensation algorithm with a target impedance model, the impedance model directly inputs the attitude adjustment calculated based on the force error into the compensated attitude control loop, overcoming the limitation of separating these two technologies in existing technologies. In high-altitude structural installation and other operations, this approach balances flight attitude stability and compliant interactive force control, avoiding impacts on the manipulated object. Based on the LSTM-Kalman filter model, environmental disturbances are predicted, and force stability constraints are combined to generate disturbance-resistant optimized trajectories. Force setpoints and impedance parameters are adjusted according to the disturbance intensity levels to solve the problems of path planning deviating from force constraints and parameter rigidity in existing technologies. Key data is recorded throughout the process, and indicators such as force tracking error and attitude stability are calculated through the performance evaluation module to achieve online fine-tuning and offline optimization. This makes up for the lack of monitoring and feedback in existing technologies, avoids anomalies in a timely manner, and supports continuous improvement of system performance. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a force feedback-based flight robot control method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a real-time mechanical feedback sensing and adjustment mechanism based on a six-dimensional force sensor, according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the integrated and improved flight attitude adaptive compensation algorithm and impedance control according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a data-driven control strategy based on real-time mechanical state according to an embodiment of the present invention; Figure 5 This is a schematic diagram of an online navigation path optimization strategy to resist disturbances, according to an embodiment of the present invention. Figure 6 Diagram of closed-loop monitoring and performance feedback. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0014] Method Implementation Examples According to an embodiment of the present invention, a flight robot control method based on force feedback is provided. Figure 1 This is a flowchart of a force feedback-based flight robot control method according to an embodiment of the present invention. Figure 1 As shown, the force feedback-based flight robot control method of this invention specifically includes: Step S1: Obtain torque data of the flying robot through the six-dimensional force and torque sensor deployed on the end effector of the flying robot, and obtain effective torque data after filtering and screening the torque data. Compare the effective torque data with the preset torque data to obtain the force error signal. The specific implementation process of step S1 is as follows: Figure 2 ,according to Figure 2 As can be seen, the embodiments of the present invention acquire force error signals by constructing a real-time mechanical feedback sensing and adjustment mechanism based on a six-dimensional force sensor, specifically including: Sensor deployment and data acquisition: A high-precision six-dimensional force / torque sensor is deployed at the connection point between the end effector and the robot body, or directly integrated into the end effector. The sensor's deployment position is optimized through dynamic simulation to ensure accurate perception of the mechanical information of all six degrees of freedom generated by the interaction between the flying robot and the manipulated object, while avoiding interference from inertial forces generated by the robot arm's movement. The end effector can be a robotic arm gripper or a tool interface. Real-time acquisition of the mechanical information generated by the interaction between the flying robot and the manipulated object is achieved. Triaxial Orthogonal Force and torque around the three axes ; Signal preprocessing and coordinate transformation: The acquired triaxial orthogonal force data and torque data around the three axes undergo multi-level signal processing, including: Adaptive low-pass filtering: The cutoff frequency is dynamically adjusted according to the current motion state of the flying robot. When moving at high speed, the cutoff frequency is appropriately increased to retain the effective signal, and when hovering, the cutoff frequency is decreased to enhance the filtering effect. Wavelet noise reduction: Wavelet transform is used to perform multi-resolution analysis on the signal, effectively separating noise components from useful signals; Temperature drift compensation: Real-time temperature compensation is performed on the force sensor output based on the readings of the built-in temperature sensor.

[0015] By using a coordinate transformation algorithm based on quaternions, the processed triaxial orthogonal force data and torque data around the three axes are transformed from the sensor coordinate system to the body coordinate system or world coordinate system of the flying robot, establishing a unified mechanical information expression benchmark and providing a foundation for multi-source information fusion.

[0016] Effective contact state identification is performed based on a preset contact force threshold to obtain effective torque data, specifically including: When the force value in at least one direction of the converted triaxial orthogonal force data and the torque data around the three axes exceeds a preset threshold and the duration is ≥2 sampling cycles of the six-dimensional force and torque sensor, it is determined as effective contact and the force feedback control loop is activated. The corresponding triaxial orthogonal force data and the torque data around the three axes at this time are taken as the effective torque data. This design avoids interference from non-contact noise on the force feedback control loop.

[0017] S2. Through a compensation layer based on Lyapunov adaptive law, the total disturbance of the flying robot is estimated in real time and injected into the attitude controller as a feedforward quantity. Based on the target impedance model and the measured force error, the adjustment amount data required to achieve the desired force interaction is calculated. The adjustment amount data is input to the attitude controller enhanced by feedforward compensation to generate control commands for driving the actuator of the flying robot. The total disturbance includes effective torque data and external environmental disturbance data.

[0018] according to Figure 3 It can be seen that the real-time estimation of the total disturbance of the flying robot through the compensation layer based on the Lyapunov adaptive law specifically includes: A dual-observer collaborative estimation architecture is constructed, which includes a nonlinear perturbation observer based on a dynamic model and a parameter adaptor based on Lyapunov adaptive law. The nonlinear disturbance observer is used to quickly estimate lumped disturbances caused by the external environment and unmodeled dynamics. The parameter adaptive device identifies and compensates online for slow-varying system disturbances caused by robotic arm motion, load changes, and model parameter uncertainties. The lumped disturbance estimated by the nonlinear disturbance observer is fused with the systematic disturbance identified by the parameter adaptor to obtain the equivalent total disturbance of the flying robot. This dual-observer cooperative estimation architecture achieves independent estimation and fusion compensation of interactive force disturbances and environmental disturbances.

[0019] The external environmental disturbance data is obtained through online estimation by an adaptive gain dynamic disturbance observer. Specifically, this includes online estimation of unmodeled aerodynamic disturbances caused by high wind speed, turbulence, humidity changes, airflow vortices, and sudden gusts, as well as their impact on attitude angles. The gain coefficient of the observer is dynamically adjusted according to the absolute value of the disturbance estimation error. When the error increases, the gain is increased to speed up the response, and when the error decreases, the gain is decreased to suppress noise.

[0020] For the desired interactive operation behavior, such as compliant contact and force tracking, a target impedance model is defined, which is expressed as follows: ; Where M, D, and K are the set virtual inertia, virtual damping, and virtual stiffness matrices, respectively. It's a positional deviation. This is the force error signal. Based on the target impedance model and real-time measurements. The position / attitude adjustment required to achieve the desired impedance characteristics is calculated and denoted as . ; The calculated The input is fed into an attitude control loop that incorporates adaptive compensation. This controller drives the flight robot's actuators to dynamically adjust their flight attitude and / or the position of the end effector, ensuring the actual interaction force / torque... Approaching the expected value This ensures overall flight stability while maintaining stability. The key innovation of this combination lies in the fact that the adjustment amount of the impedance control output directly acts on the attitude control loop, which already has strong disturbance rejection capabilities, thus achieving tight coupling between force control and attitude stability.

[0021] Figure 4 This is a schematic diagram of a data-driven control strategy according to an embodiment of the present invention. In a specific implementation of the present invention, the force feedback-based flight robot control method further includes a data-driven control strategy: The data-driven control strategy specifically includes: Step S3: Construct a mechanical state feature vector for the operation. The feature vector includes triaxial orthogonal force data, torque data around the three axes, flight attitude angles, angular velocities, and position information. The mechanical state feature vector is processed based on an LSTM-Kalman filter fusion prediction model to output prediction information for environmental disturbances. The LSTM-Kalman filter fusion prediction model learns the temporal correlation characteristics of the mechanical state feature vector through an LSTM network and uses Kalman filtering to correct errors in the prediction results. Figure 5As an embodiment of the present invention, the disturbance-resistant online navigation path optimization strategy, in a specific implementation of the present invention, further includes a disturbance-resistant online navigation path optimization strategy: The disturbance-resistant online navigation path optimization strategy specifically includes: Step S4: Input the predicted environmental disturbance information into the path planning module. Based on the preset global path, use a model predictive control framework or local trajectory optimization algorithm to perform online replanning of the local path, generating an optimized trajectory resistant to disturbances online. Input the optimized trajectory into the position or velocity control loop of the flight control system for execution. The objective function of the planning simultaneously considers: flight stability indicators, operational stability constraints, environmental disturbance avoidance / utilization, and obstacle avoidance safety distance. Among them, flight stability indicators include: minimizing the rate of change of attitude angles and minimizing the positional deviation from the desired trajectory; operational stability constraints include ensuring that path adjustments do not cause the end effector to disengage or generate excessive impact force; environmental disturbance avoidance / utilization involves actively avoiding predicted strong turbulence areas or adjusting the path in the downwind direction to save energy.

[0022] After outputting the predicted information on environmental disturbances, it further includes: A graded adjustment strategy is adopted based on the disturbance intensity level of the predicted information to dynamically adjust the expected force setting value, target impedance parameter, and gain parameter of the adaptive compensation algorithm. The disturbance intensity level is determined based on the magnitude of the impact of environmental disturbances on the interaction force deviation between the flying robot and the manipulated object. This division is based on the stability requirements of the flying robot's interactive operation. A low disturbance level is defined as a level where the proportion of the interaction force deviation caused by the disturbance is within a first preset interval; a medium disturbance level is defined as a level where the deviation proportion is within a second preset interval; and a high disturbance level is defined as a level where the deviation proportion is within a third preset interval. Furthermore, the upper limit of the first preset interval is less than the lower limit of the second preset interval, and the upper limit of the second preset interval is less than the lower limit of the third preset interval. The first preset interval is the interval where the deviation proportion is lower than the minimum disturbance impact threshold allowed for operational stability; the second preset interval is the interval where the deviation proportion is between the minimum disturbance impact threshold and the maximum acceptable disturbance impact threshold; and the third preset interval is the interval where the deviation proportion is higher than the maximum acceptable disturbance impact threshold. At low disturbance levels, the adjustment range of each parameter is based on not causing oscillations in the control parameters and maintaining the stability of the interaction between the flying robot and the manipulated object; When the disturbance level is medium, the adjustment range of each parameter is based on balancing the disturbance response speed and control stability, and the adjustment range is greater than the adjustment range corresponding to the low disturbance level. At high disturbance levels, the adjustment range of each parameter prioritizes rapid disturbance suppression and avoids the expansion of interaction force deviations, and this adjustment range is greater than that corresponding to medium disturbance levels. This data-driven control strategy, based on real-time mechanical states, significantly improves the system's adaptability and intelligence in complex dynamic environments.

[0023] Figure 6 This invention provides a closed-loop monitoring and performance feedback mechanism based on [specific embodiment]. Figure 6 As shown in a specific implementation of the present invention, the force feedback-based flight robot control method further includes a closed-loop monitoring and performance feedback mechanism: The closed-loop monitoring and performance feedback mechanism specifically includes: during system operation, continuously recording all key state variables and control variables such as raw and processed data from the six-dimensional force / torque sensor, attitude angle, position, control commands, adaptive compensation amount, impedance parameters, and path planning results; the state variables include the angular acceleration, force change rate, and ambient air pressure value of the flying robot; The performance evaluation module calculates the power tracking error and attitude stability index. The performance evaluation module includes a data preprocessing unit, an error calculation unit, and an index analysis unit. The data preprocessing unit performs noise reduction and standardization on the recorded data. The error calculation unit calculates the power tracking error based on the processed data. The index analysis unit calculates the attitude stability index in combination with the flight attitude fluctuation range. Adjusting control parameters or optimizing the model based on the evaluation results includes: fine-tuning control parameters online when the force tracking error is within a preset small error range and the attitude stability index fluctuation is within a preset small fluctuation range; and optimizing the model offline when the force tracking error exceeds the preset small error range or the attitude stability index fluctuation exceeds the preset small fluctuation range.

[0024] The embodiments of the present invention have the following beneficial effects: By deploying a six-dimensional force and torque sensor on the end effector, dynamic changes in interactive force / torque are accurately captured, addressing the issues of insufficient force feedback or low data utilization in existing technologies. Based on a Lyapunov adaptive law compensation layer, combined with dual observers, the total disturbance is estimated collaboratively. An adaptive gain dynamic observer optimizes the disturbance estimation accuracy, providing real-time feedforward compensation to the attitude controller. This overcomes the shortcomings of insufficient disturbance immunity in existing technologies, ensuring hovering accuracy and attitude stability in high wind speeds and turbulent environments, preventing operational deviations or collisions. The adaptive disturbance compensation algorithm is combined with a target impedance model. The impedance model, based on the attitude adjustment calculated from the force error, is directly input into the compensated attitude control loop, overcoming the limitation of separating these two technologies in existing ones. In high-altitude structural installation and other operations, this approach balances flight attitude stability and compliant interactive force control, avoiding impacts on the workpiece. Based on the LSTM-Kalman filter model, environmental disturbances are predicted, and force stability constraints are combined to generate disturbance-resistant optimized trajectories. Force setpoints and impedance parameters are adjusted according to the disturbance intensity levels to solve the problems of path planning deviating from force constraints and parameter rigidity in existing technologies. Key data is recorded throughout the process, and indicators such as force tracking error and attitude stability are calculated through the performance evaluation module to achieve online fine-tuning and offline optimization. This makes up for the lack of monitoring and feedback in existing technologies, avoids anomalies in a timely manner, and supports continuous improvement of system performance.

[0025] Device Example 1 An electronic device is provided according to an embodiment of the present invention, comprising: Processor; and, A memory is configured to store computer-executable instructions, which, when executed, cause the processor to perform the steps of the force feedback-based flight robot control method described above.

[0026] Device Example 2 According to an embodiment of the present invention, a storage medium is provided for storing computer-executable instructions, which, when executed, implement the steps of the above-described force feedback-based flight robot control method.

[0027] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A force feedback based flying robot control method, characterized by The method comprises the following steps: S1, acquiring torque data of the flying robot through a six-dimensional force torque sensor deployed on a flying robot operation end effector, filtering and screening the torque data to obtain effective torque data, comparing the effective torque data with preset torque data, and acquiring force error signals; S2, estimating the total disturbance of the flying robot in real time through a compensation layer based on Lyapunov adaptive law, and injecting the total disturbance as a feedforward quantity into an attitude controller, calculating the adjustment data required to achieve the expected force interaction according to the target impedance model and the measured force error, inputting the adjustment data into the attitude controller enhanced by feedforward compensation to generate control instructions for driving the flying robot actuator, wherein the total disturbance includes effective torque data and external environmental disturbance data.

2. The method of claim 1, wherein, The acquisition of the torque data of the flying robot and the filtering and screening of the torque data to obtain the effective torque data specifically comprises: On the working end effector of the flying robot, a six-dimensional force torque sensor is deployed to obtain the interaction force between the flying robot and the operating object in real time Three-axis orthogonal force And the torque around the three axes ; Performing low-pass filtering and noise reduction processing on the collected three-axis orthogonal force data and the torque data around the three axes; Converting the processed three-axis orthogonal force data and the torque data around the three axes from the sensor coordinate system to the body coordinate system or the world coordinate system of the flying robot through a coordinate transformation algorithm; Based on the preset contact force threshold, the effective contact state is identified to obtain the effective torque data, specifically including: When the force value in at least one direction of the converted three-axis orthogonal force data and the torque data around the three axes exceeds the preset threshold and the duration is greater than or equal to 2 sampling periods of the six-dimensional force torque sensor, it is determined that there is effective contact and the force feedback control loop is activated, and the corresponding three-axis orthogonal force data and the torque data around the three axes at this time are taken as the effective torque data.

3. The method of claim 1, wherein, The external environmental disturbance data is obtained by online estimation of an adaptive gain dynamic disturbance observer, specifically including: estimating the unmodeled aerodynamic disturbance caused by high wind speed, turbulence, humidity change, airflow vortex and sudden gust and its influence on the attitude angle in real time through the adaptive gain dynamic disturbance observer; wherein the gain coefficient of the observer is dynamically adjusted according to the absolute value of the disturbance estimation error, and the gain is increased to speed up the response when the error increases, and the gain is reduced to suppress noise when the error decreases.

4. The method of claim 1, wherein, The target impedance model is expressed as: ； wherein M, D, K are respectively the set virtual inertia, virtual damping and virtual stiffness matrices, is the position error, is the force error signal.

5. The method of claim 1, wherein, The method further comprises: Constructing a work mechanics state feature vector, which includes three-axis orthogonal force data, torque data around the three axes, flying attitude angle, angular velocity and position information; Processing the mechanics state feature vector based on an LSTM-Kalman filter fusion prediction model to output prediction information of the environmental disturbance; the LSTM-Kalman filter fusion prediction model learns the time sequence correlation characteristics of the mechanics state feature vector through an LSTM network, and combines Kalman filtering to correct the prediction result; Feeding the prediction information of the environmental disturbance to a local path planner, and combining work force stability constraints and flight stability indicators to generate an anti-disturbance optimized trajectory online; Converting the optimized trajectory into control instructions and sending them to the control system of the flying robot for execution.

6. The method of claim 5, wherein, After outputting the prediction information of the environmental disturbance, further comprising: The hierarchical adjustment strategy is adopted according to the disturbance intensity level of the prediction information, and the expected force setting value, the target impedance parameter and the gain parameter of the adaptive compensation algorithm are dynamically adjusted; wherein the disturbance intensity level is divided based on the influence amplitude of the environmental disturbance on the interaction force deviation of the flying robot and the operating object, and the division is based on the stability requirement of the flying robot interaction work, the low disturbance level is the level of the interaction force deviation caused by the disturbance, the proportion of which is in the first preset interval of the preset interaction force threshold, the medium disturbance level is the level of the deviation proportion in the second preset interval, and the high disturbance level is the level of the deviation proportion in the third preset interval, and the first preset interval upper limit < second preset interval lower limit, second preset interval upper limit < third preset interval lower limit; the first preset interval is the interval in which the deviation proportion is lower than the minimum disturbance influence threshold allowed by the work stability, the second preset interval is the interval in which the deviation proportion is between the minimum disturbance influence threshold and the maximum acceptable disturbance influence threshold, and the third preset interval is the interval in which the deviation proportion is higher than the maximum acceptable disturbance influence threshold; When the disturbance level is low, the adjustment amplitude of each parameter is based on the premise that the control parameter is not oscillated and the stability of the flying robot and the operating object interaction is maintained; When the disturbance level is medium, the adjustment amplitude of each parameter is based on the balance between the disturbance response speed and the control stability, and the adjustment amplitude is greater than the adjustment amplitude corresponding to the low disturbance level; When the disturbance level is high, the adjustment amplitude of each parameter is based on the premise of giving priority to guaranteeing the rapid suppression of disturbance and avoiding the expansion of interaction force deviation, and the adjustment amplitude is greater than the adjustment amplitude corresponding to the medium disturbance level.

7. The method of claim 1, wherein, The method further comprises: Real-time recording of sensor data, control instructions and state variables; the state variables include the angular acceleration of the flying robot, the force change rate and the environmental air pressure value; Calculating the force tracking error and the attitude stability index based on the performance evaluation module; the performance evaluation module includes a data preprocessing unit, an error calculation unit and an index analysis unit, the data preprocessing unit performs noise reduction and standardization processing on the recorded data, the error calculation unit calculates the force tracking error based on the processed data, and the index analysis unit calculates the attitude stability index combined with the flying attitude fluctuation range; Adjusting the control parameters or optimizing the model according to the evaluation results, specifically including: when the force tracking error is in the preset small error interval and the attitude stability index fluctuation is in the preset small fluctuation interval, the control parameters are adjusted online; when the force tracking error exceeds the preset small error interval or the attitude stability index fluctuation exceeds the preset small fluctuation interval, the model is optimized offline.

8. The method of claim 1, wherein, The compensation layer based on Lyapunov adaptive law is used to estimate the total disturbance of the flying robot in real time, specifically including: A dual observer cooperative estimation architecture is constructed, which includes a nonlinear disturbance observer based on a dynamic model and a parameter adaptive device based on Lyapunov adaptive law; The nonlinear disturbance observer is used to quickly estimate the lumped disturbance caused by external environment and unmodeled dynamics; The parameter adaptive device is used to identify and compensate the slow-varying systematic disturbance caused by the movement of the robot arm, the change of the load and the uncertainty of the model parameters. The lumped disturbance estimated by the nonlinear disturbance observer is fused with the systematic disturbance identified by the parameter adaptive to obtain an equivalent total disturbance of the flying robot. 9.An electronic device comprising: a processor; and a memory arranged to store computer-executable instructions that, when executed, cause the processor to perform the steps of the force feedback based control method of a flying robot according to any one of claims 1-8. 10.A storage medium for storing computer-executable instructions that, when executed, implement the steps of the force feedback based control method of a flying robot according to any one of claims 1-8.

Images