Unmanned aerial vehicle high-precision spacer dismounting and mounting control method based on tactile feedback

Abstract

The invention discloses an unmanned aerial vehicle high-precision spacer dismounting and mounting control method based on tactile feedback, and belongs to the technical field of intelligent operation and maintenance of high-voltage power transmission and distribution lines. According to the method, a multi-mode tactile sensor array and a self-adaptive force control mechanical arm are carried through an unmanned aerial vehicle, three-dimensional force, torque and surface deformation data when the mechanical arm makes contact with a spacer are collected in real time, the action of the mechanical arm is adjusted in combination with a dynamic impedance control algorithm, and high-precision disassembly and assembly operation is achieved. According to the method, the problems of large positioning error and poor operation stability caused by lack of real-time force feedback of a traditional unmanned aerial vehicle disassembling and assembling method are solved, and the operation precision and safety in a complex environment are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent operation and maintenance of high-voltage transmission and distribution lines, and in particular to a high-precision spacer disassembly and assembly control method for an unmanned aerial vehicle (UAV) based on tactile feedback, which is suitable for automated maintenance operations of high-voltage cable spacers in complex environments. Background Art

[0002] Spacers are key components for maintaining the spacing between split conductors in high-voltage transmission lines. Their assembly and disassembly must be performed at high altitudes, and traditional manual labor is associated with low efficiency and high risk. While existing drone-based assembly and disassembly methods reduce manual intervention, they still suffer from the following drawbacks:

[0003] Reliance on visual positioning: In rain, fog, strong light, or wire obstruction, the visual system is prone to misidentification, resulting in positioning deviation;

[0004] Lack of force feedback: The robot arm cannot sense contact force during operation, which can easily cause damage to the spacer rods or wires;

[0005] Poor adaptability: When faced with abnormal conditions such as corrosion and deformation, fixed control strategies are difficult to guarantee the success rate of operations.

[0006] Therefore, there is an urgent need in this field for a technical solution that can improve the accuracy and reliability of disassembling and assembling spacer rods of drones.

[0007] The information disclosed in this background technology section is only intended to enhance understanding of the overall background of the invention and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the Invention

[0008] The purpose of the present invention is to provide a technical solution that can improve the accuracy and reliability of disassembling and assembling spacer rods of a drone.

[0009] To achieve the above object, the present invention provides the following solutions:

[0010] A high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback comprises the following steps:

[0011] S1: The drone uses a fusion positioning module based on a lidar and inertial navigation system to obtain 3D point cloud data of the high-voltage cable. Combined with the preset spacer installation coordinates, it plans the flight path and reaches the target location.

[0012] S2: The drone activates a tactile feedback module, which includes a multimodal tactile sensor array and an adaptive force-controlled robotic arm. The tactile sensor array is integrated on the surface of the robotic arm's end effector and is used to collect real-time three-dimensional force, torque, and surface deformation data when the robotic arm contacts the spacer rod, and transmits the data to the onboard processor via a high-speed communication module.

[0013] S3: Based on the tactile data, the onboard processor uses a pre-trained deep learning model to identify the type, installation status, and corrosion level of the spacer. This deep learning model extracts the spatial features of the tactile signals through a convolutional neural network and analyzes the time series data using a long-short-term memory network to output a spacer status classification result.

[0014] S4: Based on the recognition results, the dynamic impedance control algorithm calculates the desired stiffness and damping parameters of the manipulator in real time and generates force-position hybrid control instructions. The algorithm maps the tactile data into impedance parameter adjustments by establishing a dynamic model of the contact between the manipulator end and the environment, thereby achieving smooth control of the manipulator's movements.

[0015] S5: During the disassembly and assembly process, the drone simultaneously activates the visual verification module, captures images of the operating area through the binocular camera, aligns them with the tactile data in time and space, detects positioning deviations, and generates compensation instructions;

[0016] S6: When the robot arm performs a disassembly or installation operation, the tactile sensor monitors the contact force changes in real time. If the force exceeds the preset safety threshold, the emergency stop mechanism is immediately triggered and an alarm is sent back to the remote control terminal.

[0017] S7: After completing a single operation, the drone generates an operation log through the edge computing module, recording tactile data, visual images, control instructions and timestamps, and encrypts and stores them in a cloud database using blockchain technology to ensure that the data cannot be tampered with;

[0018] S8: The drone automatically switches to the next target location according to the preset task queue and repeats steps S1-S7 until all spacer disassembly and assembly tasks are completed;

[0019] S9: After the mission is completed, the remote control terminal performs a full-process retrospective analysis based on the operation log to optimize the deep learning model parameters and control algorithm strategy;

[0020] S10: When the self-repair mechanism detects an abnormality in the robotic arm or sensor, it repairs key components on-site through redundant module switching or 3D printing technology to ensure mission continuity.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The present invention provides a high-precision spacer assembly and disassembly control method for unmanned aerial vehicles (UAVs) based on tactile feedback. High-precision tactile sensors are used to capture mechanical signals in real time, improving operational robustness in complex environments. The present invention uses tactile feedback to adaptively adjust the stiffness and damping of the robotic arm to achieve smooth operation. Multi-sensor fusion verification ensures operational accuracy by synergizing tactile data with visual and positioning information. The present invention ensures mission continuity through redundant modules and 3D printing technology, supporting multi-machine collaborative optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention 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 invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 A schematic flow chart of a control method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0026] The purpose of the present invention is to provide a technical solution that can improve the accuracy and reliability of disassembling and assembling spacer rods of a drone.

[0027] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] Example 1:

[0029] This embodiment provides a high-precision control method for disassembling and assembling spacer rods of a UAV based on tactile feedback. Figure 1 Shown, including:

[0030] S1: The drone uses a fusion positioning module based on a lidar and inertial navigation system to obtain 3D point cloud data of the high-voltage cable. Combined with the preset spacer installation coordinates, it plans the flight path and reaches the target location.

[0031] S2: The drone activates a tactile feedback module, which includes a multimodal tactile sensor array and an adaptive force-controlled robotic arm. The tactile sensor array is integrated on the surface of the robotic arm's end effector and is used to collect real-time three-dimensional force, torque, and surface deformation data when the robotic arm contacts the spacer rod, and transmits the data to the onboard processor via a high-speed communication module.

[0032] S3: Based on the tactile data, the onboard processor uses a pre-trained deep learning model to identify the type, installation status, and corrosion level of the spacer. This deep learning model extracts the spatial features of the tactile signals through a convolutional neural network and analyzes the time series data using a long-short-term memory network to output a spacer status classification result.

[0033] S4: Based on the recognition results, the dynamic impedance control algorithm calculates the desired stiffness and damping parameters of the manipulator in real time and generates force-position hybrid control instructions. The algorithm maps the tactile data into impedance parameter adjustments by establishing a dynamic model of the contact between the manipulator end and the environment, thereby achieving smooth control of the manipulator's movements.

[0034] S5: During the disassembly and assembly process, the drone simultaneously activates the visual verification module, captures images of the operating area through the binocular camera, aligns them with the tactile data in time and space, detects positioning deviations, and generates compensation instructions;

[0035] S6: When the robot arm performs a disassembly or installation operation, the tactile sensor monitors the contact force changes in real time. If the force exceeds the preset safety threshold, the emergency stop mechanism is immediately triggered and an alarm is sent back to the remote control terminal.

[0036] S7: After completing a single operation, the drone generates an operation log through the edge computing module, recording tactile data, visual images, control instructions and timestamps, and encrypts and stores them in a cloud database using blockchain technology to ensure that the data cannot be tampered with;

[0037] S8: The drone automatically switches to the next target location according to the preset task queue and repeats steps S1-S7 until all spacer disassembly and assembly tasks are completed;

[0038] S9: After the mission is completed, the remote control terminal performs a full-process retrospective analysis based on the operation log to optimize the deep learning model parameters and control algorithm strategy;

[0039] S10: When the self-repair mechanism detects an abnormality in the robotic arm or sensor, it repairs key components on-site through redundant module switching or 3D printing technology to ensure mission continuity.

[0040] In one embodiment, the fusion positioning module is specifically implemented as follows:

[0041] The lidar scans high-voltage cables and the surrounding environment to generate a three-dimensional point cloud map with millimeter-level accuracy; the inertial navigation system calculates the real-time attitude and position of the drone through gyroscope and accelerometer data; the lidar and INS data are fused through the extended Kalman filter algorithm to eliminate the cumulative error of a single sensor and achieve highly robust positioning in complex electromagnetic environments.

[0042] In one embodiment, the multimodal tactile sensor array comprises:

[0043] Piezoelectric force sensor, capacitive deformation sensor and magnetostrictive torque sensor; the multimodal tactile sensor array is encapsulated with a flexible substrate material to adapt to the irregular shape of the high-voltage cable surface; the sensor data sampling frequency is higher than 1kHz to ensure accurate capture of the dynamic contact process.

[0044] In one embodiment, the deep learning model is trained by:

[0045] A dataset of tactile signals from spacers under different working conditions is collected, including normal installation, looseness, corrosion, and deformation. The data is normalized, time-domain filtered, and feature enhanced. Transfer learning technology is used to extract tactile image features using a pre-trained ResNet-50 model, combined with a bidirectional LSTM network to capture temporal dependencies, and finally the classification probability is output through a fully connected layer.

[0046] In one embodiment, the dynamic impedance control algorithm is formulated as follows:

[0047] M x +B x +K(xx d )=F ext

[0048] Among them, M, B, and K are the inertia matrix, damping matrix, and stiffness matrix respectively; x is the actual position of the end of the robot arm, and x is the d is the desired position, F ext The external contact force fed back by the tactile sensor; the algorithm is based on F ext Adjust B and K according to the real-time changes to achieve a dynamic balance between force and position.

[0049] In one embodiment, the binocular camera captures RGB-D images at a rate of 30fps and aligns them with the timestamp of the tactile data through a feature matching algorithm. Stereo vision technology is used to reconstruct the three-dimensional structure of the operating area, which is compared with the lidar point cloud map. The positioning deviation is calculated and PID compensation instructions are generated to correct the drone's hovering position and the posture of the end of the robotic arm.

[0050] In one embodiment, the specific application of blockchain technology in S7 is:

[0051] The operation log is divided into multiple data blocks, each of which contains a hash value, timestamp, and hash pointer to the previous block. Smart contracts are used to verify data integrity, and only authorized terminals can decrypt and access log contents, preventing operation data from being maliciously tampered with or leaked.

[0052] In one embodiment, the self-repair mechanism comprises:

[0053] Redundantly driven robotic arm joint module, spare tactile sensor unit and micro 3D printing device; when a joint motor failure is detected, it switches to the redundant drive module to maintain operation; if the sensor is damaged, the 3D printing device uses the thermoplastic material it carries to repair the sensor packaging layer on site and restore its function.

[0054] In one embodiment, the drone is further equipped with an environment adaptation module, specifically comprising:

[0055] Temperature and humidity sensors, anemometers, and electromagnetic field strength detectors adjust drone flight parameters and robotic arm control strategies based on real-time environmental data to ensure operational stability in strong winds, rain, fog, or high electromagnetic interference conditions.

[0056] In one embodiment, the method supports a multi-machine collaborative operation mode, specifically:

[0057] Multiple drones share mission maps and operation logs through the 5G communication network. The master drone dynamically allocates disassembly and assembly tasks, and the slave drones collaborate to complete the spacer replacement operations of a large-scale high-voltage line based on priority. The tactile data of each drone is aggregated through federated learning technology to continuously optimize the global control model.

[0058] Example 2: Application of a high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback in a high-voltage transmission line in mountainous areas.

[0059] The drone uses a six-rotor platform and is equipped with a lidar (Velodyne VLP-16), a high-precision inertial navigation system (INS), a binocular camera (Intel RealSense D455), a multimodal tactile sensor array (integrated at the end of the robotic arm) and an edge computing module (NVIDIA Jetson AGX Xavier).

[0060] The remote control terminal uploads the three-dimensional map of the target line and the spacer installation coordinates (accuracy ±5cm) and generates a task queue (replacing 20 spacers).

[0061] The drone hovers 20 meters above the cable, and the lidar scan generates three-dimensional point cloud data (resolution 2mm) of the high-voltage cable and surrounding terrain, identifying the direction of the wires and the initial position of the spacers.

[0062] The extended Kalman filter (EKF) algorithm is used to fuse the lidar point cloud with the gyroscope and accelerometer data of the INS to calculate the drone's attitude in real time (pitch angle error of ±0.1°), eliminating positioning drift caused by strong winds.

[0063] Using algorithms to avoid mountain obstacles, the drone approached the first target spacer at a speed of 3m / s and hovered 1.2 meters below it.

[0064] The tactile sensor array at the end of the robotic arm is activated, the piezoelectric force sensor (range 0-50N, accuracy ±0.5N) detects the contact force, the capacitive deformation sensor (sensitivity 0.1μm) captures the concave and convex features of the spacer surface, and the magnetostrictive torque sensor (range 0-10Nm) monitors the rotational resistance of the mounting bolt.

[0065] The tactile data is normalized and input into the pre-trained model (ResNet-50+BiLSTM), which outputs the recognition result: the target spacer is "rusted type" (confidence level 92%), and there is a risk of the middle bolt getting stuck.

[0066] Based on the identification results, the stiffness parameter K of the robotic arm was dynamically adjusted from 500N / m to 200N / m, and the damping parameter was increased from 80Ns / m to 120Ns / m to ensure smooth force application during disassembly and avoid bolt breakage.

[0067] The binocular camera captures the operating area at 30fps to generate an RGB-D image. Feature point matching (ORB algorithm) is synchronized with the tactile data timestamp to detect the relative position deviation between the end of the robotic arm and the spacer (lateral offset 3mm, angular deviation 1.5°).

[0068] Generate compensation commands to adjust the drone's hovering position (lateral movement 3mm, yaw angle correction 1.5°), and realign the end of the robotic arm with the spacer rod installation hole.

[0069] The robotic arm performs a counterclockwise rotation operation, and the tactile sensor provides real-time feedback of the torque value; when the torque suddenly increases to 8.5Nm (the preset threshold is 8Nm), an emergency stop is triggered and an alarm is sent back to the remote terminal.

[0070] The operator remotely switches to "flexible mode", and the robotic arm rotates in pulses (pausing for 0.5 seconds every 5° rotation) to gradually loosen the bolts and finally complete the disassembly.

[0071] The robotic arm clamps the new spacer and aligns it with the wire. The tactile sensor monitors that the installation force is stable within the range of 15±2N. The binocular vision confirms that the installation angle error is <1°, completing the locking.

[0072] The operation log (including tactile force curves, visual images, and control instructions) is divided into 128KB data blocks, encrypted using SHA-256 hash, and uploaded to the Huawei Cloud blockchain platform through smart contracts to ensure that the data cannot be tampered with.

[0073] During the disassembly process, the magnetostrictive torque sensor was damaged due to overload; the micro 3D printing device (equipped with PLA material) repaired the sensor packaging layer within 15 seconds, and the redundant tactile unit was simultaneously activated to ensure the continuity of subsequent tasks.

[0074] The master drone synchronizes the status of the completed five spacer replacements to the three slave drones via the 5G network, and dynamically allocates the remaining 10 tasks (slave drone B is responsible for areas with higher altitudes).

[0075] The tactile data of each machine (such as corrosion and deformation) is used to train a lightweight model locally, and the global dynamic impedance control parameters are updated through encrypted gradient aggregation, which reduces the overall disassembly force exceeding limit events by 35%.

[0076] When the anemometer detected an instantaneous wind speed of 12 m / s, the drone automatically increased the rotor speed to 8500 rpm and the robotic arm switched to "high damping mode" (B = 150 Ns / m) to ensure operational stability.

[0077] The electromagnetic field strength around the wire was detected to be 50μT, and the carbon fiber shield was activated to reduce the sensor signal noise to less than 5%.

[0078] The remote terminal retrieved the blockchain storage data and found that the installation of the No. 3 spacer rod took too long (due to severe rust). The deep learning model training set was optimized and 600 sets of "severe rust" category data were added.

[0079] The stiffness adjustment response time of dynamic impedance control is shortened from 200ms to 100ms, improving operational efficiency under complex working conditions.

[0080] In the 15 spacer replacement tasks, the positioning deviation was <2mm, the installation force fluctuation range was <±10%, and no wire damage or component falling off occurred.

[0081] Compared with traditional single-machine operations, the multi-machine collaborative mode saves 45% of working hours and achieves a task completion rate of 97% in complex mountainous environments.

[0082] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0083] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback, characterized in that: The following steps are involved: S1: The drone uses a fusion positioning module based on a lidar and inertial navigation system to obtain 3D point cloud data of the high-voltage cable. Combined with the preset spacer installation coordinates, it plans the flight path and reaches the target location. S2: The drone activates a tactile feedback module, which includes a multimodal tactile sensor array and an adaptive force-controlled robotic arm. The tactile sensor array is integrated on the surface of the robotic arm's end effector and is used to collect real-time three-dimensional force, torque, and surface deformation data when the robotic arm contacts the spacer rod, and transmits the data to the onboard processor via a high-speed communication module. S3: Based on the tactile data, the onboard processor uses a pre-trained deep learning model to identify the type, installation status, and corrosion level of the spacer. This deep learning model extracts the spatial features of the tactile signals through a convolutional neural network and analyzes the time series data using a long-short-term memory network to output a spacer status classification result. S4: Based on the recognition results, the dynamic impedance control algorithm calculates the desired stiffness and damping parameters of the manipulator in real time and generates force-position hybrid control instructions. The algorithm maps the tactile data into impedance parameter adjustments by establishing a dynamic model of the contact between the manipulator end and the environment, thereby achieving smooth control of the manipulator's movements. S5: During the disassembly and assembly process, the drone simultaneously activates the visual verification module, captures images of the operating area through the binocular camera, aligns them with the tactile data in time and space, detects positioning deviations, and generates compensation instructions; S6: When the robot arm performs a disassembly or installation operation, the tactile sensor monitors the contact force changes in real time. If the force exceeds the preset safety threshold, the emergency stop mechanism is immediately triggered and an alarm is sent back to the remote control terminal. S7: After completing a single operation, the drone generates an operation log through the edge computing module, recording tactile data, visual images, control instructions and timestamps, and encrypts and stores them in a cloud database using blockchain technology to ensure that the data cannot be tampered with; S8: The drone automatically switches to the next target location according to the preset task queue and repeats steps S1-S7 until all spacer disassembly and assembly tasks are completed; S9: After the mission is completed, the remote control terminal performs a full-process retrospective analysis based on the operation log to optimize the deep learning model parameters and control algorithm strategy; S10: When the self-repair mechanism detects an abnormality in the robotic arm or sensor, it repairs key components on-site through redundant module switching or 3D printing technology to ensure mission continuity.

2. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The specific implementation of the fusion positioning module is as follows: The lidar scans high-voltage cables and the surrounding environment to generate a three-dimensional point cloud map with millimeter-level accuracy; the inertial navigation system calculates the real-time attitude and position of the drone through gyroscope and accelerometer data; the lidar and INS data are fused through the extended Kalman filter algorithm to eliminate the cumulative error of a single sensor and achieve highly robust positioning in complex electromagnetic environments.

3. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The multimodal tactile sensor array comprises: Piezoelectric force sensor, capacitive deformation sensor and magnetostrictive torque sensor; the multimodal tactile sensor array is encapsulated with a flexible substrate material to adapt to the irregular shape of the high-voltage cable surface; the sensor data sampling frequency is higher than 1kHz to ensure accurate capture of the dynamic contact process.

4. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The training method of the deep learning model is: Collect tactile signal data sets of spacer bars under different working conditions, including normal installation, looseness, corrosion, and deformation; perform data normalization, time domain filtering, and feature enhancement processing; Transfer learning technology is used to extract tactile image features using a pre-trained ResNet-50 model, which is then combined with a bidirectional LSTM network to capture temporal dependencies, and finally the classification probability is output through a fully connected layer.

5. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The formula for the dynamic impedance control algorithm is: M x +B x +K(x-x d )=F ext Among them, M, B, and K are the inertia matrix, damping matrix, and stiffness matrix respectively; x is the actual position of the end of the robot arm, and x is the d is the desired position, F ext The external contact force fed back by the tactile sensor; the algorithm is based on F ext Adjust B and K according to the real-time changes to achieve a dynamic balance between force and position.

6. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The binocular camera captures RGB-D images at a rate of 30fps and aligns them with the timestamp of the tactile data through a feature matching algorithm. Stereo vision technology is used to reconstruct the three-dimensional structure of the operating area, which is compared with the lidar point cloud map. The positioning deviation is calculated and PID compensation instructions are generated to correct the drone's hovering position and the posture of the end of the robotic arm.

7. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The specific applications of blockchain technology in S7 are: The operation log is divided into multiple data blocks, each of which contains a hash value, timestamp, and hash pointer to the previous block. Smart contracts are used to verify data integrity, and only authorized terminals can decrypt and access log contents, preventing operation data from being maliciously tampered with or leaked.

8. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1 is characterized in that: The self-repair mechanism comprises: Redundantly driven robotic arm joint module, spare tactile sensor unit and micro 3D printing device; when a joint motor failure is detected, it switches to the redundant drive module to maintain operation; if the sensor is damaged, the 3D printing device uses the thermoplastic material it carries to repair the sensor packaging layer on site and restore its function.

9. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1, characterized in that: The drone is also equipped with an environment adaptation module, specifically including: Temperature and humidity sensors, anemometers, and electromagnetic field strength detectors adjust drone flight parameters and robotic arm control strategies based on real-time environmental data to ensure operational stability in strong winds, rain, fog, or high electromagnetic interference conditions.

10. The high-precision spacer assembly and disassembly control method for a UAV based on tactile feedback according to claim 1, characterized in that: The method supports multi-machine collaborative operation mode, specifically: Multiple drones share mission maps and operation logs through the 5G communication network. The master drone dynamically allocates disassembly and assembly tasks, and the slave drones collaborate to complete the spacer replacement operations of a large-scale high-voltage line based on priority. The tactile data of each drone is aggregated through federated learning technology to continuously optimize the global control model.

Images