An unmanned aerial vehicle path planning and precision dropping method based on power transmission line visual guidance

Abstract

This invention provides a method for UAV path planning and precise delivery based on power transmission line vision guidance. The invention constructs a multi-source perception and state feedback system integrating vision, RTK, and IMU to achieve stable perception of power transmission line targets and UAV motion states. An improved vision processing method is used to obtain the main direction of the power transmission line, key structural points, and visual perception confidence levels. This is combined with a camera imaging model, fused pose, and safe altitude offset to generate an approach target point. Furthermore, a global approach path constrained by the target point and a local path smoothing optimization mechanism driven by fused state feedback are constructed to achieve efficient coordination between path planning and end-point delivery control. Based on this, the delivery position is compensated and corrected according to the UAV's flight state and environmental disturbances. Through delivery feasibility judgment and precise end-point alignment control, high-precision and safe delivery operations are achieved in complex power transmission line environments.

Description

Technical Field

[0001] This invention relates to the field of intelligent inspection of power transmission lines and autonomous operation of unmanned aerial vehicles (UAVs), and in particular to a method for UAV path planning and precise deployment based on visual guidance of power transmission lines. Background Technology

[0002] With the continuous expansion of power systems, the demand for transmission line inspection and maintenance is increasing. Traditional manual inspection methods suffer from low efficiency, high labor intensity, and high safety risks, especially in complex terrain and high-altitude operating environments, making it difficult to meet the safety and intelligence requirements of modern power grids. In recent years, drones, due to their advantages of high mobility, wide coverage, and flexible deployment, have been gradually applied to scenarios such as live-line defect detection and auxiliary operations on transmission lines, becoming an important technical means for intelligent operation and maintenance of transmission lines.

[0003] During power transmission line operations, drones are required to perform tasks such as deploying inspection robots, delivering traction ropes, or assisting in installation at specific locations. However, power transmission lines typically have a slender structure, small diameter, and complex spatial distribution. They often form a dense working environment together with components such as towers, fittings, and insulator strings. In addition, there may be irregular obstacles such as trees and buildings in the surrounding area, which places high demands on the drones' environmental perception, path planning, and precise deployment capabilities.

[0004] Existing UAV obstacle avoidance and path planning methods mostly rely on a single sensor. For example, vision-based methods are easily affected by changes in lighting, background interference, and insufficient long-range resolution, making it difficult to reliably identify small power transmission line targets. While methods relying on complex 3D mapping equipment can acquire spatial information, they increase the UAV's payload, power consumption, and system integration complexity, making them unsuitable for lightweight operation platforms with limited payload. Furthermore, traditional path planning methods often employ single shortest path or local path optimization strategies, lacking a collaborative mechanism between target point generation, fusion of state feedback, and proximity path planning. This makes it difficult to balance path accessibility, terminal stability, and operational safety in power transmission line proximity operations.

[0005] On the other hand, in precision delivery operations, existing technologies typically rely on GPS positioning or simple visual guidance for delivery control, lacking comprehensive compensation for factors such as drone attitude changes, flight speed, and environmental disturbances. This can easily lead to significant deviations in delivery location, making it difficult to meet the requirements of precision operations on power transmission lines. Furthermore, the lack of effective coordination between obstacle avoidance control and delivery control may result in conflicts between obstacle avoidance strategies and delivery requirements when approaching the target area, further impacting operational safety and accuracy.

[0006] While existing technologies can utilize sensor information such as vision, RTK, and IMU for UAV positioning, obstacle avoidance, and power line identification, most solutions still treat target recognition, state estimation, path planning, and delivery control as relatively independent functional modules. For precise power line delivery scenarios, the conductor targets are slender, and their visual features are easily affected by background interference and aircraft attitude disturbances. Generating target points solely based on single-frame recognition results can easily lead to approach direction deviations. If the path planning stage relies solely on a fixed safety distance or fixed weights for searching, it becomes difficult to adjust the approach strategy in real time based on the reliability of visual perception and the aircraft's motion state. If the delivery stage is triggered solely by the UAV's current position, it is difficult to compensate for delivery deviations caused by flight speed, attitude changes, and release delays. Therefore, the key to UAV delivery operations on power lines lies not in improving the accuracy of a single sensor, but in establishing a closed-loop collaborative mechanism between conductor visual perception results, UAV state feedback, path planning weights, and delivery triggering conditions.

[0007] Therefore, how to achieve stable identification, target point generation, and high-precision fusion positioning of small transmission lines in complex transmission line environments by relying on lightweight sensing information such as vision, RTK, and IMU, and on this basis, construct an approach path planning and local smoothing optimization mechanism under target point constraints, while completing high-precision delivery control, has become a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0008] Purpose of the Invention: Addressing the lack of closed-loop coordination between transmission line target identification, UAV state estimation, approach path planning, and delivery control in existing technologies, which leads to target point instability in complex backgrounds, kinematic abrupt changes in approach trajectories, and the susceptibility of delivery triggering timing and delivery location to attitude disturbances, this invention proposes a UAV path planning and precise delivery method based on transmission line vision guidance. This method does not simply superimpose vision, RTK, and IMU; instead, it incorporates visual perception confidence, comprised of conductor characteristic response, line characteristic continuity, and multi-frame stability, into the processes of approach target point generation, path planning weight adjustment, local trajectory smoothing, and delivery feasibility determination. Simultaneously, it combines state feedback information such as UAV position, speed, attitude, and attitude angle change rate to dynamically adjust the approach direction, safety distance, trajectory smoothness, and delivery compensation, thereby forming a closed-loop control mechanism for perception-localization-planning-delivery operations on transmission lines.

[0009] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0010] A method for UAV path planning and precise deployment based on power transmission line vision guidance includes the following steps: S1. Construct a multi-source perception and operation execution system consisting of a visual sensor, an RTK positioning module, an IMU inertial measurement unit, a flight control module, and a delivery execution mechanism. Collect visual images, RTK positioning data, IMU attitude data, and UAV flight status data of the power transmission line operation area, and complete time synchronization and coordinate system one. S2. Perform adaptive edge detection and weighted line feature extraction on the visual image, and perform attitude compensation on the line feature extraction results in combination with the IMU attitude data to obtain the main direction of the conductor, key structural points of the conductor, visual relative observation of the conductor, and visual perception confidence. S3. Determine the reference position of the traverse based on the main direction of the traverse and the key structural points of the traverse, and apply a safety height offset along the safety offset direction at the reference position of the traverse to generate a point close to the target operation; at the same time, fuse the RTK positioning data, the IMU attitude data and the visual relative observation of the traverse to obtain state estimation results including the position, velocity, attitude and attitude angle change rate of the UAV. S4. Generate a global approach path based on the approach target point, the state estimation result, the preset guide safety distance constraint and the guide approach direction consistency constraint; and dynamically adjust the target tracking weight, safety distance weight and trajectory smoothing weight based on the visual perception confidence and the attitude angle change rate. Based on the adjusted weights, perform local smoothing optimization on the global approach path to obtain a smooth approach trajectory with continuous position, velocity and acceleration. S5. When the UAV enters the preset delivery window along the smooth approach trajectory, and the visual perception confidence, attitude angle change rate, trajectory safety margin determined based on the smooth approach trajectory, and position error all meet the corresponding thresholds, the target delivery position is compensated and corrected according to the speed, attitude, and visual observation residual in the state estimation results to obtain the corrected target delivery position, and the delivery execution mechanism is controlled to complete the precise delivery of the power transmission line at the corrected target delivery position.

[0011] As a preferred embodiment, the visual perception confidence level It is obtained using the following formula:

[0012] in, This represents the normalized value of the linear characteristic response intensity. Indicators representing the continuity of line characteristics This represents the stability index of the main direction of the conductor and the key structural points of the conductor across multiple consecutive frames. , and Here are the confidence weighting coefficients, and When the visual perception confidence level When the target tracking weight is lower than the preset confidence threshold, the target tracking weight is reduced, while the safety distance weight and the trajectory smoothing weight are increased.

[0013] As a preferred embodiment, the attitude compensation includes: constructing an attitude rotation matrix based on the roll angle, pitch angle, and yaw angle output by the IMU attitude data; transforming the characteristic direction of the traverse line in the image coordinate system to the body coordinate system or the world coordinate system; and using the attitude change amount in consecutive frames to correct the main direction of the traverse line, so as to reduce the impact of UAV body jitter and viewpoint changes on the traverse line recognition results.

[0014] As a preferred embodiment, the approach to the target work point Generate according to the following formula:

[0015] in, Indicates the reference position of the conductor. Indicates the safety height offset. Represents the unit vector in the vertical direction of the world coordinate system. This indicates the lateral safety offset. The unit vector representing the safety offset direction perpendicular to the main direction of the conductor; the approach target point is located in the safety space above or to the side of the conductor area.

[0016] As a preferred embodiment, the state estimation result is obtained by fusing the RTK positioning data, the IMU attitude data, and the visual relative observation of the traverse; wherein, the RTK positioning data is used to constrain the global position drift of the UAV, the IMU attitude data is used for high-frequency state propagation, and the visual relative observation of the traverse is used to correct the local position and attitude deviation of the UAV relative to the target area of ​​the traverse.

[0017] As a preferred embodiment, the path cost function for generating the global proximity path is... It includes path length cost, safety distance cost, and proximity direction consistency cost, and its expression is:

[0018] in, This represents the path length cost. This represents the safety distance cost of a path node relative to the transmission line safety buffer zone. This represents the proximity direction consistency cost between the path terminal velocity direction and the main direction of the conductor. , and The corresponding weighting coefficients are used; the UAV enters the preset deployment window along a direction that is consistent with or approximately consistent with the main direction of the guide wire by means of the proximity direction consistency cost constraint.

[0019] As a preferred embodiment, the target tracking weight, the safety distance weight, and the trajectory smoothing weight are based on the visual perception confidence level. and the rate of change of attitude angle Dynamic adjustment: When rise and When the target tracking weight is reduced, the target tracking weight is increased; when Reduce or When the distance increases, the safety distance weight and the trajectory smoothing weight are increased.

[0020] As a preferred embodiment, the local smoothing optimization uses quintic spline interpolation to fit local path segments in the global approach path, so that the generated smooth approach trajectory is continuous in terms of position, velocity, and acceleration, and infeasible trajectories are eliminated based on the maximum speed, maximum acceleration, maximum climb angle, minimum turning radius of the UAV, and the safety distance constraints of the guide wire.

[0021] As a preferred embodiment, the triggering conditions for the preset deployment window include: the position error between the current position of the UAV and the target deployment position before compensation is less than the position tolerance threshold; the attitude angle change rate is less than the attitude stability threshold; the minimum safe distance between the smooth approach trajectory and the power transmission line and surrounding obstacles in the prediction time domain is greater than the safe distance threshold to meet the trajectory safety margin; and the visual perception confidence is greater than the confidence threshold.

[0022] As a preferred embodiment, the modified target delivery location The compensation and correction shall be performed according to the following formula:

[0023] in, This indicates the target deployment location before compensation. This represents the velocity vector in the state estimation result. Indicates the delay time for deployment execution. Represents the roll angle from the state estimation results. Pitch angle and heading angle The constructed attitude rotation matrix, This represents the unit vector indicating the direction of delivery. This represents the visual observation residual. , and These represent the velocity compensation coefficient, attitude compensation coefficient, and visual residual compensation coefficient, respectively.

[0024] As a preferred embodiment, step S1 specifically includes: Construct a UAV multi-source perception and state feedback system consisting of the visual sensor, the RTK positioning module, the IMU inertial measurement unit, and the flight control module; The modules within the multi-source sensing and state feedback system are synchronized in time and calibrated in space to unify the acquired data into the same coordinate system. The drone is controlled to fly to the target power transmission line operation area and continuously scans and collects visual images during the flight. The RTK positioning data, IMU attitude data, and flight status data of the UAV are acquired simultaneously. The data collected at the same or adjacent times are time-aligned and correlated to form a multi-source sensing dataset under the current operating state for use in subsequent steps.

[0025] As a preferred embodiment, step S2 specifically includes: The visual images in the multi-source perception dataset are preprocessed, and candidate edge features of the transmission line are extracted through adaptive edge detection. Based on the candidate edge features, weighted line feature extraction is performed to obtain the transmission line route feature parameters; IMU attitude compensation and sliding filtering are performed by combining the IMU attitude data in the multi-source sensing dataset to correct the impact of UAV flight disturbances on the line feature extraction results. Based on the attitude compensation results, the characteristic parameters of the power transmission line and the camera imaging model, the target point is initially calculated to obtain the visual relative observation of the conductor, the main direction of the conductor and the key structural points of the conductor. The visual perception confidence is calculated based on the line feature response intensity, line feature continuity, and multi-frame stability. The main direction of the traverse, the key structural points of the traverse, the visual relative observation of the traverse, and the visual perception confidence are encapsulated into path planning perception information and output to the next stage.

[0026] As a preferred embodiment, step S3 specifically includes: The reference position of the traverse is determined based on the key structural points of the traverse in the path planning perception information, and the tangential direction of the local coordinate system is determined based on the main direction of the traverse. Based on the reference position of the guide wire, a safety height offset and a lateral safety offset are applied to generate the approach target point; Construct a fusion state vector that includes the UAV's position, velocity, attitude, gyroscope bias, and accelerometer bias; The state estimation result at the current moment is obtained by using the IMU attitude data for high-frequency state propagation, using the RTK positioning data to constrain the global position drift, and using the visual relative observation of the guide line to correct the local pose deviation of the UAV relative to the guide line target area. The approach target point, the state estimation result, the main direction of the traverse, and the visual perception confidence are uniformly encapsulated and output.

[0027] As a preferred embodiment, step S4 specifically includes: Based on the current location of the UAV, the approach target point, the safety buffer zone of the guide wire, and the surrounding obstacles, a global approach path search space is constructed. Under the conditions of satisfying all dynamic constraints and the safety distance constraint of the conductor, the global proximity path is generated by searching using the path cost function; The target tracking weight, the safety distance weight, and the trajectory smoothing weight are dynamically adjusted based on the visual perception confidence level and the attitude angle change rate in the state estimation result. The local reference path segment near the current time is extracted from the global approach path, and the local smoothing optimization is performed using the fifth-order spline interpolation to generate the smooth approach trajectory. The safety of the smooth approach trajectory is determined, and the conflict-free trajectory with the highest comprehensive evaluation value is output as the optimal local approach trajectory.

[0028] As a preferred embodiment, step S5 specifically includes: Based on the state estimation results and the local reference path segment, it is determined whether the UAV has entered the preset deployment window; When all the triggering conditions related to the visual perception confidence level, the attitude angle change rate, and the trajectory safety margin are met, it is determined that the current moment meets the deployment requirements. The corrected target deployment position is calculated and obtained by using the current velocity, attitude angle change and visual observation residuals from the state feedback. The drone is controlled to perform end-point precision alignment based on the corrected target deployment location. When the positional error of the end-point precision alignment is less than the preset precision alignment error threshold, the delivery execution mechanism is controlled to release the target object, thus completing the precise delivery operation of the power transmission line.

[0029] Compared with the prior art, the present invention has at least the following beneficial effects: 1. This invention uses the characteristic response of the conductor line, the continuity of the line feature, and the stability of multiple frames to form the visual perception confidence, and introduces it into the path planning weight adjustment and deployment feasibility judgment process, thus avoiding the target point jump and false triggering deployment problems caused by the traditional method relying only on the visual recognition result of a single frame.

[0030] 2. This invention generates approach target points with safety height offset and lateral safety offset based on the main direction of the conductor and key structural points, and introduces approach direction consistency constraints in the path planning cost function, enabling UAVs to enter the operation window along a direction that conforms to the spatial orientation of the transmission line, thereby reducing the safety risks caused by vertically cutting into the conductor area.

[0031] 3. This invention dynamically adjusts the target tracking weight, safe distance weight, and trajectory smoothing weight based on visual perception confidence and the UAV attitude angle change rate. When visual observation is reliable and the aircraft is in a stable state, it improves approach efficiency and improves approach conservatism when visual confidence is low or attitude disturbance is large, thereby taking into account path accessibility, flight stability, and operational safety.

[0032] 4. This invention comprehensively considers UAV speed, deployment delay, attitude changes, and visual observation residuals during the deployment phase, dynamically compensating and correcting the target deployment position. Furthermore, it uses visual confidence, attitude stability, trajectory safety margin, and position error as deployment triggering conditions, thereby improving deployment accuracy and operational reliability in complex power transmission line environments. Attached Figure Description Figure 1 This is a flowchart of the method proposed in this invention.

[0033] Figure 2 This is a flowchart of the UAV path planning and precise delivery algorithm for power transmission lines in this invention.

[0034] Figure 3 This is a flowchart of the calculation of visual confidence of the traverse and the generation of target points in this invention.

[0035] Figure 4 This is a schematic diagram illustrating the generation of the conductor safety offset target point in this invention.

[0036] Figure 5 This is a schematic diagram illustrating the collaborative adjustment of local path weights by visual confidence and state feedback in this invention.

[0037] Figure 6 This is a schematic diagram of the dynamic compensation of the target placement location and the determination of the placement trigger in this invention. Detailed Implementation

[0038] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0039] like Figure 1 As shown, this invention presents a method for UAV path planning and precise deployment based on power transmission line vision guidance, which specifically includes the following steps: S1. Construct a multi-source perception and state feedback system consisting of a visual sensor, an RTK positioning module, an IMU inertial measurement unit, and a flight control module to collect visual images, RTK positioning data, IMU attitude data, and UAV flight status data of the power transmission line operation area. Step S1 specifically includes: S11. Construct a UAV multi-source perception and state feedback system consisting of a visual sensor, an RTK positioning module, an IMU inertial measurement unit, and a flight control module. This system is used to collect visual images, positioning data, attitude data, and velocity status information of the power transmission line operation area. Specifically, the visual sensor is used to collect image information of the power transmission line operation area, the RTK positioning module is used to provide global position constraints, the IMU inertial measurement unit is used to acquire real-time attitude, angular velocity, and acceleration information of the UAV, and the flight control module is used to realize flight state calculation and trajectory control.

[0040] S12. Perform time synchronization and spatial extrinsic parameter calibration on the visual sensor, RTK positioning module, IMU inertial measurement unit, and flight control module to unify image data, positioning data, attitude data, and UAV pose information into the same coordinate system. Specifically, calibrate the extrinsic parameter relationships between the camera coordinate system, body coordinate system, and world coordinate system to satisfy:

[0041] Where is the point coordinate in the camera coordinate system, and is the spatial transformation matrix from the body coordinate system to the camera coordinate system. Furthermore, the target direction in the world coordinate system is projected onto the image plane, satisfying:

[0042] Where is the image pixel coordinate, is the camera intrinsic parameter matrix, and are the rotation matrix and translation vector from the world coordinate system to the camera coordinate system, respectively, is the target point coordinate in the world coordinate system, and is the scaling factor. Through the above calibration relationship, unified registration of visual data, RTK positioning data, and IMU attitude data is achieved.

[0043] S13. Control the drone to enter the power transmission line operation area and perform patrol data collection; according to the preset inspection route or mission path, control the drone to fly to the target power transmission line operation area, continuously scanning and collecting data on the environment in front of and around the operation area during flight, and acquiring real-time environmental information on power transmission lines, towers, fittings, tension clamps, and surrounding obstacles. The drone is constantly... The pose can be represented as:

[0044] Here, represents the UAV's position coordinates in three-dimensional space, and represents the roll angle, pitch angle, and yaw angle, respectively. These pose parameters are used for subsequent image ranging compensation, state information coordinate transformation, and path planning control.

[0045] S14. Acquire visible light image data of the transmission line operation area; continuously acquire image sequences of the transmission line operation area using the visible light camera to obtain visual information about the transmission line itself, the target operation location, and the surrounding environment. (Time recorded) The captured image frames are:

[0046] in, Indicates the first Frame image, Indicates the camera imaging process, This indicates the acquisition time. The image sequence is used for subsequent transmission line feature extraction and target recognition.

[0047] S15. Collect RTK positioning data, IMU attitude data, and UAV velocity and status data; obtain the UAV's global position through the RTK positioning module, obtain attitude angle, angular velocity, and acceleration information through the IMU inertial measurement unit, and output the UAV's current velocity and control status in conjunction with the flight control module. The fused status data can be represented as:

[0048] in, This indicates the UAV's position, speed, attitude, and related sensor status information at any given time. The fused status data is used to characterize the UAV's real-time motion state during the approach operation, providing status feedback for subsequent fusion odometry, target point tracking, and path planning.

[0049] S16. Synchronously acquire UAV pose information; acquire the UAV's position, velocity, and attitude information in real time through the inertial measurement unit and positioning module, wherein the velocity vector can be expressed as:

[0050] The attitude angle information is output by the inertial measurement unit, and combined with the positioning module, the complete state variables of the UAV can be obtained:

[0051] in, Indicates the drone at a certain time The status information is used for subsequent attitude compensation ranging, trajectory prediction, and control decisions.

[0052] S17. Construct a multi-source sensing dataset for the current moment; perform time alignment and data association on visible light images, traverse visual features, RTK positioning data, IMU attitude data, and UAV flight status information acquired at the same or adjacent moments to form a multi-source sensing dataset for the current operational state:

[0053] in, This represents the multi-source sensing dataset at time k. This multi-source sensing dataset is input into subsequent power transmission line identification, target point generation, fusion positioning, path planning, and precision operation modules to achieve smooth approach and high-precision operation of UAVs for precise deployment to power transmission lines.

[0054] Based on the multi-source sensing dataset constructed in step S1, step S2 further performs visual recognition, target point generation, and status feedback on the transmission lines, providing a preliminary sensing foundation for subsequent multi-source fusion localization, local path smoothing optimization, and precise deployment. Figure 2 As shown, the algorithm block diagram of this invention includes a hardware layer, a conductor identification and target point generation module, a multi-source fusion odometry module, and a path planning module. The hardware layer includes a visual sensor, an RTK positioning module, and an IMU inertial measurement unit, used to input visual images, positioning data, attitude data, and flight status information. The conductor identification and target point generation module sequentially includes adaptive edge detection, line feature extraction, multi-frame information fusion, and conductor principal direction estimation, used to generate approach target points and visual perception confidence scores. The multi-source fusion odometry module integrates RTK, IMU, and visual observations to output high-precision UAV position, velocity, and attitude. The path planning module includes global approach path generation under target point constraints and local path smoothing optimization combined with state feedback, used to output an executable smooth approach trajectory that eliminates kinematic abrupt changes and continuous control commands, ultimately controlling the UAV to complete the power transmission line path planning and precise delivery operation.

[0055] S2. Input the visual image into the conductor recognition and target point generation module, and sequentially perform adaptive edge detection, line feature extraction, multi-frame information fusion, and conductor main direction estimation to generate target points for the approach operation; step S2 specifically includes: S21. The visual image is preprocessed, and the edge features of the transmission line are extracted using adaptive edge detection. Specifically, for each time... Collected visual images Grayscale conversion and filtering / denoising are performed to obtain the preprocessed image. Let the mean gradient magnitude of the preprocessed image be... The standard deviation of the gradient magnitude is Then the high threshold and low threshold of adaptive edge detection are expressed as follows:

[0056]

[0057] in, Indicates time High threshold, Indicates time The low threshold For high threshold adjustment coefficient, It is the ratio coefficient between high and low thresholds, and satisfies... Edge detection is performed based on the high and low thresholds to obtain a candidate edge set for the transmission line. .

[0058] S22. Based on the candidate edge set, perform line feature extraction to obtain the transmission line route feature parameters. Specifically, the candidate edge set... The gradient strength and orientation consistency of each edge point are introduced into the parameter space as weights to construct a weighted accumulator function:

[0059] in, Indicates time The weighted cumulative value, and These represent the distance and angle parameters of the line, respectively. This represents the cumulative decision function. Represents edge points The weights. Represented as:

[0060] in, Represents edge points gradient magnitude, This indicates the consistency index of edge point orientation. and Let be the weight coefficient, and satisfy... By extracting the peak value corresponding to the weighted accumulator This allows us to obtain the line characteristic parameters of the transmission line.

[0061] S23. Combine the attitude data to perform IMU attitude compensation and sliding filtering to correct the impact of UAV flight disturbances on the line feature extraction results. Specifically, suppose the UAV at time... The attitude angle vector is:

[0062] in, , and These represent the roll angle, pitch angle, and yaw angle, respectively. To reduce the impact of flight vibrations and instantaneous attitude fluctuations on the recognition results, a sliding filter is applied to the attitude angle vector to obtain a smoothed attitude angle vector:

[0063] in, This represents the filtered attitude angle vector. This represents the length of the sliding window. Attitude compensation is performed on the transmission line feature parameters based on the smoothed attitude angle vector to obtain the compensated line feature parameters. .

[0064] S24. Based on the attitude compensation results and the characteristic parameters of the power transmission line, perform improved center projection ranging to obtain the target distance and precise location information between the UAV and the power transmission line. Specifically, let the camera focal length be... The reference dimensions for transmission lines are The feature pixel difference in the online image of the transmission line is The target distance between the UAV and the power transmission line is then obtained based on the improved central projection model. for:

[0065] in, Here is the attitude compensation factor. The attitude compensation factor is expressed as:

[0066] To improve ranging stability, the ranging results at consecutive time points are smoothed to obtain the corrected target distance:

[0067] in, This indicates the corrected target distance. This represents the length of the ranging smoothing window. By combining the current pose of the UAV with the compensated line characteristic parameters, the precise position information of the power transmission line relative to the UAV can be further determined.

[0068] S25. Calculate the visual perception confidence score based on the transmission line identification results, ranging results, and continuous frame stability. Specifically, the visual perception confidence score is constructed based on the response intensity of line feature extraction, the continuity of the conductor centerline, and the stability of target point generation across multiple frames. Its expression is:

[0069] in, The normalized index representing the response of line feature extraction. Indicators representing the geometric continuity of transmission line routes. This indicates the spatial stability index generated near the target point of the operation. and Let be the confidence weighting coefficient, and satisfy . The visual perception confidence score is used for adaptive adjustment of the evaluation weights of the subsequent local path smoothing optimization algorithm, as well as for preliminary determination of the feasibility of precise delivery at the last mile.

[0070] S26. Output the precise location of the transmission line, the target distance, and the visual perception confidence level, and generate path planning perception information. Specifically, the compensated line feature parameters, the corrected target distance, and the visual perception confidence level are uniformly encapsulated to form a visual ranging output set:

[0071] in, Indicates time The perception output set includes compensated line feature parameters, confidence scores, and generated proximity target points. This perception output set serves as input to the path planning module, used to generate a global proximity path under target point constraints, and provides a highly reliable perception foundation for subsequent local trajectory smoothing optimization and precise end-point delivery.

[0072] S3. Based on the main direction of the transmission line, key structural points, RTK positioning data, and IMU attitude data, generate an approach target point, and integrate RTK, IMU, and visual observations to construct a multi-source fusion odometry, outputting the UAV's position, velocity, and attitude status; Step S3 specifically includes: S31. Construct a conductor geometric constraint model based on the main direction and key structural points of the transmission line. Specifically, extract key points on the centerline of the conductor in the image domain, and map the key points in the image domain to spatial direction vectors by combining the camera imaging model, the current pose of the UAV, and the extrinsic parameters of the body coordinate system.

[0073] in, K represents the normalized direction vector of the key points of the traverse in the camera coordinate system, and K represents the camera intrinsic parameter matrix. This represents the pixel coordinates of key points on the conductor.

[0074] S32. Generate a near-operation target point based on the conductor geometric constraint model. Specifically, according to the conductor's main direction, spatial direction vector, and transmission line operation safety requirements, apply a safety height offset along the vertical direction of the world coordinate system above the conductor reference point to obtain the terminal target point for path planning;

[0075] The terminal target point is used to constrain the UAV to stably approach the safe space above the guide wire area from the takeoff point, and to provide a navigation target for subsequent path planning and terminal deployment. Further, the safety cost under the target point constraint is defined as:

[0076] in, Indicates the path length cost. This represents the distance cost of a path node relative to the transmission line safety buffer zone. This represents the cost of approaching directional consistency. and Let these be the path planning weight coefficients, and satisfy the following:

[0077] The cost of the safe distance can be expressed as:

[0078] Here, represents the minimum distance from the safety node unit to the centerline of the transmission line, and is the buffer attenuation coefficient. Based on the above definition, the closer a safety node is to the transmission line or a high-safety-constraint area, the greater its safety constraint cost.

[0079] S33. Determine the starting node and target node for global path planning; specifically, determine the starting node based on the current pose of the UAV. Determine the target node based on the target location or the preset operation location. Let any node in the path search space be denoted as . The candidate path from the starting node to the target node is denoted as:

[0080] in, This represents a candidate global proximity path from the starting node to the target node. Indicates a path node; S34. Construct a proximity path planning cost function based on the target point constraints and safety costs. Specifically, define a comprehensive evaluation function for each path node as follows:

[0081] Where g(n) represents the cumulative actual cost from the starting node to the current node, and h(n) represents the heuristic cost from the current node to the target node. This represents the security constraint cost of the current node;

[0082] in, This represents the Euclidean distance between adjacent nodes. Represents a node The safety constraint cost is λ, where λ is the safety constraint adjustment coefficient and m represents the number of path segments from the starting node to the current node.

[0083] The heuristic cost is expressed as:

[0084] in, This represents the Euclidean distance from the current node to the target point. This represents the safe distance heuristic within the current node's neighborhood, where η is the heuristic adjustment coefficient. The safe distance heuristic is defined as follows:

[0085] Here, represents the neighborhood set corresponding to the node, and represents the normalized safety distance constraint value. By introducing the target point constraint and safety distance constraint into the heuristic function, the path planning algorithm actively maintains a safe distance from the dangerous boundary of the transmission line during the search process.

[0086] S35. Perform a global path search using a path planning algorithm to generate a global approach path. Specifically, under the premise of satisfying the safety distance threshold of the guide wire and flight kinematic constraints, search for the path with the minimum comprehensive cost from the takeoff area to the point of approach to the operational target: in, This represents the generated globally optimal nearest path.

[0087] S36. Perform local path smoothing optimization by fusing multi-source state feedback. Specifically, to eliminate abrupt path transitions and suppress body oscillations, the global approach path is smoothed using fifth-order spline interpolation, combining real-time pose and velocity feedback from multi-source fusion odometry outputs, to obtain an executable trajectory with continuous position, velocity, and acceleration.

[0088]

[0089] in, The coefficients are polynomials, which are solved by the current flight state, target point constraints, and dynamic constraints to ensure smooth and stable kinematics during the approach process.

[0090] S37. Output the global proximity path and input it into the local path optimization module. Specifically, the smoothed optimized global proximity path, safety constraint information, and target point information are uniformly encapsulated into a global planning output set:

[0091] in, The global planning output set represents the time step. This global planning output set is input into the subsequent local path optimization module, where it, along with visual perception confidence and fused state feedback, participates in the approach operation path optimization decision.

[0092] S4. Based on the approach target point and the multi-source fusion odometer output status, generate a global approach path under the target point constraint, and perform local path smoothing optimization by combining the safety distance constraint, approach direction consistency constraint, and status feedback. Step S4 specifically includes: S41. Establish a local path optimization model suitable for near-operation scenarios on power transmission lines. Specifically, based on the current position, velocity, attitude, and flight control constraints of the UAV, construct a set of feasible trajectories for the UAV within the current local space. Predict the future short-time domain motion trajectory of the UAV based on each set of candidate control commands, forming a set of candidate local trajectories. Let the candidate control command be , where represents the linear velocity of the UAV and represents the angular velocity of the UAV. Then, the comprehensive evaluation function of the candidate trajectory can be expressed as:

[0093] in, This represents the comprehensive evaluation value of the candidate trajectory τ. Let S(τ) represent the target point tracking evaluation term, S(τ) represent the safety distance evaluation term, V(τ) represent the velocity evaluation term, and C(τ) represent the trajectory smoothness evaluation term. and These represent the weight coefficients of each evaluation item.

[0094] S42. Adaptively adjust the weights for local path optimization based on visual perception confidence and fusion state feedback. Specifically, the visual perception confidence output in step S2 and the UAV position, velocity, and attitude state output in step S3 are incorporated into the weight adjustment process. When the visual perception confidence is high and the state estimation is stable, increase the weights of the target point tracking evaluation item and the approach direction consistency evaluation item; when the visual perception confidence is low or the attitude fluctuation is large, increase the weights of the safety distance evaluation item and the smoothness evaluation item to enhance the conservatism and stability of the approach process. The weight coefficients satisfy the following relationship:

[0095]

[0096]

[0097] in, and Let represent the initial weights, and α, β, and γ represent adjustment coefficients. To ensure the stability of the comprehensive evaluation, it is preferable to normalize each weight to satisfy the following:

[0098] S43. Construct a local trajectory tracking reference based on the global approach path, and establish a collaborative mechanism between the global approach path and local smoothing optimization. Specifically, select a local reference path segment near the current time from the global approach path output in step S3 as the heading guidance basis for local path optimization. The collaborative mechanism adopts a combination of global target point guidance and local trajectory correction, where the global approach path is used to provide the overall path trend towards the target area, and local path optimization is used to perform smoothing correction under real-time status feedback.

[0099] During the flight of the UAV, if the flight status is stable and the safety margin is sufficient, it will preferentially fly along the global approach path; if local environmental constraints, attitude fluctuations or target point observations change, local trajectory replanning will be performed under the global path constraints to output a local approach trajectory that meets the safety distance and smoothness constraints.

[0100] S44. Integrate visual ranging results and state feedback information to calculate the safety constraint index of candidate local trajectories. Specifically, the distance information between key structural points and targets of the transmission line output in step S2, and the fused state information output in step S3, are used together for local trajectory safety assessment. For each candidate trajectory, calculate its minimum safe distance from the transmission line and surrounding obstacles in the prediction time domain, and determine whether it meets the preset safety threshold. Let a candidate trajectory be , then its safety margin can be expressed as:

[0101] in, Representing candidate trajectories Minimum safety margin, Represents trajectory points Distance to the nearest obstacle or danger boundary of the transmission line. When If the value is less than a preset safety threshold, the candidate trajectory is determined to be infeasible and removed from the candidate trajectory set.

[0102] S45. Select the optimal local approach trajectory from the candidate trajectories that meet the safety constraints, and output the local path optimization result. Specifically, calculate the comprehensive evaluation value for each candidate trajectory that meets the safety constraints, select the candidate trajectory with the largest comprehensive evaluation value as the optimal local approach trajectory at the current moment, and generate the corresponding speed control command to send to the flight control module. The optimal local trajectory satisfies:

[0103] in, This represents the set of candidate trajectories that satisfy the safety constraints. This represents the comprehensive evaluation value of the candidate trajectory. This yields the local path optimization result at the current moment, which is used to guide the UAV in maintaining a safe distance from power lines and surrounding obstacles.

[0104] S46. The local path optimization results are output in conjunction with the global path information. Specifically, the optimal local approach trajectory at the current moment, the corresponding speed control command, and the global path reference segment are uniformly encapsulated to form a path planning output set:

[0105] in, This represents the optimal local approximation trajectory. This indicates the corresponding speed control command. This represents a local reference segment of the global approach path. The path planning output set serves as the motion decision-making basis for subsequent target placement position correction and precise placement control, enabling the UAV to complete power transmission line operation tasks while ensuring a safe distance.

[0106] S5. Based on the precise location, target distance, global approach path, and local path optimization results, determine the target delivery location and perform compensation corrections, then control the drone to complete the precise delivery operation of the power transmission line. Step S5 specifically includes: S51. Determine the target deployment area and deployment triggering conditions. Specifically, based on the precise location of the power transmission line and the target distance information output in step S2, combined with the global approach path output in step S3 and the local path optimization results output in step S4, determine the current spatial position of the UAV relative to the power transmission line, and select a target deployment area that meets the operational requirements. The deployment triggering conditions include at least: the UAV is located within a preset deployment window, the flight attitude meets stability requirements, the safe distance between the UAV and the power transmission line and surrounding obstacles meets a threshold requirement, and the target deployment area is accessible.

[0107] Let the current position of the drone be... The initial location of the target delivery point is Then, if the following conditions are met:

[0108] Furthermore, when the drone's attitude fluctuations meet a preset attitude threshold, the drone can be considered to have entered the deployment window. This indicates the tolerance threshold for the placement location.

[0109] S52. Compensate and correct the target delivery location based on the UAV's flight status and environmental disturbances. Specifically, because the UAV is affected by changes in flight speed, attitude, and external disturbances during flight, there may be a deviation between the actual delivery point and the ideal delivery point. Therefore, it is necessary to compensate and correct the target delivery location. The compensation factors include the UAV's current speed, flight altitude, attitude angle, and environmental disturbances.

[0110] Let the current velocity vector of the UAV be... The environmental disturbance compensation vector is The corrected target deployment location It can be represented as:

[0111] in, and These represent the velocity compensation coefficient and the environmental disturbance compensation coefficient, respectively. By compensating and correcting the target deployment location, the consistency between the deployment landing point and the target area can be improved.

[0112] S53. Establish a deployment feasibility determination model. Specifically, before executing the deployment action, a comprehensive assessment is made of the current flight state, the local path optimization state, and the safety of the deployment area. When the UAV is in a locally stable flight state, and the current optimal local trajectory does not conflict with power transmission lines and surrounding obstacles within the prediction time domain, the deployment feasibility requirements are met at the current moment. Let the deployment feasibility determination function be , then it can be expressed as:

[0113] in, Indicators representing the degree of location matching Indicates the flight attitude stability index, Indicates local environmental safety indicators. , and For the corresponding weight coefficients, and satisfying:

[0114] when If the value exceeds the preset delivery threshold, the current time is considered to meet the delivery requirements.

[0115] S54. Control the UAV to adjust to the delivery posture and perform terminal precision alignment. Specifically, after meeting the delivery feasibility judgment conditions, control the UAV to perform fine-tuning flight according to the corrected target delivery position, so that the UAV gradually converges within the target delivery window and maintains a relatively stable flight posture. The fine-tuning process can be combined with the local path optimization results output in step S4 to avoid entering the dangerous area of ​​the power transmission line or getting close to the surrounding obstacles during the adjustment process. By performing closed-loop control on the error between the current position of the UAV and the corrected target delivery position, it is ensured that:

[0116] in, This represents the position error vector at the current moment. When... Less than the preset error threshold At that time, it was believed that the drone had completed the terminal precision alignment.

[0117] S55. Control the delivery mechanism to perform precise delivery. Specifically, when the UAV completes precise end-effector alignment and the delivery feasibility assessment meets the requirements, control the delivery mechanism to activate and release the target object. The delivery mechanism can be an electrically controlled mounting release mechanism, a mechanical locking release mechanism, or other execution mechanisms suitable for UAV payload delivery. After the delivery action is executed, record the UAV pose, target delivery location, and local environmental conditions at the time of delivery for subsequent delivery result evaluation.

[0118] S56. Confirm the status of the delivery result and output the accurate delivery operation result. Specifically, after the delivery is completed, the target delivery area is observed again based on visual sensor and UAV status feedback information to determine whether the target object has been successfully delivered to the preset area. If the delivery result meets the preset accuracy requirements, the delivery is considered successful; if the requirements are not met, a decision on subsequent flight adjustments or re-delivery is generated based on the current mission status. The delivery result is encapsulated into a delivery output set:

[0119] in, This indicates the corrected target deployment location. This indicates the result of the feasibility assessment for deployment. Indicates the status of the delivery result; when When the delivery is successful, it indicates that the delivery was successful. The time indicates that the delivery was unsuccessful. The delivery output set is used to characterize the accuracy of the UAV's delivery operation at the current moment.

[0120] In summary, this invention proposes a method for UAV path planning and precise delivery based on power transmission line vision guidance. It constructs a multi-source perception and state feedback system that integrates visual images, RTK positioning data, IMU attitude data, and UAV flight status information to achieve collaborative acquisition of the conductor target and the UAV's state. An improved visual processing method is used to complete stable identification of the power transmission line, target point generation, and visual perception confidence calculation. A multi-source fusion odometry system is constructed by combining RTK, IMU, and visual observations to provide continuous and stable position, velocity, and attitude states for path planning. Based on this, a global approach path constrained by the target point is generated, and local path smoothing optimization is driven by fusion state feedback. Furthermore, the target delivery position is compensated and corrected, and delivery feasibility judgment and end-point precision alignment control are performed, thereby achieving safe approach and high-precision delivery operations in complex power transmission line environments. This method has good engineering application value and promising prospects for widespread application.

[0121] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention. For example, the source of the construction of the conductor centerline topology model, the measurement reliability evaluation index, the covariance expansion factor, the gating threshold, and the sensor combination method can all be adjusted according to different voltage levels, conductor types, robot structures, and operating environments; the evaluation dimensions and scoring rules can be customized according to the needs of different industries, and the differentiated inference strategy can also be expanded and optimized according to actual application scenarios. These adjustments and optimizations all fall within the protection scope of the present invention.

Claims

1. A method for UAV path planning and precise delivery based on power transmission line vision guidance, characterized in that, Includes the following steps: S1. Construct a multi-source perception and operation execution system consisting of a visual sensor, an RTK positioning module, an IMU inertial measurement unit, a flight control module, and a delivery execution mechanism. Collect visual images, RTK positioning data, IMU attitude data, and UAV flight status data of the power transmission line operation area, and complete time synchronization and coordinate system one. S2. Perform adaptive edge detection and weighted line feature extraction on the visual image, and perform attitude compensation on the line feature extraction results in combination with the IMU attitude data to obtain the main direction of the conductor, key structural points of the conductor, visual relative observation of the conductor, and visual perception confidence. S3. Determine the reference position of the conductor based on the main direction of the conductor and the key structural points of the conductor, and apply a safety height offset along the safety offset direction at the reference position of the conductor to generate a point close to the target work point; Simultaneously, by fusing the RTK positioning data, the IMU attitude data, and the relative visual observations of the traverse line, a state estimation result including the UAV's position, velocity, attitude, and rate of change of attitude angles is obtained; S4. Generate a global approach path based on the approach target point, the state estimation result, the preset traverse safety distance constraint, and the traverse approach direction consistency constraint; The target tracking weight, safety distance weight, and trajectory smoothing weight are dynamically adjusted based on the visual perception confidence and the attitude angle change rate. Based on the adjusted weights, the global approach path is locally smoothed to obtain a smooth approach trajectory with continuous position, velocity, and acceleration. S5. When the UAV enters the preset delivery window along the smooth approach trajectory, and the visual perception confidence, attitude angle change rate, trajectory safety margin determined based on the smooth approach trajectory, and position error all meet the corresponding thresholds, the target delivery position is compensated and corrected according to the speed, attitude, and visual observation residual in the state estimation results to obtain the corrected target delivery position, and the delivery execution mechanism is controlled to complete the precise delivery of the power transmission line at the corrected target delivery position.

2. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The confidence level of visual perception It is obtained using the following formula: in, This represents the normalized value of the linear characteristic response intensity. Indicators representing the continuity of line characteristics This represents the stability index of the main direction of the conductor and the key structural points of the conductor across multiple consecutive frames. , and Here are the confidence weighting coefficients, and When the visual perception confidence level When the target tracking weight is lower than the preset confidence threshold, the target tracking weight is reduced, while the safety distance weight and the trajectory smoothing weight are increased. The attitude compensation includes: constructing an attitude rotation matrix based on the roll angle, pitch angle and yaw angle output by the IMU attitude data, transforming the characteristic direction of the guide line in the image coordinate system to the body coordinate system, and using the attitude change amount in consecutive frames to correct the main direction of the guide line, so as to reduce the impact of UAV body jitter and viewpoint changes on the guide line recognition results.

3. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The approach to the target point Generate according to the following formula: in, Indicates the reference position of the conductor. Indicates the safety height offset. Represents the unit vector in the vertical direction of the world coordinate system. This indicates the lateral safety offset. The unit vector representing the safety offset direction perpendicular to the main direction of the conductor; the approach target point is located in the safety space above or to the side of the conductor area.

4. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The state estimation result is obtained by fusing the RTK positioning data, the IMU attitude data, and the visual relative observation of the traverse line; wherein, the RTK positioning data is used to constrain the global position drift of the UAV, the IMU attitude data is used for high-frequency state propagation, and the visual relative observation of the traverse line is used to correct the local position and attitude deviation of the UAV relative to the target area of ​​the traverse line. The path cost function for generating the global proximity path It includes path length cost, safety distance cost, and proximity direction consistency cost, and its expression is: in, This represents the path length cost. This represents the safety distance cost of a path node relative to the transmission line safety buffer zone. This represents the proximity direction consistency cost between the path terminal velocity direction and the main direction of the conductor. , and The corresponding weighting coefficients are used; the UAV enters the preset deployment window along a direction that is consistent with or approximately consistent with the main direction of the guide wire by means of the proximity direction consistency cost constraint.

5. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The target tracking weight, the safety distance weight, and the trajectory smoothing weight are based on the visual perception confidence level. and the rate of change of attitude angle Dynamic adjustment: When rise and When the target tracking weight is reduced, the target tracking weight is increased; when Reduce or When the distance increases, the safety distance weight and the trajectory smoothing weight are increased.

6. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The local smoothing optimization uses fifth-order spline interpolation to fit local path segments in the global approach path, making the generated smooth approach trajectory continuous in terms of position, velocity, and acceleration. Infeasible trajectories are eliminated based on the UAV's maximum speed, maximum acceleration, maximum climb angle, minimum turning radius, and the safety distance constraint of the guide wire.

7. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The triggering conditions for the preset deployment window include: the position error between the current position of the UAV and the target deployment position before compensation is less than the position tolerance threshold; the attitude angle change rate is less than the attitude stability threshold; the minimum safe distance between the smooth approach trajectory and the power transmission line and surrounding obstacles in the prediction time domain is greater than the safe distance threshold to meet the trajectory safety margin; and the visual perception confidence is greater than the confidence threshold.

8. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, The corrected target delivery location The compensation and correction shall be performed according to the following formula: in, This indicates the target deployment location before compensation. This represents the velocity vector in the state estimation result. Indicates the delay time for deployment execution. Represents the roll angle from the state estimation results. Pitch angle and heading angle The constructed attitude rotation matrix, This represents the unit vector indicating the direction of delivery. This represents the visual observation residual. , and These represent the velocity compensation coefficient, attitude compensation coefficient, and visual residual compensation coefficient, respectively.

9. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 1, characterized in that, Step S1 specifically includes: Construct a UAV multi-source perception and state feedback system consisting of the visual sensor, the RTK positioning module, the IMU inertial measurement unit, and the flight control module; The modules within the multi-source sensing and state feedback system are synchronized in time and calibrated in space to unify the acquired data into the same coordinate system. The drone is controlled to fly to the target power transmission line operation area and continuously scans and collects visual images during the flight. The RTK positioning data, IMU attitude data, and flight status data of the UAV are acquired simultaneously. The data collected at the same or adjacent times are time-aligned and correlated to form a multi-source sensing dataset under the current operating state for use in subsequent steps.

10. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 9, characterized in that, Step S2 specifically includes: The visual images in the multi-source perception dataset are preprocessed, and candidate edge features of the transmission line are extracted through adaptive edge detection. Based on the candidate edge features, weighted line feature extraction is performed to obtain the transmission line route feature parameters; IMU attitude compensation and sliding filtering are performed by combining the IMU attitude data in the multi-source sensing dataset to correct the impact of UAV flight disturbances on the line feature extraction results. Based on the attitude compensation results, the characteristic parameters of the power transmission line and the camera imaging model, the target point is initially calculated to obtain the visual relative observation of the conductor, the main direction of the conductor and the key structural points of the conductor. The visual perception confidence is calculated based on the line feature response intensity, line feature continuity, and multi-frame stability. The main direction of the traverse, the key structural points of the traverse, the visual relative observation of the traverse, and the visual perception confidence are encapsulated into path planning perception information and output to the next stage.

11. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 10, characterized in that, Step S3 specifically includes: The reference position of the traverse is determined based on the key structural points of the traverse in the path planning perception information, and the tangential direction of the local coordinate system is determined based on the main direction of the traverse. Apply a safety height offset and a lateral safety offset at the guide reference position to generate the approach target point; Construct a fusion state vector that includes the UAV's position, velocity, attitude, gyroscope bias, and accelerometer bias; The state estimation result at the current moment is obtained by using the IMU attitude data for high-frequency state propagation, using the RTK positioning data to constrain the global position drift, and using the visual relative observation of the guide line to correct the local pose deviation of the UAV relative to the guide line target area. The approach target point, the state estimation result, the main direction of the traverse, and the visual perception confidence are uniformly encapsulated and output.

12. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 11, characterized in that, Step S4 specifically includes: Based on the current location of the UAV, the approach target point, the safety buffer zone of the guide wire, and the surrounding obstacles, a global approach path search space is constructed. Under the conditions of satisfying all dynamic constraints and the safety distance constraint of the conductor, the global proximity path is generated by searching using the path cost function; The target tracking weight, the safety distance weight, and the trajectory smoothing weight are dynamically adjusted based on the visual perception confidence level and the attitude angle change rate in the state estimation result. The local reference path segment near the current time is extracted from the global approach path, and the local smoothing optimization is performed using the fifth-order spline interpolation to generate the smooth approach trajectory. The safety of the smooth approach trajectory is determined, and the conflict-free trajectory with the highest comprehensive evaluation value is output as the optimal local approach trajectory.

13. The method for UAV path planning and precise delivery based on power transmission line vision guidance according to claim 12, characterized in that, Step S5 specifically includes: Based on the state estimation results and the local reference path segment, it is determined whether the UAV has entered the preset deployment window; When all the triggering conditions related to the visual perception confidence level, the attitude angle change rate, and the trajectory safety margin are met, it is determined that the current moment meets the deployment requirements. The corrected target deployment position is calculated and obtained by using the current velocity, attitude angle change and visual observation residuals from the state feedback. The drone is controlled to perform end-point precision alignment based on the corrected target deployment location. When the positional error of the end-point precision alignment is less than the preset precision alignment error threshold, the delivery execution mechanism is controlled to release the target object, thus completing the precise delivery operation of the power transmission line.

Images