Obstacle avoidance control method and system for power transmission line maintenance robot

Abstract

The invention discloses a power transmission line maintenance robot obstacle avoidance control method and system, and the method comprises the steps: constructing a multi-mode sensing network through integrating a laser radar, an infrared sensor, a millimeter wave radar and a visual camera, and collecting 0-20 m three-dimensional operation environment data in real time; dynamically dividing high, medium and low risk areas according to the electric field intensity and the obstacle density, and setting response grades; a dynamic weighted fusion algorithm is adopted to generate far, medium and close distance sensing data streams; implementing a hierarchical control strategy according to the risk level, wherein the hierarchical control strategy comprises path planning, fine adjustment avoidance and three-dimensional bypass braking; the system has the functions of adaptive sensor weight adjustment, risk-driven resource arbitration mechanism and fault-tolerant control and feedback learning, and can improve the obstacle avoidance reliability and intelligent level in a complex power grid environment. The scheme has the technical advantages of high sensing redundancy, high response precision and high system stability.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission line maintenance robots, and in particular to an obstacle avoidance control method and system for a power transmission line maintenance robot. Background Art

[0002] Transmission lines are a vital component of the power system, and their safe operation is directly linked to the stability and reliability of the entire power grid. Traditional transmission line maintenance relies primarily on manual inspections, which, affected by factors such as terrain, climate, and high-altitude operations, result in low efficiency, high risk, and high labor costs. In recent years, with the advancement of robotics technology, transmission line inspection robots have begun to be used in high-voltage line maintenance, replacing manual labor in some high-risk tasks.

[0003] Although existing power transmission line maintenance robots have certain perception and obstacle avoidance capabilities, they still face the following technical bottlenecks:

[0004] Single perception and insufficient redundancy: Currently, most systems only use lidar or a single visual sensor, which cannot achieve cross-verification of multi-source information and makes it difficult to achieve stable perception in complex scenarios such as strong electromagnetic interference and severe weather.

[0005] Insensitive response to risk scenarios: Most systems only respond to obstacles according to fixed thresholds and lack a hierarchical control strategy based on dynamic risk assessment, resulting in delayed obstacle avoidance responses or misjudgment of risk levels.

[0006] Static fusion processing method: Sensor data fusion mostly adopts fixed weight and fixed priority schemes, without considering the dynamic adjustment of sensor confidence under changing environmental conditions, resulting in insufficient processing efficiency and accuracy.

[0007] The control system lacks closed-loop feedback and fault tolerance mechanisms: When a sensor fails or misreports, existing systems simply skip or terminate execution, lacking a fusion compensation mechanism, which affects the stable operation of the system.

[0008] Therefore, there is an urgent need for an obstacle avoidance control system and method for a transmission line maintenance robot that can integrate multimodal perception, adaptive decision-making, real-time feedback and fault tolerance to improve the robot's autonomous operation capability and reliability in complex power grid environments. Summary of the Invention

[0009] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of this application to avoid obscuring the purpose of this section, the abstract and the title of the invention, and such simplifications or omissions should not be used to limit the scope of the present invention.

[0010] Therefore, in order to solve the above technical problems, the present invention provides the following technical solutions: a method for controlling an obstacle avoidance system of a power transmission line maintenance robot, the method comprising the following steps:

[0011] (a) Real-time data collection of the robot's surrounding environment is achieved through a multimodal sensor system. The sensors, including lidar, infrared sensors, millimeter-wave radar, and visual cameras, are deployed on the robot's head, wings, and chassis, covering a three-dimensional space of 0 to 20 meters around the wire.

[0012] (b) Based on the electric field strength, obstacle density and safety regulations, the working environment is divided into high-risk area, medium-risk area and low-risk area, and the corresponding safety distance threshold and response level are set respectively;

[0013] (c) generating a fused data stream using a dynamic weighted filtering algorithm, wherein the weights are dynamically adjusted based on the risk level, sensor confidence, and environmental interference factors;

[0014] (d) Implementing hierarchical obstacle avoidance strategies in different areas, where:

[0015] Low-risk areas: Millimeter-wave radar and vision fusion to perform dynamic obstacle recognition and path planning;

[0016] Medium-risk area: LiDAR and infrared sensors cross-verify obstacle locations and implement path fine-tuning;

[0017] High-risk areas: Force sensors and infrared sensors work together to trigger reverse braking and three-dimensional avoidance;

[0018] (e) Differentiated control strategies are set based on risk levels. In high-risk areas, multi-level redundancy checks are enabled and speed reduction is forced. In low-risk areas, a rapid response mechanism is enabled, and control parameters and judgment thresholds are dynamically adjusted through a feedback mechanism.

[0019] As a preferred solution of the obstacle avoidance control method for the power transmission line maintenance robot of the present invention, the multimodal sensor has the following configuration:

[0020] The laser radar adopts linear array scanning mode and is installed on the head of the robot. The axial resolution is no more than 0.1° and is used for close-range high-precision positioning.

[0021] The infrared sensor is a dual-band structure and is located at the end of the operating arm. It is suitable for nighttime low-light detection and insulator overheating identification.

[0022] The millimeter-wave radar is in 77GHz FMCW mode and supports multi-target trajectory prediction;

[0023] The visual camera is equipped with a polarizing filter and adaptive dimming to resist electromagnetic interference.

[0024] As a preferred solution of the obstacle avoidance control method for the power transmission line maintenance robot of the present invention, the dynamic weighted filtering algorithm includes:

[0025] In high-risk areas, the weight of lidar is 0.6-0.8, and the weight of infrared sensor is 0.2-0.4;

[0026] In low-risk areas, the weight of millimeter-wave radar is 0.7 to 0.9, and the weight of visual imagery is no less than 0.1;

[0027] If a high voltage electric field or rainfall environment is detected, the system will automatically increase the infrared sensor weight to no less than 0.6.

[0028] As a preferred solution of the obstacle avoidance control method for the power transmission line maintenance robot of the present invention, the hierarchical obstacle avoidance strategy includes:

[0029] In high-risk areas: LiDAR and force sensors are checked for consistency, infrared sensor data is compared with the environmental map, and wire model deviation analysis is performed in conjunction with visual images to achieve three-level redundant obstacle avoidance judgment.

[0030] In low-risk areas: If the millimeter-wave radar detects an obstacle at a distance less than 120% of its set threshold, path deviation is immediately triggered. If visual recognition simultaneously detects a static obstacle, lateral avoidance is prioritized.

[0031] As a preferred solution of the obstacle avoidance control method for the power transmission line maintenance robot of the present invention, when a sensor anomaly occurs, the control system automatically triggers the fault tolerance mechanism according to the current risk level:

[0032] In high-risk areas, the system reconstructs missing perception data through interpolation compensation and forces speed reduction control;

[0033] In low-risk areas, the control logic automatically switches to a single-sensor dominant decision-making mode and extends the path decision cycle to maintain stability;

[0034] All fault records are synchronously written into the black box module, and health assessment reports based on historical data are generated periodically.

[0035] As a preferred embodiment of the obstacle avoidance control method for the power transmission line maintenance robot of the present invention, it includes an adaptive control parameter adjustment step for dynamically optimizing the obstacle avoidance strategy based on real-time environmental information and control feedback, wherein:

[0036] When the wind speed exceeds level 8, the system automatically increases the safety distance threshold of high-risk areas by 20%;

[0037] When the ambient temperature is lower than –20°C, the infrared sampling frequency is increased to 15Hz;

[0038] When the obstacle avoidance success rate is continuously lower than 90% or sensor abnormalities occur frequently, the system will proactively prompt for manual intervention.

[0039] As a preferred solution of the obstacle avoidance control method for the power transmission line maintenance robot of the present invention, the system includes a closed-loop feedback and learning mechanism:

[0040] Build a dynamic feedback model based on obstacle avoidance results and sensor health status to optimize sensor weights and path strategies;

[0041] When the obstacle avoidance success rate is continuously less than 90%, manual intervention will be prompted;

[0042] Generate self-diagnosis reports and push them to the operation and maintenance platform.

[0043] The present invention also provides the above-mentioned power transmission line maintenance robot obstacle avoidance control system, comprising

[0044] Multimodal perception module: Integrates lidar, infrared sensor, millimeter-wave radar and visual camera, and is installed on the robot body according to geometric layout rules;

[0045] Risk area division module: dynamically divide high / medium / low risk areas and output control level parameters;

[0046] Fusion processing module: generates far / mid / near data streams based on sensor confidence weighting;

[0047] Hierarchical control module: performs path planning, fine-tuning and braking according to risk level;

[0048] Priority arbitration module: Dynamically allocates computing resources based on risk levels to ensure that the computing priority of high-risk areas is not less than 70%.

[0049] As a preferred solution of the obstacle avoidance control system of the power transmission line maintenance robot of the present invention, the close-range emergency braking submodule includes:

[0050] Reverse thrust mechanism triggering conditions: contact force > 5N and infrared temperature rise > 10°C;

[0051] When the insulator disc diameter deviation is greater than 5%, a detour path with a Z-axis offset greater than 30 cm is generated by visual recognition;

[0052] Generates an optimal braking distance curve based on the obstacle closing rate.

[0053] As a preferred solution of the obstacle avoidance control system of the power transmission line maintenance robot of the present invention, it also includes the following functional modules:

[0054] Anti-interference communication module: uses TTP / C bus to transmit sensor data, with a bit error rate of no more than 10 -9 ;

[0055] Fault recovery module: When the lidar fails, the system automatically enables visual and infrared combined detection, shields the millimeter-wave radar in strong electromagnetic environments, and switches to optical recognition;

[0056] Energy consumption optimization module: When the control system runs continuously in the low-risk zone for more than 60 seconds, it can automatically shut down some redundant sensors to reduce the power consumption of the entire machine to below 120W.

[0057] Beneficial effects of the present invention:

[0058] 1. This invention integrates multiple sensors, including lidar, infrared sensors, millimeter-wave radar, and visual cameras, to build a multimodal environmental perception system covering a spatial range of 0 to 20 meters. These sensors complement and collaborate under varying conditions of distance, weather, and electromagnetic interference, enabling robust perception of targets such as birds, wires, drones, and insulators, addressing the perception failures of existing systems.

[0059] 2. This solution dynamically divides high-, medium-, and low-risk areas into different categories based on electric field strength, obstacle density, and operational safety regulations. Different control strategies are implemented based on the regional level, including path fine-tuning, emergency braking, and three-dimensional avoidance. This significantly improves obstacle avoidance accuracy and control response flexibility, avoiding unnecessary avoidance or control lag caused by unified control logic.

[0060] 3. Compared to existing fixed-weight fusion methods, this invention proposes a fusion strategy that dynamically adjusts weighting coefficients based on sensor confidence, environmental conditions, and historical accuracy. For example, in electromagnetic interference scenarios, it automatically reduces visual weight and increases infrared weight, while in rainy and foggy environments, it introduces millimeter-wave-based judgment, achieving dual guarantees for perception accuracy and response speed.

[0061] 4. When a sensor is abnormal or fails completely, the system can perform interpolation compensation based on historical models, or call other sensor data for redundant decision-making, and at the same time activate the deceleration mechanism to ensure safe operation; through the black box mechanism, the system operating status and perception of abnormal information are recorded, forming an autonomous learning and health assessment mechanism, which significantly enhances system reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0063] Figure 1 It is the overall workflow diagram of the present invention. DETAILED DESCRIPTION

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0065] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0066] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0067] Reference Figure 1 This embodiment of the present invention targets 220kV transmission line maintenance operations in mountainous areas. The on-site environment features pine forests, rugged terrain, complex obstacle types (such as branches, bird nests, foreign objects, and broken wires), and unpredictable weather. The maintenance robot, mounted on a maintenance cableway support, moves along the conductors to perform inspections and maintenance tasks. The system must possess highly robust obstacle avoidance capabilities and support precise operations in high-risk areas.

[0068] 1. The implementation process is as follows:

[0069] Step (a): Environmental data collection and synchronization;

[0070] Start multi-sensor collaborative data collection 10 minutes before the robot starts operating to build a real-time environment perception system:

[0071] 1. LiDAR

[0072] Used to scan the area 0 to 5 meters ahead, generate a 3D point cloud map, and accurately identify the spatial position and material of obstacles (such as metal residues and branches);

[0073] The collected data includes:

[0074] Distance value: the actual distance from each scanning point to the object (unit: meter);

[0075] Point cloud angle coordinates: horizontal scanning angle, resolution 0.1°;

[0076] Reflection intensity: The strength of the echo signal is used to determine the material of the object (distinguishing between metal and non-metal materials);

[0077] Example:

[0078] The reflection intensity of scanning point A is 70% at a distance of 2.8m, which is judged to be the residue of a metal tool;

[0079] The reflection intensity of point B is 15%, and the distance is 1.3m. It may be a leaf or a bird's nest, and further visual identification is required.

[0080] 2. Infrared sensor

[0081] Collect dual-band thermal imaging data to identify abnormal heating of electrical equipment (such as heating of wire joints and arc faults), supporting nighttime or low-visibility scenarios;

[0082] The types of data collected include:

[0083] Thermal Image: Temperature distribution map in pixels (unit: °C);

[0084] Dual-band imaging (e.g., 8-14 μm and 3-5 μm);

[0085] Example:

[0086] The temperature in the cable joint area rises to 75°C, compared to the normal conductor temperature of 40°C, marking it as a warning area;

[0087] The infrared image shows a high-temperature foreign object 1.2 meters ahead, and combined with the lidar, it is determined to be a flying bird.

[0088] Millimeter-wave radar (77GHz FMCW)

[0089] Detect high-speed dynamic targets (such as birds and drones) at long distances (5 to 20 meters) and predict their trajectory;

[0090] The types of data collected include:

[0091] Relative speed of the target object (m / s);

[0092] The distance between the object and the radar (Range);

[0093] Multipath echo intensity (for determining the contours of dynamic objects);

[0094] Example:

[0095] The flying object was detected at a distance of 15.7m and a speed of approximately 6.3m / s (approaching), triggering a medium-risk warning.

[0096] Vision Camera

[0097] Identify obstacle types (such as bird nests and insulator pollution) using HDR images and deep learning models;

[0098] The data types collected include: target edge contour map, texture classification information;

[0099] Example:

[0100] The shape of a suspected bird's nest was detected in the visual image. Combined with the overlapping area of ​​the lidar point cloud, its position was determined to be 2.3m in front and 15° to the right.

[0101] 5. Data storage: All sensor data is stored in the local cache in real time for subsequent fusion processing.

[0102] Step (b): Risk area division;

[0103] The main purpose of this step is to:

[0104] Achieve "graded response" for obstacle avoidance control: allocate different control cycles and safety redundancy based on risk levels, improving the system's adaptability and fault tolerance to different scenarios;

[0105] Improve the ability to respond to the "unstructured environment" around transmission lines: such as tree canopy, flying birds, floating foreign objects, electric field interference and other multi-source heterogeneous risk sources;

[0106] Ensure the electromagnetic safety and flight stability of the maintenance robot: avoid entering high-field strength areas that may cause control failure or structural damage.

[0107] Based on electric field strength, obstacle density and safety regulations, three risk areas are divided:

[0108] 1. High-risk area: 0 to 3 meters;

[0109] Judgment conditions: There are dynamic obstacles within the above range or the electric field strength is greater than 4kV / m;

[0110] Basis for setting this limit: The "State Grid Corporation of China Live Line Working Specifications" and the "IEC 61472" standard stipulate the induced electric field limit for humans and equipment working within 3 meters of high-voltage transmission conductors (>4kV / m may cause electric shock hazards). Empirical data shows that under normal operating conditions on a standard 220kV line, the electric field at 1 meter is approximately 67kV / m, dropping to approximately 34kV / m at 3 meters. To account for factors such as flying birds and wind-induced sway errors, 3 meters is set as the boundary of the mandatory intervention zone.

[0111] 2. Medium risk area: 3 to 10 meters;

[0112] Judgment conditions: Dense static objects (such as tree canopies) within the above range are affected by strong light or electromagnetic interference and visual recognition is affected;

[0113] Basis for setting: The 3-10 meter range is the "sensing warning zone" near charged conductors; within this area, radar and LiDAR have higher detection stability;

[0114] Low-risk area: 10 to 20 meters;

[0115] Judgment conditions: The environment within the above range is open and obstacles are sparse;

[0116] Setting basis: This area is at the edge of the visual field of power transmission poles or high-altitude equipment, and the radar perception data is stable; the electric field strength has limited interference with communication and control signals; the millimeter-wave radar detection range (identifying flying targets within 20 meters), LiDAR effective point cloud density distribution and other parameters set the threshold of the low-risk zone (i.e., 10 to 20 meters).

[0117] These intervals are derived from field tests and power system conductor protection standards, taking into account both the risk of false touch and the need for control redundancy;

[0118] It should be noted that the above-mentioned risk areas and control periods are divided into the following: the present invention is designed in combination with the 220kV transmission line environment, typical sensor parameters and current operating specifications. Other embodiments may appropriately adjust the partition boundaries and corresponding response periods according to the actual transmission level and sensor accuracy, and still fall within the scope of protection of the technical solution of the present invention;

[0119] Step (c): sensor fusion and dynamic weighting algorithm;

[0120] 1. Standardization of sensor perception output;

[0121] In order to solve the problem of different physical units and dimensions of data collected by various sensors, a standardized process is used to obtain the unitized perception intensity score S i (t), which is calculated as follows:

[0122] 1.1、Infrared thermal imaging (IR) output;

[0123] The current detection target temperature T obj (t) is normalized to the interval [0,1]:

[0124]

[0125] Among them, T min 、T max The actual minimum and maximum values ​​of the scene temperature distribution are taken respectively (e.g. 20℃~80℃);

[0126] 1.2. Laser radar distance detection (LiDAR);

[0127] The distance score formula is as follows:

[0128] Among them, d(t) is the distance between the current obstacle and the aircraft, d max =30m is the upper limit of the equipment detection capability;

[0129] 1.3. Visual recognition sensor confidence (Vision): Directly use the target sensor confidence R output by deep learning i .

[0130] 1.4 Millimeter-wave radar echo response (mmWave)

[0131] The score is constructed using the normalized signal-to-noise ratio (SNR) and echo stability, which can be defined as:

[0132] 2. Basic weight distribution;

[0133] Setting basis: Determined by expert experience rules based on sensor performance parameters (such as accuracy and anti-interference ability), historical measured data and power industry standards (such as DL / T 639).

[0134] Regional risk level Infrared (IR) LiDAR mmWave Vision High risk (<3m) 0.9 0.6 0.4 0.3 Medium risk (3-10m) 0.5 0.8 0.95 0.6 Low risk (10-20m) 0.3 0.9 0.6 0.85

[0135] The classification is based on the following principles:

[0136] Infrared reacts quickly to localized heating in high-risk areas (such as near high-voltage equipment);

[0137] LiDAR has higher mapping accuracy at medium and long distances;

[0138] mmWave is sensitive to small targets (birds, drones);

[0139] Vision has more advantages in medium and long-distance contour recognition.

[0140] 3. Dynamic weight update;

[0141] At each moment t, the system outputs the intensity S according to the perception i (t) and the basic sensor confidence R of the channel i , calculate its weighting factor:

[0142] Where: i represents the current channel, j is the index of all channels, used for normalization; ∑ j ω i (t) = 1, which meets the normalization requirement.

[0143] Dynamic adjustment mechanism:

[0144] Normal working conditions: The weight is directly calculated by the above formula.

[0145] Abnormal working conditions (such as sensor failure): trigger expert rules to force weight adjustment.

[0146] Rainy day scene example:

[0147] Initial parameters (medium risk area):

[0148] Original weight calculation (without rain interference):

[0149] Assuming that the current location is in the medium risk area (3 to 10 meters), the basic scoring coefficient (confidence) of each sensor is R i They are:

[0150] Infrared (IR): 0.5

[0151] LiDAR: 0.8

[0152] Millimeter wave (mmWave): 0.95

[0153] Vision: 0.6

[0154] Assume that each sensor’s perception strength score S i (t):

[0155] Infrared: 0.8 (temperature detection effective)

[0156] LiDAR: 0.7 (point cloud quality is normal)

[0157] Millimeter wave: 0.9 (dynamic target stabilization)

[0158] Vision: 0.6 (clear image)

[0159] Then, the initial weight calculation steps are as follows:

[0160] Denominator = (0.5 × 0.8) + (0.8 × 0.7) + (0.95 × 0.9) + (0.6 × 0.6) = 2.175;

[0161] ω IR =(0.5×0.8) / 2.175≈0.184;

[0162] ω LiDAR =(0.8×0.7) / 2.175≈0.258;

[0163] ω mmWave =(0.95×0.9) / 2.175≈0.393;

[0164] ω Vision =(0.6×0.6) / 2.175≈0.165;

[0165] Rain trigger adjustment:

[0166] When the system detects visual failure (e.g., rainfall causing the image signal-to-noise ratio to be less than 0.4), it performs the following operations:

[0167] Forced adjustment of infrared basic rating coefficient: R IR Increased from 0.5 to 0.7 (default rule);

[0168] Visual perception score is set to zero: S Vision (t) = 0, which is the failure state.

[0169] Recalculate R i :

[0170] Infrared: R IR =0.7, S IR (t) = 0.8;

[0171] LiDAR: R LiDAR =0.8, S LiDAR (t) = 0.7;

[0172] Millimeter wave: R mmWave =0.95, S mmWave (t) = 0.9;

[0173] Vision: Vision = 0.6, S LiDAR (t) = 0;

[0174] Recalculate weights:

[0175] Denominator = (0.7 × 0.8) + (0.8 × 0.7) + (0.95 × 0.9) + (0.6 × 0) = 1.975;

[0176] ω IR =(0.7×0.8) / 1.975≈0.283;

[0177] ω LiDAR =(0.8×0.7) / 1.975≈0.283;

[0178] ω mmWave =(0.95×0.9) / 1.975≈0.433;

[0179] ω Vision =0;

[0180] Expert rules intervene:

[0181] If the calculated infrared weight is still lower than 0.6, the system will forcibly set the infrared weight lower limit to 0.6 according to the preset rules, and the remaining weight will be distributed proportionally:

[0182] ω IR =0.6

[0183] The remaining weight is combined = 1-0.6 = 0.4;

[0184] ωmmWave =0.4×(0.433 / (0.283+0.433))≈0.4×0.605≈0.242;

[0185] ω LiDAR =0.4×(0.283 / (0.283+0.433))≈0.4×0.395≈0.1580.4×

[0186] (0.283 / (0.283+0.433))≈0.4×0.395≈0.158;

[0187] Final weight:

[0188] Infrared: 0.6

[0189] Millimeter wave: 0.242

[0190] LiDAR: 0.158

[0191] Vision: 0

[0192] Step (d): Hierarchical obstacle avoidance strategy execution;

[0193] 1. Take the example of a robot detecting a high-speed flying object (bird) 7 meters away:

[0194] In the long range (7 meters), millimeter-wave radar and visual images are combined with a path planning module to estimate its flight trajectory;

[0195] When approaching within 3 meters and entering the middle section, the LiDAR and infrared sensors work together to fine-tune the current trajectory;

[0196] If the vehicle suddenly invades within 1 meter, the close section strategy will be triggered, and the force sensor and infrared will be linked to perform reverse braking + Z-axis floating action to ensure rapid obstacle avoidance.

[0197] 2. Fault tolerance:

[0198] LiDAR failure: Switch to vision + infrared combination, and positioning accuracy drops to ±3cm.

[0199] Infrared disabled: Enable LiDAR+millimeter wave and disable temperature detection.

[0200] Step (e): response strategy differentiation;

[0201] 1. High-risk areas

[0202] The control cycle is compressed to 50ms, the speed is reduced to 0.2m / s, and dual-channel communication is enabled to ensure real-time performance.

[0203] 2. Low-risk areas

[0204] The response cycle is extended to 200ms, energy consumption is reduced, and paths are dynamically optimized.

[0205] Step (f): dynamic system parameter adjustment;

[0206] 1. Environmental adaptation:

[0207] Wind speed > Level 8: The safety distance in high-risk areas increases from 1.5 meters to 1.8 meters.

[0208] Temperature < -20°C: The infrared sampling rate is increased to 15Hz to reduce image smear.

[0209] 2. Self-learning mechanism:

[0210] When the obstacle avoidance success rate is less than 90%, optimization suggestions or manual takeover requests are triggered.

[0211] This solution builds a logical closed loop at multiple levels, including dynamic risk identification, multi-source perception fusion, adaptive control, and system fault-tolerant collaboration. It proposes a highly self-consistent, robust, and practical engineering application-rich intelligent obstacle avoidance solution for transmission line maintenance. It not only effectively solves the problems of incomplete perception, extensive control, and system fragility of existing transmission line maintenance robots, but also proposes a number of technological innovations in fusion strategy, risk zoning logic, and control decision-making. It has obvious practical value and industrial transformation potential.

[0212] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for controlling an obstacle avoidance system for a power transmission line maintenance robot, characterized by: The method comprises the following steps: (a) Real-time data collection of the robot's surrounding environment is achieved through a multimodal sensor system. The sensors, including lidar, infrared sensors, millimeter-wave radar, and visual cameras, are deployed on the robot's head, wings, and chassis, covering a three-dimensional space of 0 to 20 meters around the wire. (b) Based on the electric field strength, obstacle density and safety regulations, the working environment is divided into high-risk area, medium-risk area and low-risk area, and the corresponding safety distance threshold and response level are set respectively; (c) generating a fused data stream using a dynamic weighted filtering algorithm, wherein the weights are dynamically adjusted based on the risk level, sensor confidence, and environmental interference factors; (d) Implementing hierarchical obstacle avoidance strategies in different areas, where: Low-risk areas: Millimeter-wave radar and vision fusion to perform dynamic obstacle recognition and path planning; Medium-risk area: LiDAR and infrared sensors cross-verify obstacle locations and implement path fine-tuning; High-risk areas: Force sensors and infrared sensors work together to trigger reverse braking and three-dimensional avoidance; (e) Differentiated control strategies are set based on risk levels. In high-risk areas, multi-level redundancy checks are enabled and speed reduction is forced. In low-risk areas, a rapid response mechanism is enabled, and control parameters and judgment thresholds are dynamically adjusted through a feedback mechanism.

2. The obstacle avoidance control method for a power transmission line maintenance robot according to claim 1, wherein: The multimodal sensor has the following configurations: The laser radar adopts linear array scanning mode and is installed on the head of the robot. The axial resolution is no more than 0.1° and is used for close-range high-precision positioning. The infrared sensor is a dual-band structure and is located at the end of the operating arm. It is suitable for nighttime low-light detection and insulator overheating identification. The millimeter-wave radar is in 77GHz FMCW mode and supports multi-target trajectory prediction; The visual camera is equipped with a polarizing filter and adaptive dimming to resist electromagnetic interference.

3. The obstacle avoidance control method for a power transmission line maintenance robot according to claim 2, wherein: Dynamic weighted filtering algorithms include: In high-risk areas, the weight of lidar is 0.6-0.8, and the weight of infrared sensor is 0.2-0.4; In low-risk areas, the weight of millimeter-wave radar is 0.7 to 0.9, and the weight of visual imagery is no less than 0.1; If a high voltage electric field or rainfall environment is detected, the system will automatically increase the infrared sensor weight to no less than 0.

6.

4. The obstacle avoidance control method for a power transmission line maintenance robot according to claim 3, wherein: The hierarchical obstacle avoidance strategy includes: In high-risk areas: LiDAR and force sensors are checked for consistency, infrared sensor data is compared with the environmental map, and wire model deviation analysis is performed in conjunction with visual images to achieve three-level redundant obstacle avoidance judgment. In low-risk areas: If the millimeter-wave radar detects an obstacle at a distance less than 120% of its set threshold, path deviation is immediately triggered. If visual recognition simultaneously detects a static obstacle, lateral avoidance is prioritized.

5. The obstacle avoidance control method for a power transmission line maintenance robot according to claim 4, wherein: When a sensor anomaly occurs, the control system automatically triggers the fault tolerance mechanism based on the current risk level: In high-risk areas, the system reconstructs missing perception data through interpolation compensation and forces speed reduction control; In low-risk areas, the control logic automatically switches to a single-sensor dominant decision-making mode and extends the path decision cycle to maintain stability; All fault records are synchronously written into the black box module, and health assessment reports based on historical data are generated periodically.

6. The obstacle avoidance control method for a power transmission line maintenance robot according to claim 5, wherein: It includes an adaptive control parameter adjustment step for dynamically optimizing the obstacle avoidance strategy based on real-time environmental information and control feedback, where: When the wind speed exceeds level 8, the system automatically increases the safety distance threshold of high-risk areas by 20%; When the ambient temperature is lower than –20°C, the infrared sampling frequency is increased to 15Hz; When the obstacle avoidance success rate is continuously lower than 90% or sensor abnormalities occur frequently, the system will proactively prompt for manual intervention.

7. The obstacle avoidance control method for a power transmission line maintenance robot according to claim 6, wherein: The system includes a closed-loop feedback and learning mechanism: Build a dynamic feedback model based on obstacle avoidance results and sensor health status to optimize sensor weights and path strategies; When the obstacle avoidance success rate is continuously less than 90%, manual intervention will be prompted; Generate self-diagnosis reports and push them to the operation and maintenance platform.

8. The power transmission line maintenance robot obstacle avoidance control system according to any one of claims 1 to 7, characterized in that: include Multimodal perception module: Integrates lidar, infrared sensor, millimeter-wave radar and visual camera, and is installed on the robot body according to geometric layout rules; Risk area division module: dynamically divide high / medium / low risk areas and output control level parameters; Fusion processing module: generates far / mid / near data streams based on sensor confidence weighting; Hierarchical control module: performs path planning, fine-tuning and braking according to risk level; Priority arbitration module: Dynamically allocates computing resources based on risk levels to ensure that the computing priority of high-risk areas is not less than 70%.

9. The obstacle avoidance control system for a power transmission line maintenance robot according to claim 8, characterized in that: The close-range emergency braking submodule includes: Reverse thrust mechanism triggering conditions: contact force > 5N and infrared temperature rise > 10°C; When the insulator disc diameter deviation is greater than 5%, a detour path with a Z-axis offset greater than 30 cm is generated by visual recognition; Generates an optimal braking distance curve based on the obstacle closing rate.

10. The power transmission line maintenance robot obstacle avoidance control system according to claim 9, characterized in that: It also includes the following functional modules: Anti-interference communication module: uses TTP / C bus to transmit sensor data, with a bit error rate of no more than 10 -9 ; Fault recovery module: When the lidar fails, the system automatically enables visual and infrared combined detection, shields the millimeter-wave radar in strong electromagnetic environments, and switches to optical recognition; Energy consumption optimization module: When the control system runs continuously in the low-risk zone for more than 60 seconds, it can automatically shut down some redundant sensors to reduce the power consumption of the entire machine to below 120W.

Images