Obstacle avoidance methods, devices, storage media, and computer equipment for unmanned aerial vehicles (UAVs) in substations

Abstract

The obstacle avoidance method for unmanned aerial vehicles (UAVs) within a substation provided in this application includes: when a UAV performs a patrol mission according to a preset patrol route within the substation, using a binocular camera mounted on the UAV, determining the current distance between the UAV and obstacles within the substation; determining the safe zone, alarm zone, and obstacle avoidance zone of the UAV's forward movement space based on the UAV's current position and flight direction; if the obstacle is determined to be in the alarm zone based on the current distance, reducing the UAV's flight speed and updating the current distance; if the obstacle is determined to be in the obstacle avoidance zone based on the updated current distance after speed reduction, adjusting the UAV's flight speed and flight direction according to the relative position of the UAV and the obstacle to avoid the obstacle, and updating the current distance; if the obstacle is determined to be in the safe zone based on the updated current distance after obstacle avoidance, triggering the UAV to return to the patrol route. This improves the obstacle avoidance capability of UAVs within substations in complex environments.

Description

Technical Field

[0001] This application relates to the field of intelligent inspection technology for substations, and in particular to a method, device, storage medium, and computer equipment for drone obstacle avoidance in a substation. Background Technology

[0002] With the continuous advancement of intelligent power systems, substation operation and maintenance are gradually moving towards unmanned and automated processes. Drones, due to their high efficiency and precision, have become an important technological means for substation inspections. Currently, drone inspections mainly rely on pre-set fixed routes, combining multiple sensors such as infrared, ultrasonic, and laser sensors for obstacle detection, enabling autonomous flight and initial obstacle avoidance. Under normal conditions, this method can meet the basic needs of daily inspections.

[0003] However, in practical applications, substations often contain dynamic obstacles such as temporary scaffolding, safety fences, and aerial work platforms. Traditional static flight paths cannot detect and avoid these changes in real time, which can easily lead to flight path deviations, collisions, or even mission interruptions. At the same time, existing sensors are limited by their technical principles, and their obstacle avoidance performance is not ideal when facing low temperatures, reflective surfaces, or small obstacles, further exposing the reality that UAVs have insufficient obstacle avoidance capabilities in complex environments. Summary of the Invention

[0004] The purpose of this application is to at least address one of the aforementioned technical deficiencies, particularly the technical deficiency of insufficient obstacle avoidance capability of UAVs in complex environments within substations in the prior art.

[0005] Firstly, this application provides a method for obstacle avoidance by unmanned aerial vehicles (UAVs) within a substation, the method comprising:

[0006] When the drones inside the substation perform their inspection tasks according to the preset inspection route, the binocular camera on the drones is used to determine the current distance between the drones and obstacles inside the substation.

[0007] Based on the drone's current position and flight direction, determine the safe zone, warning zone, and obstacle avoidance zone in the drone's forward movement space;

[0008] If the obstacle is determined to be in the warning zone based on the current distance, the drone's flight speed is reduced and the current distance is updated.

[0009] If the obstacle is determined to be in the obstacle avoidance zone based on the updated current distance after deceleration, the drone's flight speed and direction are adjusted according to the relative position of the drone and the obstacle to avoid the obstacle, and the current distance is updated.

[0010] If the obstacle is determined to be within the safe zone based on the updated current distance after obstacle avoidance, the drone will be triggered to return to its patrol route.

[0011] In one embodiment, the step of determining the current distance between the drone and obstacles within the substation using a binocular camera mounted on the drone includes:

[0012] After calibrating the binocular camera, images from the left and right perspectives of the drone are acquired simultaneously, and stereo correction processing is performed on the left and right perspective images to align them on the same horizontal line.

[0013] A semi-global matching algorithm is used to perform stereo matching on the left and right view images after stereo correction to obtain a disparity map containing obstacles.

[0014] Based on the disparity map and combined with the triangulation principle of binocular ranging, the position information of the obstacle in the left and right view images is extracted, and the current distance between the UAV and the obstacle is calculated.

[0015] In one embodiment, the step of determining the safe zone, warning zone, and obstacle avoidance zone of the drone's forward movement space based on the drone's current position and flight direction includes:

[0016] Using the drone's current position as a reference point and combining the drone's flight direction, a forward spatial projection area for the drone is constructed on a two-dimensional plane.

[0017] The projection area is divided into a safe zone, an alarm zone, and an obstacle avoidance zone according to their distance from the drone. The safe zone is located at the farthest end, the alarm zone is located in the middle, and the obstacle avoidance zone is located at the closest end.

[0018] In one embodiment, the step of reducing the flight speed of the drone includes:

[0019] Based on the drone's current flight speed and the changing trend of the distance between the drone and obstacles, the drone is controlled to slow down its flight speed to a level lower than its original speed.

[0020] In one embodiment, the step of adjusting the drone's flight speed and direction based on the drone's relative position to the obstacle includes:

[0021] When the obstacle is determined to be a static obstacle, the drone is controlled to slow down and fly in the opposite azimuth and altitude directions to the obstacle as the drone's flight direction;

[0022] When an obstacle is determined to be a dynamic obstacle, the drone's flight speed is increased to be higher than the obstacle's moving speed, and the azimuth and altitude directions opposite to those of the obstacle are taken as the drone's flight direction.

[0023] In one embodiment, the step of determining that an obstacle is a dynamic obstacle includes:

[0024] Based on images containing obstacles captured continuously by a binocular camera, if the position of the obstacle shifts in the continuously captured images, the obstacle is considered a dynamic obstacle.

[0025] or

[0026] Based on the infrared thermal images of obstacles continuously collected by the dual-light infrared camera on the drone, if the position of the infrared feature of the obstacle shifts in the continuously collected infrared thermal images, then the obstacle is a dynamic obstacle.

[0027] In one embodiment, the step of triggering the drone to return to its patrol route includes:

[0028] Control the drone to continue flying at the adjusted speed during obstacle avoidance, and adjust the drone's flight direction in the opposite direction based on the direction of deviation when the drone avoids the obstacle, so that the drone returns to the patrol route.

[0029] Secondly, this application provides an obstacle avoidance device for unmanned aerial vehicles (UAVs) in a substation, the device comprising:

[0030] The current distance determination module is used to determine the current distance between the drone and obstacles in the substation when the drone is performing an inspection mission according to the preset inspection route.

[0031] The area division module is used to determine the safe zone, warning zone, and obstacle avoidance zone of the drone's forward movement space based on the drone's current position and flight direction;

[0032] The first distance update module is used to reduce the drone's flight speed and update the current distance if it is determined that an obstacle is located in the alarm zone based on the current distance.

[0033] The second distance update module is used to adjust the drone's flight speed and direction according to the relative position of the drone and the obstacle if it is determined that the obstacle is located in the obstacle avoidance zone based on the current distance updated after deceleration, so as to avoid the obstacle and update the current distance.

[0034] The cruise return module is used to trigger the drone to return to the patrol route if the obstacle is determined to be in the safe zone based on the updated current distance after obstacle avoidance.

[0035] Thirdly, this application provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of any of the above embodiments of the UAV obstacle avoidance method in a substation.

[0036] Fourthly, this application provides a computer device, including: one or more processors, and a memory;

[0037] The memory stores computer-readable instructions, which, when executed by one or more processors, perform the steps of any of the above embodiments of the UAV obstacle avoidance method in a substation.

[0038] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0039] The obstacle avoidance method for UAVs in substations provided in this application utilizes binocular cameras to acquire real-time distance information to obstacles. Combined with the UAV's current position and flight direction, it dynamically divides the area into safe zones, alarm zones, and obstacle avoidance zones, enabling tiered responses to obstacles. When an obstacle is detected entering the alarm zone, the UAV promptly reduces its flight speed and dynamically updates the distance to the obstacle. Furthermore, when an obstacle enters the obstacle avoidance zone, the UAV intelligently adjusts its flight speed and direction based on its relative position to effectively avoid the obstacle, preventing path deviation or collision risks. Once obstacle avoidance is successful and the obstacle is out of danger, the UAV automatically returns to its original patrol route, ensuring the continuity and integrity of the patrol mission. This significantly improves the obstacle avoidance flexibility and safety of UAVs in substations when facing temporary structures, dynamic obstacles, and complex environmental changes. It effectively addresses the technical shortcomings of traditional methods in obstacle avoidance under complex environments, enhancing the reliability and intelligence of unmanned patrol operations. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart illustrating the obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation provided in an embodiment of this application.

[0042] Figure 2 Example diagram of the triangulation principle provided in the embodiments of this application;

[0043] Figure 3 Example diagram of the projection of the drone and its surrounding area provided in the embodiments of this application;

[0044] Figure 4 Example diagram showing the relative positions of the UAV and obstacles in the patrol route provided in this application embodiment;

[0045] Figure 5 This is an example diagram of the drone obstacle avoidance process provided in the embodiments of this application;

[0046] Figure 6 A schematic diagram of the structure of the UAV obstacle avoidance device in the substation provided in an embodiment of this application;

[0047] Figure 7 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0048] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0049] This application provides a method for obstacle avoidance by unmanned aerial vehicles (UAVs) within a substation. The following embodiments illustrate this method using a computer device as an example. It is understood that the computer device can be any device with data processing capabilities, including but not limited to a single server, server cluster, personal laptop, desktop computer, etc. Figure 1 As shown, the method may include the following steps:

[0050] S101: When a drone in a substation performs an inspection mission according to a preset inspection route, the drone's binocular camera is used to determine the current distance between the drone and obstacles in the substation.

[0051] Among them, "drone" refers to an aircraft device capable of autonomously flying and performing operations according to preset tasks without direct human control. "Binocular camera" refers to a sensor device equipped on a drone that acquires image information through two independent camera components with a fixed baseline distance, and calculates the distance to objects based on image parallax information. "Obstacles within the substation" refers to physical objects existing inside the substation that may affect the drone's flight trajectory, including but not limited to temporary scaffolding, safety fences, and aerial work platforms. "Current distance" refers to the actual spatial distance between the drone and the obstacle, measured and determined by the drone at a specific moment using the binocular camera. "Preset inspection route" refers to the flight path pre-planned and stored in the drone's control system based on the substation layout and inspection requirements, for the drone to autonomously follow during task execution.

[0052] In this step, when the drone is performing its inspection mission within the substation according to the preset inspection route, the computer equipment calls the onboard binocular camera module to simultaneously collect left and right view images of the substation environment, and determines the positional relationship of obstacles based on the image information, thereby obtaining the current distance data between the drone and the obstacles.

[0053] Considering the diverse types and sizes of obstacles within substations, image recognition algorithms can be further integrated to preprocess images captured by binocular cameras, such as obstacle contour extraction, edge detection, and volume estimation, to improve the accuracy of obstacle recognition and the stability of ranging. If multiple obstacles are identified in the image, a priority ranking strategy, such as prioritizing closest obstacles, can be used to determine the target objects requiring focused monitoring, and the obstacle list can be maintained in real time for subsequent action calls. Simultaneously, to ensure sufficient reliability of ranging data in actual flight control, the computer equipment can perform multi-frame fusion processing on the current distance data, such as smoothing noise through sliding window averaging and outlier removal, reducing the risk of misjudgment of flight commands due to single mismeasurements. If an abnormally rapid change in the current distance is detected, such as due to camera obstruction or reflection causing recognition failure, a backup obstacle avoidance mechanism can be automatically triggered, such as short-term hovering and re-acquiring images, further improving the robustness of ranging in complex environments.

[0054] The computer equipment will then continuously output updated current distance data, providing real-time basis for subsequent decisions such as flight speed adjustment, direction control, and path avoidance, ensuring that the UAV can flexibly adjust its flight behavior according to changes in the surrounding environment, thereby safely and efficiently completing the inspection mission within the substation.

[0055] It is understandable that by using the onboard binocular camera to determine the current distance to obstacles in real time when the drone is performing a preset inspection route, it is possible to accurately perceive dynamic or temporary obstacles in the complex environment of the substation, and avoid flight deviations or collision risks caused by the inability of traditional static routes to reflect environmental changes in a timely manner.

[0056] S102: Based on the drone's current position and flight direction, determine the safe zone, warning zone, and obstacle avoidance zone in the drone's forward movement space.

[0057] The current location refers to the real-time spatial position information of the drone, acquired by its own positioning module, such as GPS or inertial navigation system. Flight direction refers to the drone's current forward orientation, determined by the flight control system based on flight attitude data. Forward space refers to the spatial area the drone will traverse along its flight direction. The safe zone is the area within the forward space where obstacles are at a distance greater than a preset safety threshold and pose no threat to flight. The warning zone is the area within the forward space where obstacles are at a distance between the warning threshold and the obstacle avoidance threshold, indicating a need for attention but not yet a direct threat. The obstacle avoidance zone is the area within the forward space where obstacles are at a distance less than the obstacle avoidance threshold, requiring emergency avoidance measures.

[0058] In this step, as the UAV performs its patrol mission along a preset patrol route, the computer equipment receives the UAV's position information in real time and, combined with flight direction data, dynamically constructs a forward spatial model within a certain range ahead of the UAV. This forward spatial model can be a three-dimensional spatial region extended along the current flight direction from the UAV's current position as a reference point, forming a defined forward distance and width for obstacle detection and flight path evaluation. To make the forward spatial model more consistent with actual flight requirements, the computer equipment can comprehensively consider the UAV's own size parameters, current flight speed, and the control response time of the flight control system, and reasonably set the forward detection radius and depth, enabling the model to cover all spatial areas that the UAV may reach in a short time, thus providing an accurate spatial basis for subsequent obstacle recognition and obstacle avoidance decisions.

[0059] After the forward space model is constructed, the computer equipment further performs safety zoning on the space based on preset multi-level distance thresholds. Specifically, the computer equipment can divide the space into multiple continuous spatial segments or concentric layers along the flight direction, with each spatial segment corresponding to a different safety level. Spatial segments that are far away and have not detected obstacles are designated as safe zones, meaning the drone can pass through normally in its current flight state; if an obstacle is detected within a medium distance range, it corresponds to an alarm zone, indicating a potential collision risk; if an obstacle approaches to within a preset minimum safe distance threshold, it corresponds to an obstacle avoidance zone, requiring immediate obstacle avoidance actions, such as reducing flight speed, adjusting flight direction, or even emergency hovering, to avoid a collision.

[0060] Furthermore, to further enhance the adaptability and intelligence of UAVs, computer equipment can dynamically optimize and adjust the detection range, safety zone, and alarm zone boundary thresholds of the forward space model based on historical inspection data, flight logs, and obstacle distribution patterns. For example, the safety zone can be appropriately narrowed in areas with dense obstacles to improve sensitivity, while the detection range can be widened in open areas to improve flight efficiency. Through this dynamic adaptive mechanism, the continuity of inspection missions and flight safety can be effectively balanced, thereby improving the UAV's automatic inspection capabilities and overall stability in various complex substation environments.

[0061] It is understandable that by dynamically determining the safe zone, warning zone, and obstacle avoidance zone of the drone's forward flight space based on its current position and flight direction, computer equipment can reasonably assess the flight risk level based on the real-time distance between obstacles and the drone, thereby providing early warnings or proactively avoiding potential threats. This zoned management approach improves environmental adaptability and responsiveness, enabling drones to perform inspection tasks more stably in the complex and dynamic environment of substations. This effectively reduces the risk of flight accidents caused by sudden obstacles, and comprehensively improves the safety and reliability of inspection operations.

[0062] S103: If the obstacle is determined to be in the warning zone based on the current distance, reduce the drone's flight speed and update the current distance.

[0063] During the drone's patrol mission along a preset route, the computer equipment receives environmental data in real time from binocular cameras, lidar, or other ranging sensors. Based on the current location and flight direction information, it continuously monitors the distribution of obstacles in the space in front of the drone. When the computer equipment determines that an obstacle is within the warning zone based on the detected current distance data, it immediately triggers a flight speed adjustment mechanism to ensure flight safety and allow sufficient time and space for subsequent obstacle avoidance maneuvers.

[0064] Specifically, the computer equipment first generates a flight speed adjustment command, instructing the drone to reduce its current flight speed by a preset percentage. The percentage reduction can be flexibly set according to actual application needs; for example, reducing the current speed to 70%-80% of the original speed to effectively slow the drone's approach to obstacles and increase reaction time when dynamic changes in obstacles are detected. To ensure flight stability, flight speed adjustment can employ a uniform deceleration mode, gradually reducing the flight speed at a constant deceleration within a set time window; or a graded deceleration mode, reducing the flight speed in stages, flexibly adjusting the deceleration amplitude according to the proximity of obstacles at different stages, thereby achieving a smooth transition and avoiding problems such as drone attitude instability and flight vibration caused by sudden deceleration.

[0065] As the flight speed decreases, the computer equipment continues to monitor the surrounding environment at a high frequency, acquiring new obstacle distance data in real time. Based on the latest sensor measurements, the computer equipment recalculates the current distance between the drone and obstacles and dynamically updates the current distance parameters. The updated current distance serves as the real-time basis for flight control decisions, used to determine changes in the obstacle's state, i.e., whether the obstacle has moved out of the warning zone due to relative motion, whether it has further entered the obstacle avoidance zone, or whether it remains within the warning zone.

[0066] In specific application scenarios, such as when a drone patrols near a substation work area and detects temporary scaffolding within the warning zone, the computer equipment triggers a speed reduction action based on the current distance. While slowing down, the drone continuously monitors the scaffolding's position in real time. If the obstacle remains relatively stationary and beyond a safe distance after the speed reduction, the drone can continue its patrol mission at low speed. Through dynamic flight speed management and real-time obstacle tracking mechanisms, the stability and resilience of drones performing patrol missions in complex environments can be effectively improved.

[0067] It is understandable that by reducing flight speed and updating the current distance in real time when an obstacle is determined to be in the warning zone at the current distance, the UAV's reaction time during the approach to the obstacle can be effectively extended, reducing flight risks, while ensuring that the computer equipment makes more accurate flight decisions based on the latest environmental data. This speed reduction mechanism based on dynamic distance updates improves the UAV's adaptability and flexibility in handling sudden environmental changes, thereby significantly enhancing the safety and mission continuity of inspections in the complex environment of substations.

[0068] S104: If the obstacle is determined to be in the obstacle avoidance zone based on the current distance updated after deceleration, the drone's flight speed and direction are adjusted according to the relative position of the drone and the obstacle to avoid the obstacle, and the current distance is updated.

[0069] The relative position refers to the spatial orientation and distance relationship between the obstacle and the current position of the drone, which may include directional information such as left and right, up and down, and front and back.

[0070] During the drone's patrol along the preset route, the computer system receives environmental data from sensors in real time and continuously updates the current distance between the drone and obstacles based on the decelerated flight status. When the updated current distance is detected to be less than the obstacle avoidance zone threshold, the computer system immediately activates the obstacle avoidance control logic, prioritizing the relative position of obstacles to formulate a detailed flight adjustment plan.

[0071] Specifically, the computer equipment first uses a multi-sensor fusion algorithm to accurately extract the three-dimensional spatial position of the obstacle relative to the drone, including horizontal and vertical angles. Based on the relative position characteristics, the computer equipment determines which side of the flight path the obstacle is located on, such as left front, right front, directly front, above, or below. If the obstacle is located to the left front, the computer equipment can prioritize instructing the drone to yaw to the right to avoid it; if the obstacle is directly in front and there is no clear path to the left or right, it further determines whether an ascending or descending path is feasible, and chooses to avoid it vertically. While adjusting the flight direction, the computer equipment comprehensively evaluates the remaining obstacle avoidance time window based on the obstacle's approach speed and its own flight speed, and dynamically adjusts the flight speed. Specifically, when the obstacle approaches the drone at a relatively high speed, the flight control system is instructed to further reduce the flight speed, and if necessary, reduce to a low-speed cruise state or temporarily hover; when the obstacle's position is relatively stable and there is sufficient avoidance space, a low-speed movement can be maintained to ensure the continuity and smoothness of control during the avoidance process.

[0072] To ensure smooth maneuverability, flight speed and direction adjustments can be issued using a parallel command mechanism. This means that deceleration is performed simultaneously with heading adjustments, preventing missed opportunities for obstacle avoidance due to delays in single actions. Meanwhile, the computer system is configured with a dynamic adjustment cycle for maneuver execution, such as resampling obstacle positions and updating the current distance every 50ms, to correct flight adjustment strategies in real time and ensure precise synchronization between the UAV's flight status and environmental changes. During flight adjustment maneuvers, if obstacles remain within the obstacle avoidance zone, the computer system can appropriately increase the magnitude of avoidance maneuvers, such as increasing yaw angles or climb / descent rates, to further increase the safe distance from the obstacle.

[0073] Taking a real-world application scenario as an example, during power transmission line inspection, if the boom of an aerial work platform used for temporary lifting operations suddenly extends into the inspection route, and the drone detects the boom entering the obstacle avoidance zone even after slowing down, the computer equipment can quickly analyze that the boom's position is to the left front and slightly higher than the current flight altitude. It then generates a right yaw and slight descent flight adjustment command, simultaneously instructing the flight speed to be reduced to 50% of its original speed. During obstacle avoidance, the computer equipment continuously tracks the relative movement trajectory of the boom. If it detects that the boom continues to extend, it commands a further right turn and descent to ensure safe passage. If it detects that the boom has stabilized or retracted, it commands the drone to smoothly return to the inspection route based on the current distance data, ensuring the smooth execution of the entire inspection mission and flight safety.

[0074] It is understandable that by adjusting flight speed and direction in a timely manner when obstacles are located within the obstacle avoidance zone, drones can flexibly avoid potential collision risks within a limited space, maximizing flight safety. Dynamic adjustment of flight speed provides ample operational time and control margin for flight path adjustments, preventing obstacle avoidance maneuvers from being unable to be completed in time due to excessive speed. Dynamic adjustment of flight direction allows the drone to flexibly select the optimal avoidance path based on changes in obstacle position, enhancing environmental adaptability. Continuously updating the current distance and dynamically adjusting the flight status accordingly ensures that obstacle avoidance decisions are synchronized with environmental changes in real time, improving the drone's response speed and autonomous decision-making capabilities in complex environments. This effectively reduces the risk of flight accidents caused by sudden obstacles, ensuring the successful completion of patrol missions.

[0075] S105: If the obstacle is determined to be in the safe zone based on the updated current distance after obstacle avoidance, the drone will be triggered to return to the patrol route.

[0076] After the drone completes obstacle avoidance maneuvers, the computer continuously updates and monitors the current distance data based on environmental information fed back from the sensors. When it detects that the obstacle has moved away from the safe zone and its position is stable with no tendency to continue approaching, the computer triggers the return-to-course logic, initiating the recovery control of the drone's heading and speed.

[0077] Specifically, the computer equipment first calculates the regression path based on the deviation between the current flight status, such as current position, heading angle, and flight speed, and the reference point of the preset patrol route. The regression path can be generated using shortest path algorithms, such as straight-line interpolation or curved path planning, to balance flight stability and mission efficiency. If the deviation is small, the UAV can be directly instructed to adjust its heading to align with the original route; if the deviation is large, adjustments need to be made in stages, including first approaching horizontally and then correcting in the forward direction, to avoid sharp turns that could cause flight attitude instability.

[0078] After path planning is completed, the computer equipment instructs the flight control system to adjust the flight speed synchronously. To ensure a smooth transition, a speed ramp-up curve can be set, that is, gradually recovering from the low speed used during obstacle avoidance to the normal cruise speed over a certain time or distance. For example, linear acceleration in 5% increments per second can be used to avoid flight jitter or inertial yaw caused by sudden acceleration.

[0079] During flight status adjustments, the computer equipment continues to maintain real-time awareness of obstacles and the surrounding environment to ensure that no new obstacles interfere with the return process. If a new obstacle is detected entering the warning zone or obstacle avoidance zone during the return process, the return command is interrupted first, and obstacle avoidance actions are re-executed, prioritizing flight safety over route restoration.

[0080] For example, in a power transmission line inspection scenario, after avoiding the boom of a temporarily raised aerial work platform, the drone detects through sensors that the boom has retracted and moved to a safe distance. At this point, the computer generates the nearest return path, guiding the drone to make a slight right turn and ascend slowly, gradually returning to the original inspection route. Simultaneously, the flight speed is adjusted to return to cruising speed, for example, 3 meters per second. The entire process is conducted based on dynamic environmental perception, ensuring the continuity, smoothness, and safety of the return maneuver.

[0081] It is understandable that adopting the method of triggering the return patrol route after an obstacle enters the safe zone ensures that the UAV can promptly resume its original patrol mission after completing obstacle avoidance maneuvers, thereby minimizing mission downtime and improving patrol efficiency. By assessing environmental safety based on real-time updated current distances, it can dynamically adapt to different obstacle movement states, improving the accuracy of environmental perception and the reliability of decision-making. By calculating the return path and adjusting the heading and speed in stages, risks such as unstable flight attitude and sharp turns can be effectively avoided, ensuring the stability and safety of the flight process. At the same time, continuously sensing environmental changes during the return process allows for timely interruption of the return action in abnormal situations, prioritizing the flight safety of the UAV and further enhancing its robustness and intelligence in complex patrol environments.

[0082] In the above embodiments, by utilizing binocular cameras to acquire real-time distance information to obstacles, and combining this with the drone's current position and flight direction, safe zones, alarm zones, and obstacle avoidance zones are dynamically divided to achieve graded responses to obstacles. When an obstacle is detected entering the alarm zone, the drone promptly reduces its flight speed and dynamically updates the distance to the obstacle. Furthermore, when an obstacle enters the obstacle avoidance zone, the drone intelligently adjusts its flight speed and direction based on its relative position to effectively avoid the obstacle and prevent path deviation or collision risks. Once obstacle avoidance is successful and the obstacle is out of danger, the drone can automatically return to its original inspection route, ensuring the continuity and integrity of the inspection mission. This significantly improves the obstacle avoidance flexibility and safety of substation drones when facing temporary structures, dynamic obstacles, and complex environmental changes, effectively addressing the technical shortcomings of traditional methods in obstacle avoidance under complex environments, and enhancing the reliability and intelligence level of unmanned inspection operations.

[0083] In one embodiment, the step of determining the current distance between the drone and obstacles within the substation using a binocular camera mounted on the drone includes:

[0084] After calibrating the binocular camera, images from the left and right perspectives of the drone are acquired simultaneously, and stereo correction processing is performed on the left and right perspective images to align them on the same horizontal line.

[0085] A semi-global matching algorithm is used to perform stereo matching on the left and right view images after stereo correction to obtain a disparity map containing obstacles.

[0086] Based on the disparity map and combined with the triangulation principle of binocular ranging, the position information of the obstacle in the left and right view images is extracted, and the current distance between the UAV and the obstacle is calculated.

[0087] Camera calibration refers to the process of measuring known calibrated objects captured by a camera and using calibration algorithms to calculate the camera's internal and external parameters, thereby eliminating lens distortion and accurately describing the camera's position and orientation in three-dimensional space. Stereo correction refers to preprocessing the left and right view images acquired by a binocular camera to align them on the same plane, eliminating image distortion and viewpoint differences, thus providing a more accurate foundation for subsequent stereo matching. The semi-global matching algorithm is an algorithm used to calculate disparity maps. It optimizes the cost function to accurately match pixels in the binocular images, thereby calculating the disparity value. The disparity map is an image generated by the stereo matching algorithm, containing disparity information for each pair of pixels, and is typically used to represent the depth relationship between objects and the camera. Triangulation is based on known camera positions and disparity information, using geometric derivation to calculate the three-dimensional position of obstacles, and thus obtain the distance to the obstacles.

[0088] During drone flight, the first step is to calibrate the binocular cameras to ensure that the acquired image data accurately reflects the true spatial information. The calibration process involves calculating the camera's intrinsic and extrinsic parameters to eliminate distortion in the camera's imaging. This can be achieved by photographing a calibration board of known size and location to acquire image data from multiple perspectives, and then using calibration algorithms to calculate precise camera parameters. After calibration, the computer performs stereo correction processing based on the simultaneously acquired left and right view images from the drone. The core of this processing is aligning the left and right view images on the same plane, ensuring that the two images are not misaligned in the horizontal coordinate system. Using image transformation algorithms in computer vision technology, such as feature point matching or region alignment methods, the computer can eliminate parallelism differences between the images.

[0089] Next, the computer device employs a semi-global matching algorithm to perform stereo matching on the corrected left and right view images. This algorithm optimizes the cost function for disparity calculation, enabling precise matching of pixels in the left and right images and generating a disparity map. The disparity value of each pixel in the disparity map represents the depth information of that point, reflecting the relative distance between the obstacle and the camera. To improve computational efficiency and accuracy, the SGM algorithm considers global consistency between pixels, enabling accurate acquisition of depth information in complex scenes. Based on the generated disparity map, the computer device utilizes the triangulation principle of binocular ranging to extract the obstacle's position information in the left and right view images. Specifically, by analyzing the disparity values ​​of the obstacle's position in the left and right images, combined with the camera's baseline distance and calibration parameters, the computer device calculates the obstacle's three-dimensional coordinates using geometric principles. In this way, the computer device can accurately estimate the current distance between the drone and the obstacle.

[0090] In this application scenario, suppose a drone is conducting an inspection of a building's exterior facade. Its binocular cameras capture real-time image data of the building's exterior walls. After calibration and stereo correction, the computer generates a disparity map and uses this map to calculate the accurate distances to obstacles on the building's surface. During flight, if obstacles appear in the drone's path, the computer can quickly assess their hazard level based on their calculated positions and distances, and take appropriate obstacle avoidance actions.

[0091] By calibrating and stereo correcting the binocular cameras, camera distortion and viewpoint differences can be eliminated, ensuring the accuracy of image data. A semi-global matching algorithm is used to perform stereo matching on the left and right view images, efficiently acquiring obstacle depth information, generating disparity maps, and extracting the three-dimensional position information of obstacles. Based on this, the current distance calculated using triangulation principles provides real-time and accurate obstacle distance data for the UAV's flight decisions, thereby improving the perception of the flight environment and the accuracy of obstacle avoidance decisions. This processing method not only improves the accuracy of obstacle detection but also optimizes the UAV's autonomous flight capabilities in complex environments, ensuring the safety and efficiency of the flight process.

[0092] In one example, binocular ranging is based on the principle of triangulation, using the relative positions of the left and right cameras and the parallax in the image to calculate the distance between the target object and the cameras. This requires that the two cameras be in the same plane, with parallel optical axes and the same focal length. Specifically, as... Figure 2 As shown, L1 and L2 are the optical centers of the left and right cameras, respectively, f is the focal length of the left and right cameras, T is the distance between the optical centers of the two cameras, and y1 and y2 represent the x-coordinates of the image points of the target point O on the left and right cameras, respectively. The parallax S is defined as y1 minus y2, and the distance to the target to be measured is d. According to the similar triangle theorem, the following formula can be derived:

[0093]

[0094] The focal length f and optical center distance T can be obtained through camera calibration. By calculating the parallax value y1 - y2, the distance information of the target object can be calculated using this formula.

[0095] The binocular ranging process typically involves four steps: camera calibration, stereo correction, stereo matching, and distance calculation. First, camera calibration quantifies the camera's geometric and optical characteristics using mathematical models. The aim is to determine the camera's intrinsic parameters (such as focal length, principal point coordinates, and distortion coefficients) and extrinsic parameters (such as the transformation relationship between the camera coordinate system and the world coordinate system). Calibration ensures the camera's measurement accuracy, providing a solid foundation for subsequent calculations. Next, stereo correction maps the images from the left and right cameras onto the same plane, ensuring that points on these images are geometrically coplanar and parallel. Since in reality, binocular cameras may not perfectly meet the ideal conditions of coplanarity, parallel optical axes, and identical focal lengths, stereo correction eliminates this error, guaranteeing the accuracy of the subsequent matching process.

[0096] After stereo correction, stereo matching is performed, which involves finding corresponding matching points in the left and right images and calculating the disparity value. To improve the accuracy and efficiency of matching, this invention employs the SGBM (Semi-Global Block Matching) algorithm, which comprehensively considers matching accuracy and computational complexity, thus providing high accuracy while ensuring real-time performance. Finally, based on the obtained disparity map, the disparity value is converted into the three-dimensional spatial coordinates of the target object using the triangulation principle of binocular ranging, and the actual distance between the obstacle and the UAV is calculated. Through this process, the computer equipment can obtain the distance information between the UAV and surrounding obstacles in real time, thereby supporting obstacle avoidance and path planning tasks for the UAV in complex environments.

[0097] In one embodiment, the step of determining the safe zone, warning zone, and obstacle avoidance zone of the drone's forward movement space based on the drone's current position and flight direction includes:

[0098] Using the drone's current position as a reference point and combining the drone's flight direction, a forward spatial projection area for the drone is constructed on a two-dimensional plane.

[0099] The projection area is divided into a safe zone, an alarm zone, and an obstacle avoidance zone according to their distance from the drone. The safe zone is located at the farthest end, the alarm zone is located in the middle, and the obstacle avoidance zone is located at the closest end.

[0100] The forward spatial projection area of ​​the drone is a two-dimensional area projected onto the ground or horizontal plane based on the drone's current position and flight direction, representing the horizontal area that the drone will cover along its flight path.

[0101] Specifically, to ensure the safety of drones in flight, computer equipment needs to calculate and update the projected area of ​​the forward trajectory space in real time based on the drone's current position and flight direction. First, by acquiring the drone's current GPS positioning information and flight direction, the computer equipment can determine its position and orientation on a two-dimensional plane. Then, based on this data, the computer equipment constructs an area on the two-dimensional plane representing the drone's flight path, typically a rectangular or fan-shaped area extending along the flight direction with the drone at its center. The size and shape of this area depend on factors such as the drone's flight speed, altitude, and control response time.

[0102] After constructing the forward flight path projection area, the computer system divides this area into different safety level zones. Specifically, the furthest area is defined as the safe zone, characterized by a relatively large distance between the drone and obstacles within this zone, allowing for minimal adjustments. Closer to the drone, an alarm zone is defined, where obstacles are close enough to potentially affect the flight path but do not yet require immediate obstacle avoidance. In the obstacle avoidance zone, obstacles are very close, requiring immediate action, such as slowing down or changing the flight path to avoid collisions. The drone's sensors monitor changes in the surrounding environment in real time, and the distance data collected by the sensors is fed back to the computer system, which adjusts the flight status accordingly. If a new obstacle appears in a zone, the computer system dynamically adjusts the zone division based on preset safety standards and takes appropriate actions based on the current flight conditions, such as activating obstacle avoidance mode or issuing an early warning.

[0103] In one example, the drone and a certain area around its direction of travel are projected onto a two-dimensional plane. This area will be divided into a safe zone, an alarm zone, and an obstacle avoidance zone, such as... Figure 3 As shown. Considering that the UAV and the live equipment in the substation must maintain a certain safe distance to avoid damage to the UAV during the autonomous inspection mission caused by induced electricity or the positioning error of the UAV, the obstacle avoidance zone of the UAV is defined as (0, 6) meters, the alarm zone is [6, 8] meters, and the safety zone is (8, 10] meters.

[0104] In this embodiment, dividing the forward space into a safe zone, an alarm zone, and an obstacle avoidance zone helps to achieve precise flight status management. By clearly defining different zones, the computer equipment can respond differently to different types of obstacles based on their distance from the drone, ensuring that the drone has good obstacle avoidance capabilities in complex environments.

[0105] In one embodiment, the step of reducing the flight speed of the drone includes:

[0106] Based on the drone's current flight speed and the changing trend of the distance between the drone and obstacles, the drone is controlled to slow down its flight speed to a level lower than its original speed.

[0107] The distance change trend between obstacles refers to the pattern of distance change between the drone and obstacles over time. Specifically, this trend can be measured in real time by distance sensors, such as lidar or ultrasonic sensors, to infer whether the obstacle is approaching or moving away from the drone.

[0108] Specifically, the computer equipment needs to monitor and adjust the drone's flight status in real time. First, the computer equipment acquires the drone's current flight speed and updates it in real time. Next, the computer equipment uses the drone's sensor data, such as lidar and visual sensors, to measure and calculate the real-time distance between the drone and surrounding obstacles.

[0109] By analyzing real-time data, the computer device can detect trends in the distance between the drone and obstacles. For example, if the distance gradually decreases, it indicates that an obstacle is approaching the drone, requiring a corresponding response. The computer device will determine, based on a set threshold, whether it is necessary to reduce the flight speed to slow the approach of the drone to the obstacle, requiring more time for subsequent obstacle avoidance maneuvers.

[0110] Specifically, if the computer determines that an obstacle is approaching at a relatively high speed, it reduces the drone's flight speed to slow its approach. At this point, the drone's power output is recalculated based on the new speed target to ensure a smooth speed reduction and avoid instability caused by excessively rapid deceleration. This process is completed through real-time feedback adjustments. Sensors continuously monitor changes in obstacle distance and transmit the data to the computer. Based on this data, the flight speed is adjusted so that the drone can fly safely at a lower speed than before, providing sufficient reaction time and space for subsequent flight adjustments or obstacle avoidance.

[0111] In this embodiment, slowing down the drone's flight speed based on the changing trend of the distance between the drone and obstacles can significantly improve the drone's flight safety. First, this adjustment allows the drone more time to make judgments and adjustments when approaching obstacles, reducing the risk of collision. Furthermore, reducing flight speed can improve flight stability, avoiding flight instability or control difficulties caused by sudden speed changes. Thus, it not only helps improve the drone's obstacle avoidance capabilities but also enhances its ability to cope with complex environments, thereby improving its safety and reliability in practical applications.

[0112] In one embodiment, the step of adjusting the drone's flight speed and direction based on the drone's relative position to the obstacle includes:

[0113] When the obstacle is determined to be a static obstacle, the drone is controlled to slow down and fly in the opposite azimuth and altitude directions to the obstacle as the drone's flight direction;

[0114] When an obstacle is determined to be a dynamic obstacle, the drone's flight speed is increased to be higher than the obstacle's moving speed, and the azimuth and altitude directions opposite to those of the obstacle are taken as the drone's flight direction.

[0115] Obstacles refer to any object in the drone's flight path that may hinder or interfere with its flight. Obstacles can be divided into two types: static obstacles and dynamic obstacles. Static obstacles are objects whose position in the environment remains fixed and does not change, such as buildings or trees; while dynamic obstacles are objects whose position changes in space, such as moving vehicles or animals. The azimuth direction refers to the angle between the drone's flight direction and true north on the ground, a two-dimensional angle value; the altitude direction refers to the drone's vertical height relative to the ground, i.e., the up and down direction in three-dimensional space.

[0116] In this embodiment, the UAV performs intelligent flight control based on the type of surrounding obstacles and its relative position to the obstacles. First, it acquires information about the obstacles to determine whether they are static or dynamic. If the obstacle is identified as static, flight control is adjusted based on the distance between the obstacle and the UAV. Specifically, when the obstacle is close to the UAV, the flight speed is automatically reduced to ensure the UAV approaches the obstacle at a low speed, thereby avoiding a collision. The flight speed is adjusted in real time based on the distance to the obstacle and the safety threshold of the flight path. Simultaneously, the azimuth and altitude directions opposite to the obstacle are calculated to guide the UAV to make appropriate flight turns, ensuring the UAV bypasses static obstacles and continues flying in a safe direction.

[0117] If the obstacle is dynamic, a more rapid and proactive response is required. In this situation, the relative position of the obstacle is constantly changing. The drone's future position is predicted by tracking its movement trajectory, and adjustments are made accordingly. To ensure the drone can avoid the obstacle in time, the flight speed is adjusted based on the obstacle's speed. If the obstacle's speed is lower than the drone's current speed, the drone's speed is appropriately increased to quickly escape the obstacle's influence. Furthermore, the drone's flight direction is adjusted in real-time based on the obstacle's position, azimuth, and altitude to ensure safe flight and avoidance of the obstacle.

[0118] To ensure flight accuracy and real-time performance, the flight control strategy is continuously updated through high-frequency data acquisition and processing. Whenever an obstacle's position changes, the flight speed and direction are dynamically adjusted based on the new data. This intelligent adjustment not only helps the drone react to a single obstacle but also handles the simultaneous presence of multiple obstacles in complex environments. It comprehensively considers the position, speed, and flight path of each obstacle to calculate the optimal obstacle avoidance strategy in real time, ensuring the drone can complete its flight mission safely and effectively. In multi-obstacle scenarios, the type and risk level of obstacles are prioritized, and the optimal flight path is dynamically planned to avoid collisions and maximize flight safety. Through this flight control strategy, the drone can achieve intelligent obstacle avoidance in complex environments and continuously optimize its flight path, improving flight safety and flexibility.

[0119] It's understandable that drones can adjust their flight speed and direction based on the type of obstacle. For static obstacles, they avoid collisions by slowing down and circumventing them; for dynamic obstacles, they accelerate and adjust their flight trajectory in real time to ensure they quickly avoid dangerous areas. This intelligent flight strategy optimizes flight paths, enhances the drone's adaptability in complex environments, improves flight safety, and ensures that the drone can effectively handle various complex situations through real-time path planning and speed control, minimizing safety hazards during flight.

[0120] In one embodiment, the step of determining that an obstacle is a dynamic obstacle includes:

[0121] Based on images containing obstacles captured continuously by a binocular camera, if the position of the obstacle shifts in the continuously captured images, the obstacle is considered a dynamic obstacle.

[0122] or

[0123] Based on the infrared thermal images of obstacles continuously collected by the dual-light infrared camera on the drone, if the position of the infrared feature of the obstacle shifts in the continuously collected infrared thermal images, then the obstacle is a dynamic obstacle.

[0124] Infrared dual-light cameras capture images by sensing the infrared radiation emitted by target objects, making them suitable for target detection in low-light or nighttime environments. Image position shift refers to the change in the relative position of obstacles between consecutive image frames, which is achieved by calculating the target point position in the image. Infrared thermal images, acquired by infrared dual-light cameras, typically display the thermal characteristics of target objects through radiation at different temperatures, helping to identify and determine the presence and location of targets.

[0125] In this embodiment, the UAV's binocular camera or infrared dual-light camera first continuously captures images containing obstacles. For binocular cameras, a stereo matching algorithm is used to synthesize the left and right view images and calculate the obstacle's position at each moment. If a shift in the relative position of the obstacle is detected in the continuous images, the degree of position change is calculated to determine that the obstacle is a dynamic obstacle. To this end, image registration technology is applied to align the images at consecutive moments and compare the obstacle's displacement in different images.

[0126] Similarly, for dual-light infrared cameras, continuously acquired infrared thermal images are analyzed. Obstacles in the infrared images are displayed as infrared radiation characteristics. By detecting changes in the position of the infrared characteristics of obstacles at different points in time, it is calculated whether the obstacle has moved. If a change in the position of the infrared characteristics of an obstacle is detected, the obstacle is also determined to be a dynamic obstacle. Unlike binocular cameras, infrared thermal images utilize thermal radiation information, making them suitable for target detection under conditions such as low light or nighttime.

[0127] It's understandable that detecting changes in image content helps determine whether an obstacle is dynamic. Image displacement detection ensures intelligent differentiation between static and dynamic obstacles, providing accurate obstacle status information. After a dynamic obstacle is identified, the drone adjusts its flight accordingly based on changes in the obstacle's speed and position, adopting a more proactive obstacle avoidance strategy. This improves the drone's obstacle avoidance capabilities in complex environments, enhancing flight safety and reliability. Simultaneously, the application of infrared thermal imaging technology allows the drone to accurately identify dynamic obstacles even in low-light or complete darkness, ensuring all-weather flight safety. Therefore, this dynamic obstacle detection scheme based on continuous image analysis effectively enhances the drone's autonomous navigation and obstacle avoidance capabilities.

[0128] In one embodiment, the step of triggering the drone to return to its patrol route includes:

[0129] Control the drone to continue flying at the adjusted speed during obstacle avoidance, and adjust the drone's flight direction in the opposite direction based on the direction of deviation when the drone avoids the obstacle, so that the drone returns to the patrol route.

[0130] Among them, deviation from direction refers to the change in course of the drone due to obstacle avoidance operations during the process of avoiding obstacles.

[0131] In this embodiment, firstly, the computer device determines whether the drone has collided with an obstacle or needs to avoid it based on the drone's real-time flight data and obstacle detection data. If an obstacle exists in the drone's flight path and the drone has already performed an obstacle avoidance maneuver, its flight speed is adjusted based on the drone's current speed and obstacle avoidance direction. Next, the drone may deviate from its original flight path after obstacle avoidance. By analyzing the drone's current position and the direction of deviation, the specific location of the deviation is identified, and the opposite correction direction is calculated. For this purpose, the drone's heading and position can be monitored in real time, and the current deviation angle can be determined by calculating the heading angle. Then, a command is issued in the opposite direction to adjust the drone's flight direction, enabling it to return to the original patrol route.

[0132] It is understandable that after obstacle avoidance, the flight direction has already deviated. Timely adjustment of the course in the opposite direction of the deviation helps the drone quickly and accurately return to its original patrol route, preventing missed inspections or mission interruptions due to prolonged deviation. This significantly improves drone safety in complex environments, reducing the risk of flight accidents caused by obstacles; it also ensures the continuity and integrity of patrol missions, reduces human intervention, and improves operational efficiency.

[0133] To facilitate understanding of the scheme in this application, specific examples are provided below.

[0134] In this embodiment, as Figure 4 As shown, a phased obstacle avoidance strategy is proposed for different scenarios when obstacles are located in the safe zone, alarm zone, and obstacle avoidance zone during autonomous inspection missions of UAVs within substations. Scenario 1 indicates that the obstacle is within the safe zone, i.e., the distance d between the UAV and the obstacle is greater than 8 meters and less than or equal to 10 meters. In this scenario, the UAV continuously detects, monitors, and measures the distance to the obstacle using its binocular camera, without needing to adjust its current flight speed and heading. Scenario 2 indicates that the obstacle enters the alarm zone, i.e., d is greater than 6 meters and less than or equal to 8 meters. In this case, the UAV actively reduces its flight speed from 1 m / s to 0.3 m / s to allow sufficient reaction time for the binocular camera to detect the obstacle and make subsequent obstacle avoidance decisions, improving system stability and obstacle avoidance accuracy. Scenario 3 indicates that the obstacle enters the obstacle avoidance zone, i.e., d is less than or equal to 6 meters. The UAV needs to adjust its flight speed, direction, and altitude to perform obstacle avoidance operations based on the obstacle type and the principle of moving away from obstacles.

[0135] In Scenario 3, if the obstacle is confirmed to be a static obstacle, the drone flies at a speed of 0.2 m / s, while adjusting its flight direction and altitude to the opposite azimuth and altitude directions to gradually move the obstacle away from the warning zone. If the obstacle is confirmed to be a dynamic obstacle, the drone increases its flight speed to be more than 0.2 m / s faster than the obstacle's relative speed, while adjusting its flight direction and altitude to the opposite azimuth and altitude directions to ensure safe and effective obstacle avoidance in a dynamic environment. Once the obstacle has successfully left the warning zone, the drone maintains a forward speed of 0.3 m / s and, based on the direction and altitude adjustments made during obstacle avoidance, corrects its flight path with opposite direction and altitude changes, thus smoothly returning the drone to its original patrol route and continuing its autonomous patrol mission.

[0136] In conjunction with the obstacle avoidance strategy described above, in this embodiment, as follows: Figure 5 The diagram illustrates the obstacle avoidance process for an autonomous UAV patrol mission. First, image preprocessing is performed on the binocular cameras mounted on the UAV. This preprocessing includes binocular camera calibration, stereo correction, stereo matching, and target disparity distance calculation based on the stereo matching results. After preprocessing, a region projection mapping is performed based on the UAV's current pose data to generate a corresponding 3D monitoring model. Subsequently, it is determined whether the UAV possesses the capability to effectively monitor and measure distances within the currently projected area. If the capability is insufficient, the image preprocessing steps are repeated to ensure the reliable operation of the perception system.

[0137] During the patrol process, the drone continuously monitors the safety zone for obstacles by measuring the distance *d* between the drone and obstacles in real time. If *d* is greater than 8 meters and less than or equal to 10 meters, a routine patrol is performed with continuous monitoring. If *d* is greater than 6 meters and less than or equal to 8 meters, the obstacle is confirmed to have entered the warning zone, and the drone automatically reduces its flight speed from 1 m / s to 0.3 m / s. If *d* is less than or equal to 6 meters, the obstacle is confirmed to be within the obstacle avoidance zone, triggering the obstacle avoidance strategy. Based on the obstacle's static or dynamic attributes, the drone adjusts its flight speed, direction, and altitude accordingly to maximize obstacle avoidance success. After completing the obstacle avoidance operation, the drone applies opposite corrective actions based on changes in the direction and altitude of the obstacle deviation, gradually returning the flight trajectory to the preset patrol route, ensuring the continuity and integrity of the overall patrol mission.

[0138] The following describes the obstacle avoidance device for unmanned aerial vehicles (UAVs) in a substation provided in the embodiments of this application. The UAV obstacle avoidance device described below and the UAV obstacle avoidance method described above can be referred to and correspond to each other. Figure 6 As shown, this application provides an obstacle avoidance device for unmanned aerial vehicles (UAVs) in a substation. The device includes:

[0139] The current distance determination module 201 is used to determine the current distance between the drone and obstacles in the substation when the drone performs its inspection mission according to the preset inspection route.

[0140] The area division module 202 is used to determine the safe zone, alarm zone and obstacle avoidance zone of the UAV's forward movement space based on the UAV's current position and flight direction;

[0141] The first distance update module 203 is used to reduce the flight speed of the drone and update the current distance if it is determined that the obstacle is located in the alarm zone based on the current distance.

[0142] The second distance update module 204 is used to adjust the flight speed and flight direction of the UAV according to the relative position of the UAV and the obstacle if it is determined that the obstacle is located in the obstacle avoidance zone based on the current distance updated after deceleration, so as to avoid the obstacle and update the current distance.

[0143] The cruise return module 205 is used to trigger the UAV to return to the patrol route if the obstacle is determined to be in the safe zone based on the updated current distance after obstacle avoidance.

[0144] In one embodiment, the current distance determination module 201 includes:

[0145] The image stereo correction unit is used to simultaneously acquire left and right view images of the drone after calibrating the binocular camera, and perform stereo correction processing on the left and right view images to make the left and right view images aligned on the same horizontal line.

[0146] The image stereo matching unit is used to perform stereo matching on the stereo-corrected left and right view images using a semi-global matching algorithm to obtain a disparity map containing obstacles.

[0147] The current distance calculation unit is used to extract the position information of the obstacle in the left and right view images based on the disparity map and the triangulation principle of binocular ranging, and to calculate the current distance between the UAV and the obstacle.

[0148] In one embodiment, the region partitioning module 202 includes:

[0149] The projection area construction unit is used to construct the forward space projection area of ​​the UAV on a two-dimensional plane, taking the current position of the UAV as a reference point and combining the flight direction of the UAV.

[0150] The area division unit is used to divide the projection area into a safe zone, an alarm zone, and an obstacle avoidance zone according to their distance from the UAV. The safe zone is located at the farthest end, the alarm zone is located in the middle, and the obstacle avoidance zone is located at the closest end.

[0151] In one embodiment, the first distance update module 203 includes:

[0152] The drone deceleration unit is used to control the drone to slow down its flight speed to a level lower than its original speed, based on the drone's current flight speed and the changing trend of the distance between the drone and obstacles.

[0153] In one embodiment, the second distance update module 204 includes:

[0154] The first obstacle avoidance unit is used to control the UAV to slow down when the obstacle is determined to be a static obstacle, and to take the azimuth and altitude directions opposite to the obstacle as the flight direction of the UAV.

[0155] The second obstacle avoidance unit is used to increase the flight speed of the UAV to a speed higher than that of the obstacle when the obstacle is determined to be a dynamic obstacle, and to take the azimuth and altitude directions opposite to those of the obstacle as the flight direction of the UAV.

[0156] In one embodiment, the second obstacle avoidance unit includes:

[0157] The first dynamic obstacle determination subunit is used to determine the obstacle based on the images containing the obstacle captured by the binocular camera at continuous time. If the position of the obstacle in the continuously captured images is detected to be shifted, the obstacle is a dynamic obstacle.

[0158] or

[0159] The second dynamic obstacle determination subunit is used to determine if the obstacle is a dynamic obstacle if the infrared feature position of the obstacle is detected to shift in the continuously acquired infrared thermal images containing the obstacle, based on the infrared dual-light camera on the UAV.

[0160] In one embodiment, the cruise return module 205 includes:

[0161] The cruise return unit is used to control the UAV to continue flying at the adjusted flight speed during obstacle avoidance, and to adjust the UAV's flight direction in the opposite direction according to the deviation direction when the UAV avoids the obstacle, so that the UAV returns to the patrol route.

[0162] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the substation drone obstacle avoidance method as described in any of the above embodiments.

[0163] In one embodiment, this application also provides a computer device storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the substation obstacle avoidance method as described in any of the above embodiments.

[0164] Indicatively, such as Figure 7 As shown, Figure 7 This is a schematic diagram of the internal structure of a computer device 300 provided in an embodiment of this application. The computer device 300 can be provided as a server. (Refer to...) Figure 7 The computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by memory 301 for storing instructions, such as application programs, that can be executed by the processing component 302. The application programs stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 302 is configured to execute instructions to perform the substation obstacle avoidance method of any of the above embodiments.

[0165] The computer device 300 may also include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate on an operating system stored in memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0166] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0167] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this document, "a," "an," "the," "the," and "its" may also include plural forms unless the context clearly indicates otherwise. "Multiple" refers to at least two, such as 2, 3, 5, or 8, etc. "And / or" includes any and all combinations of the related listed items.

[0168] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0169] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for obstacle avoidance by unmanned aerial vehicles (UAVs) within a substation, characterized in that, The method includes: When the drone in the substation performs its inspection mission according to the preset inspection route, the binocular camera on the drone is used to determine the current distance between the drone and obstacles in the substation. Based on the current position and flight direction of the UAV, determine the safe zone, warning zone, and obstacle avoidance zone of the UAV's forward movement space; If the obstacle is determined to be located in the warning zone based on the current distance, the drone's flight speed is reduced and the current distance is updated. If the obstacle is determined to be located in the obstacle avoidance zone based on the updated current distance after deceleration, the drone's flight speed and direction are adjusted according to the relative position of the drone and the obstacle to avoid the obstacle, and the current distance is updated. If the obstacle is determined to be located in the safe zone based on the updated current distance after obstacle avoidance, the UAV is triggered to return to the patrol route.

2. The obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation according to claim 1, characterized in that, The step of determining the current distance between the drone and obstacles within the substation using the binocular camera mounted on the drone includes: After calibrating the binocular camera, the left and right view images of the drone are acquired simultaneously, and stereo correction processing is performed on the left and right view images to make the left and right view images aligned on the same horizontal line. A semi-global matching algorithm is used to perform stereo matching on the left and right view images after stereo correction to obtain a disparity map containing the obstacle. Based on the disparity map and combined with the triangulation principle of binocular ranging, the position information of the obstacle in the left and right view images is extracted, and the current distance between the UAV and the obstacle is calculated.

3. The obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation according to claim 1, characterized in that, The step of determining the safe zone, warning zone, and obstacle avoidance zone of the drone's forward movement space based on the drone's current position and flight direction includes: Using the current position of the UAV as a reference point and in conjunction with the flight direction of the UAV, a forward spatial projection area of ​​the UAV is constructed on a two-dimensional plane; The projection area is divided into a safe zone, an alarm zone, and an obstacle avoidance zone according to their distance from the UAV, wherein the safe zone is located at the farthest end, the alarm zone is located in the middle, and the obstacle avoidance zone is located at the closest end.

4. The obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation according to claim 1, characterized in that, The step of reducing the flight speed of the drone includes: Based on the current flight speed of the drone and the trend of distance change between the drone and the obstacle, the drone is controlled to slow down its flight speed to a level lower than its original flight speed.

5. The obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation according to claim 1, characterized in that, The step of adjusting the flight speed and flight direction of the drone based on its relative position to the obstacle includes: When the obstacle is determined to be a static obstacle, the drone is controlled to slow down and fly, and the azimuth and altitude directions opposite to the obstacle are taken as the flight direction of the drone. When the obstacle is determined to be a dynamic obstacle, the flight speed of the UAV is increased to be higher than the moving speed of the obstacle, and the azimuth and altitude directions opposite to those of the obstacle are taken as the flight direction of the UAV.

6. The obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation according to claim 5, characterized in that, The step of determining that the obstacle is a dynamic obstacle includes: Based on the images containing the obstacle captured by the binocular camera at continuous intervals, if the position of the obstacle is detected to shift in the continuously captured images, then the obstacle is a dynamic obstacle. or Based on the infrared thermal images of the obstacle continuously collected by the infrared dual-light camera on the UAV, if the position of the infrared feature of the obstacle shifts in the continuously collected infrared thermal images, then the obstacle is a dynamic obstacle.

7. The obstacle avoidance method for unmanned aerial vehicles (UAVs) in a substation according to any one of claims 1 to 6, characterized in that, The step of triggering the UAV to return to the patrol route includes: The drone is controlled to continue flying at the adjusted flight speed during obstacle avoidance, and its flight direction is adjusted in the opposite direction according to the deviation direction when it avoids the obstacle, so that the drone returns to the patrol route.

8. A drone obstacle avoidance device for use in a substation, characterized in that, The device includes: The current distance determination module is used to determine the current distance between the drone and obstacles in the substation when the drone is performing an inspection mission according to a preset inspection route. The area division module is used to determine the safe zone, alarm zone, and obstacle avoidance zone of the UAV's forward movement space based on the UAV's current position and flight direction. The first distance update module is used to reduce the flight speed of the drone and update the current distance if it is determined that the obstacle is located in the alarm zone based on the current distance. The second distance update module is used to adjust the flight speed and flight direction of the UAV according to the relative position of the UAV and the obstacle if it is determined that the obstacle is located in the obstacle avoidance zone based on the current distance updated after deceleration, so as to avoid the obstacle and update the current distance. The cruise return module is used to trigger the UAV to return to the patrol route if it is determined that the obstacle is located in the safe zone based on the updated current distance after obstacle avoidance.

9. A storage medium, characterized in that: The storage medium stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the obstacle avoidance method for unmanned aerial vehicles in a substation as described in any one of claims 1 to 7.

10. A computer device, characterized in that, include: One or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the obstacle avoidance method for unmanned aerial vehicles in a substation as described in any one of claims 1 to 7.

Images