| With the increasing mileage of cable tunnels,the workload of cable tunnel inspection is also increasing,and there is an urgent need for a robot that can replace workers’ inspection.With the rapid development of drone technology,drones have become one of the most ideal replacements for cable tunnel inspection workers.However,the cable tunnel environment is relatively narrow and long,with dark and uneven light.Using traditional unmanned aerial vehicles directly will not be able to complete the cable tunnel inspection task due to mapping and navigation issues.Therefore,it is necessary to optimize the drone control algorithm for the cable tunnel environment in order to effectively solve the problem of autonomous inspection of cable tunnels by drones.This article takes autonomous inspection drones as the main research object.Firstly,it summarizes the research status of tunnel inspection drone technology at home and abroad,and combines the requirements of the State Grid for the inspection system to select hardware equipment models that can be suitable for this system,such as RGB D cameras,airborne computers,drone airframes,and flight controllers,A set of autonomous patrol drones was designed to obtain cable tunnel images and toxic and harmful gas concentration data.In order to solve the problem of cable tunnel mapping and unmanned aerial vehicle pose estimation,the traditional RGB D visual SLAM algorithm needs to be optimized for the cable tunnel environment.Firstly,by selecting an ORB feature point extraction algorithm with better real-time performance,a grid homogenization method is used to solve the problem of uneven distribution of feature points in the image,resulting in unstable operation of unmanned aerial vehicles in cable tunnel environments,and easy loss of current location.Then,the ORB feature point matching algorithm is improved using a cross filtering method to eliminate some obviously incorrect feature point pairs,avoiding wasting a lot of time conducting meaningless calculations when using the RANSAC algorithm iteration,thereby further improving the efficiency of the visual SLAM algorithm and the response speed of the drone.Thirdly,the IRRT * navigation planning algorithm was selected for the global planning of patrol drones,but this algorithm was not optimized for the narrow and long environment of cable tunnels.Because tunnels only move in one direction,the artificial potential field APF algorithm can be introduced to control the growth direction of random trees in the rapid expansion algorithm,guiding the random trees to expand directly from the starting point to the target point and from the target point to the starting point,The process of surrounding exploration in the IRRT * navigation planning algorithm to determine the expansion direction of a random tree is removed,reducing the number of curved routes in the planning path,and avoiding the drone hitting a wall due to turning.Therefore,APF-IRT * not only improves the real-time performance of navigation algorithms,but also improves the safety of unmanned aerial vehicles.Finally,experimental platform construction and experimental verification were carried out.Firstly,the construction of autonomous patrol UAV system and the design of system software scheme are carried out.Then,the fixed-point hover experiment and the flight trajectory accuracy evaluation experiment of UAV are designed.Finally,the map construction and autonomous patrol experiment of UAV in the cable tunnel environment are designed to verify the feasibility of the method in this paper. |
PDF Full Text Request