A Quadruped Inspection Robot Under Complex Terrain

With the continuous development of society and the continuous advancement of urbanization,more and more communication base stations,transmission lines,towers,meteorological observation stations,new energy(photovoltaic,wind energy)power generation facilities and other infrastructure have been established,and the related detection and maintenance work has followed.However,due to the limitation of use and geographical environment,especially in the mountainous and hilly areas of Guizhou,which accounts for 92.5%,if we will develop new energy including photovoltaic in Guizhou,it is necessary to build relevant power facilities which are usually located on the open mountain top or hillside,far away from the highway,in such a special geographical environment.Both inspection and maintenance need human and animal resources.In this project,a quadruped inspection robot based on complex terrain and environment is proposed and designed in view of the factors such as the inconvenience of wheeled and tracked vehicles and robots in complex terrain,the instability of air flow at high altitude,the complex electromagnetic environment near the transmission line and the difficulty of unmanned aerial vehicle operation.The mechanical structure of the robot is designed by Solid Works software,and each part is printed by 3D printer and assembled by manual operation.In the circuit hardware,the microcontroller unit of this robot is connected with wireless communication module,gyroscope module,image recognition module,bus steering-engine-control module by serial port expansion module.This control module controls the steering engines of 12 joints through the bus to drive the whole quadruped robot.In mathematical modeling,in order to control the robot,this paper uses DH matrix to establish the forward kinematics model of the robot's single leg,and obtains the numerical solution of the inverse kinematics model of the robot's single leg by iterative method.On this basis,the robot leg can fit various trajectories in space.In gait planning,in order to enable the robot to walk,this paper makes the single leg fitting trajectory into five support dispersion points and two swing dispersion points,and uses the improved Q-learning algorithm to search 24 effective combinations from 625 combinations of quadruped support phases as the robot's gait.At the same time,according to the zero moment point(ZMP)control principle,the offset correction is introduced to extend the original 16 gait to 20 gait,which ensures the stability of the robot in the swing phase.The basic function of robot walking and rotating is realized.And then,with the color recognition function of Open MV and the principle of optical imaging,the recognition of the target object is realized,and the distance between the recognized object and the robot is measured.Based on this data,combined with the gyroscope return parameters,the robot is controlled to search,track and close to the target object.In order to realize the inspection work of robot in complex terrain,this paper improves the previous ant colony algorithm to make the path planned by the new algorithm more practical.Moreover,based on the theoretical modeling,this paper designs an algorithm,which enables the robot to adjust its posture at any time for adapting to its own environmentThrough the prototype test,it is verified that the robot can walk and turn on the relatively flat mountain slope.At the same time,in the case of unmanned control,the robot can realize the function of simple search and recognition of target objects,which makes it have higher practical value in the field environment.