System Design And Detection Algorithm For High Voltage Line Inspection Robot

Using robots to finish inspections of transmission lines is an important development direction in the field of special robots,which can improve the automatic management level of overhead transmission lines and reduce the risk of safety accidents.The research in this field also promotes the development of robot control,localization and mapping,object detection and electromagnetic shielding technology.The basic design requirement of this reaserch is to make the robot walk stably and span obstacles safely on line.Whereas the core target is to detect the possible fracture,damage and other faults of the lines and armour clamps.Besides,transmission line detection robot need to find obstacles in front and get their distance.The existing line inspection robot system is mainly used for ground line patrol,which has poor line adaptability and limited obstacle-crossing ability.And the traditional on-line fault detection methods are mostly using ultrasonic and eddy current sensing devices which have problems of low efficiency and poor accuracy.This project will major in transmission line robot system design and on-line object detection.At the same time,a prototype inspection robot has been built to carry out the experimental verification.According to the requirements of high voltage line inspection task,we adopt a six arm structure scheme in which the robot is driven by six action wheels and hangs under the transmission lines.Based on the structural characteristics of the robot,obstacle crossing schemes are presented and analyzed in our project.We also put forward a kind of power take-off mechanism.The paper analyzes the force state of the robot and the data requirements in different environments,which helps finish the design of control and sensor systems.When walking along the transmission lines,the robot should recognize the armour clamps in front such as space dampers,vibration dampers and suspension clamps,judge the type and calculate motion parameters.According to the requirements of recognition of the obstacles on-line,an armour clamp recognition algorithm based on structural constraint is proposed.We construct special descriptors to find feature points on the centerline,then extract ROI areas and calculate the HOG characteristics.Train SVMs to judge the type of armour clamp in ROI areas.Then a recognition acceration method is put forward based on prior information in consecutive frames.According to the task requirements of robot acquiring the posture of its relative front obstacles in local environment.We present a 3D-reconstruction method of transmission lines based on data accumulation of Lidar sensor.Then an algorithm of obstacle crossing location based on Improved Kalman filter is proposed:Segment the point cloud datas acquired by lidar,extract the obstacle part and match it with the database,find out the robot position distribution satisfying the maximum posterior condition.We also present a global environment driving path location method based on the combination of wheel odometer and IMU inclination informationAccording to the requirements of online fault detection,an adaptive homomorphic filter is used to enhance contrast,reduce the influence of irradiation component of the image and sharpen the edge.The methods mentioned above are tested in transmission line environment in the laboratory.