Obstacle Detection And Avoidance Method For Duct Cleaning Robot Combined Optical Flow And Artificial Potential Field

In recent years, the central air-condition is widely used in the buildings,cleaning the air-conditioning duct has been token more and more seriously. The people has paid more attentions to the duct cleaning robot which is the key equipment for duct cleaning. From the current research achievement of air duct cleaning robot,we can derive that the ability of autonomous navigation, motion control and duct's interior perception of duct cleaning robots is in the low level,so it is difficult to complete the cleaning task for the robots while the ducts has a closed and complex inner structure.In this paper, according to work environment of duct cleaning robot and the characteristics of all kinds of obstacles in the environment, aimed to develop the ability of obstacle avoiding of the duct cleaning robot, the approach which is combined with optical flow and artificial potential field for obstacles inspection and avoidance of the duct cleaning robot is presented,finally,the simulation and experiment demonstrate the proposed approach is feasible. To improve the work efficient of duct cleaning robot, we devote the obstacle inspection and avoidance methods of duct cleaning robot to research and realization. The contributions of this paper are given as follows:Firstly, based on introducing the work task, environment and principle of the duct cleaning robot, We analyzed the mechanical structure and control system of the duct cleaning robot, and then present the basic scheme which can realize obstacle inspection and avoidance of the duct cleaning robot.Secondly, the obstacle inspection approach based on optical flow has been researched. First of all, executing the camera shouted image by pre-processing operation such as gradation, smoothing and denouncing and so on are executed to the image which is shot by the camera, then utilizing image gradation to inspect and track the corner, using Pyramid Lucas-Kanade algorithm to compute optical flow.Experiments showed that the obstacle detection can be implemented based on optical flow balancing algorithm.Thirdly,the approach for obstacle avoidance and path planning which is combined optical flow and artificial potential field is presented. With the estimation of image sequence, we can find there exists the bigger density of optical flow in the area of the obstacle. So at first, density of optical flow is used to construct therepulsion field, the artificial potential field is added in the image sequence,and the repulsion potential field is calculated. Experiments showed that obstacle avoidance can be achieved by the algorithm and the target location navigation and obstacle avoidance are considered simultaneously.