Dynamic Path Planning For Indoor Mobile Robot

With the development of science and technology, researches about robots have become a hot spot. Robots have already been applied to many social fields. Besides, the aging of the population and the demand of intelligent household service robots make people pay more and more attention to the research of household robots which has become an important direction of robot researches. Navigation is a basic function of the robots which contains the perception of environment, localization of the robot and path planning algorithms. This paper mainly focuses on the path plan system of the indoor robot based on the global vision and the heuristic search A*algorithm.This paper firstly introduces the technologies of environmental map construction and robot localization, which are related to robot path planning. After the overview of the existing commonly used path planning algorithms, the overall design of the path planning system based on the vision is determined. In order to revise the image distortion of the camera, the linear model and nonlinear model of camera imaging are described. Planar chessboard is used for camera calibration. In terms of environmental map construction, according to the global vision images acquired by the camera, method based on Canny edge contour is adopted to the images after the pretreatment of the images. Through external contours of the obstacles, areas of obstacles are spilt out from the images. To improve the efficiency of the path planning algorithm, the environmental images are rasterized to build the global grid map of the environment. As for the tracking and localization of the robot, the background image based on codebook model is used for robot area detection to adapt to the interference of ambient light. In the process of the robot moving, this paper uses Camshift algorithm to track the position of the robot. The accuracy and real-time performance of the localization meet the needs of the system.At last, heuristic search A*algorithm is adopted to gird map for path planning. In order to adapt to the unknown environmental obstacles, a dynamic A*search method is put forward, which can update the environment grid map. Based on Qtcreator and Opencv library, an interface for simulation is built to test the results of the path planning algorithm.