A Map Building And Navigation System Of Mobile Robot Based On Multi-Sensors

This thesis is focus on the Self-Localization, Obstacles Localization and Metric Map building of Autonomous Mobile Robot AS-RF in unknown environment.A new system which used auxiliary guide to help robots explore the indoor structural environment has been built. It mainly contains several parts:1^st, based on the area and the position of robot, a movement control policy is designed that using different tracks and "Rectangle Safety Area" which is divided into dangerous area, obstacle-avoided area and alertness area to estimate the position where the obstacle possibly exist and choose obstacle-avoided policy; A view point setting method is proposed, the key is corner points' angle divided line. After getting the probable position of robot by optical-electrical encoder, the guides stored in data chain can be searched, it can make facility to the following work such as self-localization and map-building of the area which should be explored. Besides, the data come from the optical-electrical encoder should be refurbished for avoiding error cumulation.2^nd, a suitable image process algorithm is proposed that involves graying, filter algorithm, edge detection and binary conversion. A new linear fitting algorithm which used histogram of gradient's direction to supervised hough transform and least-squares procedure is presented. After recognized guides in the image, biomimetic pattern recognition fusing data is proposed to compute relative angle and distance between guides and AS-RF robot.3^rd, after using self-adaptive valve value method to compart the data of laser finder, a new fitted linear compartmentation, which contains IEPF algorithm, is proposed to get segment equation, incertitude and point parameter. After coordinate conversion, segment matching and segment combination on parcel map, Kalman filter is applied to update segment estimate parameter. Through disposing segment point and intersection selectively, the segment points were updated according to projection theory. In order to complete the map building task, updating data of sensors in several observing position is important.The experiments in the indoor structural environment show that this map-building method realizes a nice accuracy, robustness and synchronous, and achieves design purpose.