Environment recognition with multiple sensors and mobile robot navigation based on the fuzzy via-point selection method

Abstract

Intelligent mobile robots, the ultimate goal for researchers in the mobile robotics area, can be defined as robots that have such capabilities as perception, reasoning, decision making, action and learning. As a research work toward this ultimate goal, in this thesis, two major fields for intelligent mobile robot have been studied: environment recognition and navigation algorithm. The study is based on indoor wheeled mobile robot that moves within structured but unknown 2D environment, and the mobile robot LCAR has been used as a test bed. With the aim of more robust environment recognition, a sensor fusion scheme for mobile robot environment recognition that incorporates range data and contour data is proposed. Ultrasonic sensor provides coarse spatial description but guarantees open space (with no obstacle) within sonic cone with relatively high belief. Laser structured light system provides detailed contour description of environment but prone to light noise and is easily affected by surface reflectivity. A sensor fusion scheme that can compensate the disadvantages of both sensors has been proposed. Scanning of environments is performed by rotating two sensors installed on the pan-tilt device on the top of the robot. For the purpose of not losing spatial detail and more natural environment description, line and corner models are used. Overall fusion process is composed of two stages: Noise elimination and belief updates. Dempster-Shafer``s evidential reasoning is applied at each stage. Open space estimation from sonar range measurements brings elimination of noisy lines from laser sensor. Comparing actual sonar data to the simulated sonar data enables data of two disparate sensors to be fused at the unified feature space. Transparent objects such as glass window can be recognized by ultrasonic sensor. A hypothesis test on the RCDs (regions of constant depth) acquired from scanning of sonar in multiple places is used for recognition of lines and corners. Line...