Searching Unknown Environment By A Mobile Robot Using Range Sensor --an Algorithm and Experiment

This paper presents an efficient algorithm for a autonomous mobile robot to construct a two dimensional world model consists of straight line segments, using the robot's range sensor and a suitable observation sequence planned by itself. Since the environment is observed locally at any point, global geometric model (map) will be generated incrementally. In this procedure the robot observes the environment iteratively and constructs the environment map using the all sensory information of the previous observations which has been obtained up to the current observation. Then, it finds the next suitable position in the area which is recognized safe in the constructing map and move to it until to complete the map construction. To demonstrate the process of the map construction using the proposed algorithm, a range sensor with a CCD video camera and a simulator based on the real mobile robot called Yamabico M, which has been developed by the authors' group, are introduced. The range sensor yields the boundary information around the robot as a set of line segments. In order to make this information, the range sensor takes 10 images around the robot at every 36 degrees and detects the boundary between the floor and the walls in each image, then the detected boundaries in the images are mapped onto the floor surface. The experimental results indicates that the proposed algorithm works well and efficiently in the real environment considering the sensory and control errors.