Exploration, navigation and mapping for autonomous mobile robots

Autonomous mobile robots are usually required to navigate and perform tasks in the environments in which they operate. Optimal motion strategies can be planned to execute the required tasks if environment maps are available. Since many situations exist where maps are not readily available, map-building might be one of the tasks that the robot initially performs. A good map building procedure should attempt to maximize acquisition of environment knowledge with minimal robot motion, and recognize the state when any further motion or environment sensing by the robot will not result in any significant addition to the knowledge base (map).; Mapping algorithms are influenced by the characteristics and constraints assumed of the environment, robot sensors and the robot itself. Mathematical representations of these characteristics and constraints are called models. Three mapping procedures based on different models are presented here.; The first procedure is based on an ideal sensor, a point robot and an environment that contains only polygonal objects. The second mapping procedure assumes the environment to be a two-dimensional region populated with polygonal and curvilinear objects. The sensor model is based on a bounded range sensor capable of returning accurate distance measurements to walls and objects within a circular region surrounding a point-robot. The third and last mapping procedure presented here uses a sensor model capable of detecting the presence or absence of objects within an annular ring (proximity detector) surrounding a disc-shaped robot, which operates in an environment similar to the one used in the second procedure. All the above mapping procedures were successfully simulated using computer models.