| The current research in the area of automatic robot motion planning revolves around two models. In the first model, complete information about the environment is assumed, objects in the environment are to be described algebraically and the motion planning presents an off-line one-time computational task. In the second model, the information is incomplete and supplied via an on-line (say, sensory) input, objects are allowed to be of arbitrary shapes, and motion planning presents an on-line continuous process. This dissertation relates to the second model and focuses on the development of general provable strategies for collision-free motion planning for two- and three-link robot arm manipulators operating among unknown obstacles. The robot is required to move from the given start to target configurations if possible, or to conclude in finite time that the target is not reachable from the start. The accepted model assumes that the robot knows its current and target positions and is equipped with some kind of sensors capable of detecting nearby obstacles; no information about the location and geometry of the obstacles is available apriori. |
