I. INTRODUCTION
This paper considers the navigation of a three degrees-of-freedom mobile robot in a planar environment populated by unknown obstacles. The robot has no apriori information about the environment, but may locally acquire this information using its on-board sensors. This class of on-line problems has a wide range of applications in unstructured environments where the robot must detect obstacles during task execution. Examples are material and mail delivery in factories and offices [1], medicine distribution in hospitals [6], horticulture duty in greenhouses [8], and planetary exploration and sample acquisition [10], [14]. Current sensor-based navigation algorithms usually assume that the robot moves with two translational degrees of freedom (two exceptions are discussed below). However, practical mobile robots move with three degrees of freedom involving translation and rotation. Since full maneuverability is often critical for task completion, there is a need to develop sensor based navigation algorithms that can plan the robot's full three degrees of freedom motions.