A gross sensor-based navigation algorithm for a mobile robot with self-positioning, sensing, control errors

In sensor-based navigation, a mobile robot arrives at its goal if, and only if, it never joins any deadlock, i.e., any periodic cycle, otherwise, the robot recognizes no deadlock-free path in an uncertain world. The authors call a set of alternative terminations the deadlock-free characteristic. In general, every previous sensor-based navigation algorithm keep the deadlock-free characteristic under the following assumptions: (1) a mobile robot traces an uncertain obstacle faithfully; and (2) a mobile robot identifies its position and orientation exactly. However, a mobile robot unfortunately cannot maintain these assumptions because of three kinds of error, i.e., self-positioning, sensing, and control errors. Therefore in this article, after eliminating these two assumptions, the authors design a new sensor-based navigation algorithm NC-Robust* in order for a mobile robot with these errors not to join any dead-lock in an uncertain world. Then they give a proof of the deadlock-free characteristic and ascertain the feasibility of the algorithm NC-Robust* by using a real mobile robot (Nomad) running in a practical world environment