Abstract
We present a method of route selection for a vision guided mobile robot with art odometry system, given an uncertain map. The robot first generates candidates of the routes, which are expanded to possible loci in consideration of the map uncertainty. It then makes an observation and movement plan along each locus and computes the minimum traveling time. Uncertainties of movement and visual sensing are considered in this step. Finally, for each candidate route, the robot computes its expected traveling time from probabilities of loci It selects the route with the least expected traveling time as the optimal route. The advantage of the planning method is that the observation and movement planning are incorporated in path planning to obtain the optimal plan. The three sources of uncertainty are considered in a unified manner in our method. The method has been implemented and tested to validate our claim.
