Satisficing feedback strategies for local navigation of autonomous mobile robots

Abstract

A general approach to the local navigation problem for autonomous mobile robots (AMRs) is presented and its application to omnidirectional and conventionally steered wheelbases is described. The problem of driving an AMR to a goal in an unknown environment is formulated as a dynamic feedback control problem in which local feedback information is used to make steering decisions while the AMR is moving. To obtain a computationally tractable algorithm, a class of satisficing feedback strategies that generate reasonable, collision-free trajectories to the goal using simplified representations of the AMR dynamics and constrains is proposed. Realizations of the feedback strategy are presented and illustrated by simulation under the assumptions of perfect feedback information and zero servo error. Straightforward extensions of the approach to handle uncertainties in real systems are briefly described.