Nonlinear control for reactive navigation of a nonholonomic robot for environmental nongradient‐based extremum seeking in maze‐like scenes

Summary A single Dubins car‐like mobile robot travels with a constant speed in a planar workspace cluttered with arbitrarily complex obstacles, possibly maze‐like ones. An unknown scalar field is defined on the plane. The sensors supply the robot with the field value at its current location and its distance to the nearest obstacle. We present a new control law that drives the robot to the location where the field attains its maximum, while avoiding collisions with the obstacles. This is justified by a mathematically rigorous global convergence result and is confirmed via computer simulations. The proposed algorithm does not use gradient estimation and is nondemanding with respect to computation and motion. Copyright © 2017 John Wiley & Sons, Ltd.