Bioinspired Navigation for a Nonholonomic Mobile Robot

Abstract

This work addresses the problem of navigation of an autonomous flying vehicle using an optic flow sensing approach motivated by the visuomotor system of insects. Here, the state estimation and conflict resolution aspects of this problem are considered. Specifically, the observability of the robot with an arbitrary number of sensors is investigated, and the minimum number of optic flow sensors and the sensor configuration necessary to have an observable system are determined. The system dynamics linearized about a point are not observable, but a nonlinear observability analysis based on Lie derivatives indicates that the sensor measurements provide enough information for estimating the states of the system. A control policy that guarantees a conflict-free path is also described. In this work, planar obstacles are considered, such as walls. The results are demonstrated in simulation.