Control for Navigation of a Mobile Robot Using Monocular Data

This book introduces a new strategy for mobile robot navigation. The complete navigation strategy is based on local landmark detection. However, the work developed just shows the local navigation results. In this context, local artificial potential fields are used as a way to attract the mobile robot towards a local goal that can act as a passage point and a featured landmark. In order to acomplish with the desired objective, simple perception system and reactive control behaviours were implemented and tested. Concretely, a single on-robot camera system was used to infer the closer robot environment where free aproaching path was computed. Moreover, the proposed control strategy is based on on-line model predictive control techniques where only short prediction horizons are considered for dealing with reactive behaviours and dynamic environments.