Fast vision-guided mobile robot navigation using neural networks

Based on observations of human navigational behaviors, the authors have designed a vision-guided mobile robot navigation system that uses both a knowledge base system for high level supervisory control, and a collection of neural networks that mimic human reactive navigational behaviors such as `go down the hallway´ or `follow the junction to the left´. They present the neural network component of the vision-guided mobile robot navigation system called NEURO-NAV and deal with how visual information is processed and how neural networks are used to perform primitive navigational tasks. Experimental results of neural classification and vision-guided navigation are included