Purdue Experiments in Model-based Vision for Hallway Navigation

In the Robot Vision Laboratory of Purdue University , we have developed two robust architectures for vision-guided mobile robot navigation in indoor environments. The rst approach, FINALE, enables the robot to autonomously navigate in hallways at speeds around 8m=min through vision-based position-updating, utilizing a geometrical map, model-based reasoning and Kalman ltering. The performance of the navigation is not impaired by the presence of stationary or moving obstacles. The second approach, FUZZY-NAV , performs vision-based navigation in a more human-friendly manner, utilizing simply a topo-logical map and the combination of neural networks and fuzzy logic. This approach also allows the robot to simultaneously navigate and avoid obstacles, both static and dynamic. In this paper, we describe the above two approaches and show experimental results to demonstrate the eeectiveness and the robustness of our approaches.