next up previous contents
Next: Vision-Based Enhanced Reality Up: Mobile Robotics Previous: Localizing a Robot

Spatial Abstraction and Mapping

This project involves the development of a formalism and methodology for making the transition from raw noisy sensor data collected by a roving robot to a map composed of object models and finally to a simple abstract map described in terms of discrete places of interest. An important early stage of such processing is the ability to select, represent and find a discrete set of places of interest or landmarks that will make up a map. Associated problems are those of using a map to accurately localize a mobile robot and generating intelligent exploration plans to verify and elaborate a map. This project considers the use of various technologies (sonar, vison, BIRIS) to perform robot localization and exploration. We have developed both geometric and topological approachs to sensor-based map construction using real data. One natural application of a map is to permit navigation and localization. Our current results with sonar-based and video based localization using autonomously generated maps combine good accuracy with a large region of convergence.
G. Dudek, P. Mackenzie, C. Zhang

Modelling of Range Data with Line Segments. Dots are sonar measurements, thick lines are inferred models, and the dotted line shows the path of the robot when scanning the environment. The map was constructed incrementally from various individual scans rather than formed from this data point set as a whole.



Thierry Baron
Mon Nov 13 10:43:02 EST 1995