Minimal Sensing Structures for Designing Robot Motions

Steve LaValle
University of Illinois at Urbana-Champaign
Computer Science

This talk will provide an overview of our approach to robotics, in which we consider how to get robots to accomplish tasks while using the weakest sensors possible. Questions considered include: What information is needed to enable robots to find and execute geodesic trajectories in nonconvex environments? What sensors are sufficient to infer the arrangement of obstacles up to the level of oriented matroids? The combinatorial structures obtained in this quest may yield insights into the design of mapping, filtering, and navigation techniques in more complicated, data-intensive scenarios, as in the recent work of Landa, Tsai, and Cheng.

Presentation (PDF File)

Back to Workshop II: Numerics and Dynamics for Optimal Transport