[Colloq] Dimitrios Kanoulas -- 1pm, Friday, Dec 6, 366WVH -- Curved Surface Patches for Rough Terrain Perception

[Francoise Niang]

From: Dimitrios Kanoulas



Title: Curved Surface Patches for Rough Terrain Perception
Speaker: Dimitrios Kanoulas

Date: Friday, December 6, 2013
Time: 1:00pm to 2:00pm
Location: 366WVH

Committee:
- Marsette Vona (advisor)
- Robert Platt
- Guevara Noubir
- Seungkook Yun (external member, SRI International)

Abstract:
Attaining animal-like legged locomotion on rough outdoor terrain with sparse foothold affordances---a primary use-case for legs vs other forms of locomotion---is a largely open problem.  New advancements in control and perception have enabled bipeds to walk on flat or uneven indoor environments.  But tasks that require reliable contact with unstructured world surfaces, for example walking on natural rocky terrain, may need new perception and control algorithms.  This thesis introduces 3D perception algorithms to identify, model, and perceptually validate sparse curved surface patches in the environment.  We present a bio-inspired system that uses a range sensor augmented with an inertial measurement unit for automatically finding patches intended to provide a reasonable set of choices for higher-level footfall selection.  Recordings of human subjects traversing rocky trails were analyzed to give baseline patch properties including orientation, curvature, and location.  Our system finds statistically similar patches in real time.  In current work we are integrating these sparse patch algorithms with a dense volumetric fusion of range data from a moving depth camera, with the goal of maintaining a dynamic patch map of relevant contact surfaces around a robot.  We also plan to integrate and test the algorithms as part of a real-time foothold selection system on a mini-biped robot.