Research of Benjamin D. Stewart

My research has centered around probabilistic methods for Mobile Robotics. From July of 2002 to January 2004 I worked on the DARPA funded project Centibots. The University of Washington and SRI International collaborated on the project. The goal was to have 100 robots autonomously explore and map a large indoor environment as well as perform surveillance and people tracking tasks in a distributed fashion. The system was highly multi-threaded and written in C++. The robotics research group at UW worked on coordinating teams of robots to explore and map indoor environments under the assumption of limited wireless communication. During the course of exploration, the robots must determine their relative locations in order to combine maps and explore in a coordinated fashion. My work involved developing probabilistic techniques for a robot to determine the locations of the other robots in the system using a partial map of the environment and a stochastic model of the environment's structure. Please see the publications section for papers and abstracts published as part of this work.

Currently, I am interested in applying probabilistic techniques for Location-Aware Ubiquitous Computing applications. Additionally, I see great potential for using Hierarchical Bayestain statistical methods in modeling human behavior. Things I would like to explore modeling are preferences (music, books, etc.) and user-interface interactions.

Mult-robot active colocation in action.

Click Here to download Windows AVI

Ann and Valerie set up to explore and map the Allen Center at UW.

 

Publications