This page lists the students that are currently working with me, and some of the students who have moved on to greener pastures. If you used to work in the lab, and aren't listed on this page, I apologize (but you already know what my memory is like). Email me and I'll add you. Doctoral Students Rob Glaubius Rob is looking at various questions in the area of reinforcement learning value-function approximation. He is concentrating on problems with continuous state spaces, and his work focuses on how to choose a good approximation architecture for a given value-function approximation problem. Projects: Reinforcement Learning Control of a Small Inspector Satellite, Aircraft Transparency Inspection Fritz Heckel Fritz is looking at how we can improve human-robot interaction by building and maintaining models of the human, based on sensor data and sound psychological models. In particular, he is interested in how we can adapt formal theories of mind (in particular Baron-Cohen's and Leslie's) for practical use on a robot. Projects: Software for Robots, Human-Robot Interaction Tom Erez Tom is looking at the problem of using reinforcement learning to learn good gait controllers for robot systems with high-dimensional, continuous state and action spaces. In particular, he is interested in the use of shaping techniques to scaffold the learning process, allowing controllers for hard problems to be bootstrapped from soluitions to easier problems. Projects: Learning Effective Gaits for Complex Robots Doug Few Doug is looking at how to effectively control a mobile robot using signals recorded directly from the human brain. In particular, he is looking at how we can use shared autonomy to make up for the lack of fidelity in the control signals obtained from the cortical recordings. Projects: Brain-Computer Interfaces for Robot Control Masters Students Stu Glaser Stu is working on applying reinforcement learning techniques to the problem of controlling multiple mobile robots. Projects: Reinforcement Learning Control of a Small Inspector Satellite Eitan Marder-Eppstein Eitan is using reinforcement learning to learn good reaching and grasping for a humanoid robot, and a good gait for a spideroid robot. Projects: Learning Effective Gaits for Complex Robots Chris Wilson Chris is looking at how to take insights from acting to improve human-robot interaction, by making our robots more physically expressive. Projects: Aircraft Transparency Inspection, Human-Robot Interaction Undergraduate Students No students in database. Alumni Joe Izraelevitz Joe worked as part of a team of undergraduates to design and implement a path-following system for one of our outdoor mobile robots. The goal was to have a system that can robustly follow all of the footpaths on the Washington University campus. Projects: None in database Topher McFarland Topher was the primary developer of Action Jackson, the robot abstract artist. He also worked on novel potential field navigation techniques for mobile robots. Topher is currently pursuing a Ph.D. at Johns Hopkins University. Projects: Action Jackson Tim Blakely Tim worked in collaboration with faculty in the departments of Biomedical Engineering and Neurological Surgery to develop a direct brain-computer interface for a variety of robotic systems. His thesis looked at controlling a small robot arm using a variety of biosignals, and comparing the quality of the control to more standard methods, such as a joystick. He is currently working on a Ph.D. at the University of Washington. Projects: Brain-Computer Interfaces for Robot Control

Page written by Bill Smart.