---

Research Interests: Robotics

1. Adaptive behavior in groups of intelligent agents is an important and cross-cutting topic that appears in many fields, including biology, computer
science, economics, entomology, evolution, and social systems. Individual agents interact with the environment and with one another, modifying their behaviors according to information received through those interactions.
While agents may attempt to achieve local goals that are to their own benefit, collective behaviors emerge beyond the individual agent's cognitive or perceptual capabilities. It is highly desirable to engineer local rules which result in achieving a global goal as specified by the system designer. This, however, requires understanding multiagent adaptation.
To that end, two major challenges require solving: (i) modeling single-agent learning and decision-making under uncertainty in dynamic environments which are altered by the agents, and (ii) understanding the emergence of collective behavior through the dynamics of interaction. Our approach is to recognize and utilize common principles, based on information theory.

S. Still. Statistical Mechanics approach to interactive learning. 2005/2007(revised). http://arxiv.org/abs/0709.1948


2. Walking gait control is a formidable example of adaptive behavior in a single agent. The neuronal hardware which controls movement patterns in animals is often modeled by  coupled  oscillator networks. Those can be used to control robotic movement. Robots can learn their movement behavior via self-adaptation by learning the control parameters of the oscillator networks.

S. Still, K. Hepp, and R. J. Douglas. Neuromorphic Walking Gait Control. IEEE Transactions on Neural Networks, Volume 17, Issue 2, pp. 496 - 508. 2006.

S. Still, B. Schölkopf, K. Hepp, and R. J. Douglas. Four-legged Walking Gait Control Using a Neuromorphic Chip Interfaced to a Support Vector Learning Algorithm. In Todd K. Leen, Thomas G. Dietterich, and Volker Tresp, editors, Advances in Neural Information Processing Systems 13, pp. 741--747, 2001. MIT Press.

S. Still and M. W. Tilden. Controller for a four legged walking machine. In L. S. Smith and A. Hamilton, editors, Neuromorphic Systems: Engineering Silicon from Neurobiology. World Scientific, 1998.

S. Still. Walking gait control for four-legged robots. ETH (Swiss Federal Institute of Technology) Zuerich, Department of Physics, 2000.


Student Projects (499, 699) and thesis projects:

1. Multi-agent systems
            Study adaptive behavior in groups of intelligent agents.

2. Walking robots
            Movement control and co-ordination for legs and a spine.


Facilities: We have created a new robotics lab in the department. Talented students would also have the opportunity to work with the multi-agent robot lab at UC Davis.

---