"It is important to create an environment where students can learn the problem-solving techniques that will help them to go beyond the thoughtful use of established applications into the realm of scientific discovery." Gary Parker

• Artificial Intelligence
• Cognitive science
• Colony robotics
• Learning in autonomous robots
• Evolutionary robotics
• Genetic algorithms
• Multi-legged robots

Professor Parker's research focuses on methodologies for learning in autonomous robots. Currently he is concentrating on evolutionary robotics; applying evolutionary computation to generate control programs for small, legged robots.

He developed the cyclic genetic algorithm (a variant of the standard GA) to provide a means for learning cycles of behavior. CGAs have been tested on three different levels of control complexity (individual leg cycles, gaits, and search cycles). He also developed punctuated anytime learning, which he is using to incorporate CGAs into a real-time learning system that uses periodic tests on the robot to continually improve the quality of controller learning. This new way of integrating the actual robot with its simulation during evolutionary computation allows the system to make adjustments to the onboard controller while learning is carried out off line, giving the robot the capability to adapt to changes in real time. Additional research has resulted in a method using PAL and sampling to co-evolve cooperative individuals in a team of robots.

Future research plans include expanded experimentation with colony robotics, genetic algorithms, emergent systems, neural networks, and collaborations with other departments such as Cognitive Science/ Psychology and Biology in the study of locomotion and lower levels of animal cognition.

Professor Parker believes in incorporating research into the classroom. Since the robots and controllers used are generally inexpensive and of simple construction, students in robotics courses get hands-on experience in the development of robots and the programming of their controllers. These courses allow students to observe the results of their labors in real world robotics instead of just in simulation. Students gain valuable problem solving techniques and an appreciation for some of the difficulties involved in robotics.

Professor Parker has published over 35 papers, including “Co-Evolving Team Capture Strategies for Dissimilar Robots” at the AAAI 2004 Symposium on Artificial Multiagent Learning (2004), “Evolving Towers in a 3-Dimensional Simulated Environment” at the Congress on Evolutionary Computation (2003), “Punctuated Anytime Learning for Hexapod Gait Generation” at the IEEE/RSJ International Conference on Intelligent Robots and Systems (2002), "The Incremental Evolution of Gaits for Hexapod Robots" at the Genetic and Evolutionary Computation Conference (2001), "Co-Evolving Model Parameters for Anytime Learning in Evolutionary Robotics" in the Robotics and Autonomous Systems Journal (2000).

His paper, "Locomotion Control Cycles Adapted for Disabilities in Hexapod Robots," received the World Automation Congress Best Paper Award at the Seventh International Symposium on Robotics with Applications (1998).

He has made several presentations of his research results at international conferences and invited talks and has organized and chaired conference sessions on Evolutionary Robotics and Learning Cyclic Control/Behavior in Robots.