Teaching the evolution of behavior with SuperDuperWalker

SuperDuperWalker is a software-based framework for experiments on the evolution of locomotion. It simulates the behavior of evolving agents in a 3D physical simulation environment and displays this behavior graphically in real time. A genetic algorithm controls the evolution of the agents. Students manipulate parameters with a graphical user interface and plot outputs using standard utilities. The software supports an inquiry cycle that has been piloted in CS193T: Biocomputational Developmental Ecology at Hampshire College. Keywords. Physical simulation, genetic algorithms, biology, The science curriculum at Hampshire College2 emphasizes original student inquiry at all levels, including first-semester courses and courses for non-majors. This puts a premium on tools and methodologies that allow for genuine inquiry by novices. As a result, Hampshire College faculty have developed a range of methodologies and technologies for student-active inquiry-based science education, and they have ...

• Type: Conference
• Year: 2005
• Where: AIED
• Authors: Lee Spector, Jon Klein, Kyle Harrington, Raymond Coppinger
• Keywords: AIED 2005, Artificial Intelligence, Genetic Algorithms, Graphical User Interfaces, Physical Simulations