A Pedagogical Framework for Modeling and Simulating Intelligent Agents and Control Systems Dan Tappan Assistant Professor, Computer Science Idaho State University 1 AAAI-08
Motivation ● Student programming assignments – too much troublesome administrative overhead – effort directed toward programming, not AI – non-reusable hacks ● Student-friendly AI modeling and simulation framework – manages components, agents, graphics, logging, etc. – facilitates controlled experiments – promotes meaningful analysis 2 AAAI-08
Environment ● 3D tabletop world of arbitrary scale y z x north west e a s t south 3 AAAI-08
Components ● Hierarchical building blocks / infrastructure for agents – input/processing/output functionality y – tedious, error-prone engineering aspects z ● mount points azimuth elevation ● degrees of freedom x ● speed, acceleration, timing – enforces students’ rules and constraints y yaw z pitch x roll 4 AAAI-08
Agents ● Coordinated components in task environment – AI core – command and control – intercommunication 5 AAAI-08
Simulation and Analysis ● Monte Carlo simulation – controlled experiments – automatic data collection mommy baby – manual, external analysis the herd ● Example scenarios – nature ● herding ● flocking ● following leader the herd ● predator/prey – engineering ● control systems ● robotics ● battle engagements 6 AAAI-08
Questions? Dan Tappan tappdan@isu.edu / www.isu.edu/~tappdan 7 AAAI-08
Recommend
More recommend