1CS 188: Artificial IntelligenceFall 2011Lecture 25: Conclusion12/1/2011Dan Klein – UC BerkeleyPacman Contest Challenges: Long term strategy Multiple agents Adversarial utilities Uncertainty about other agents’ positions, plans, etc.22CONTEST SLIDES HIDDEN3…and Congratulations to All! Amazing work by everyone You should all be proud of what you’ve accomplished!93Example: Starcraft10What is Starcraft?11Image from Ben Weber4Why is Starcraft Hard? Starcraft is: Adversarial Long Horizon Partially Observable Realtime Concurrent … No single algorithm (e.g. minimax) will solve it off-the-shelf12The Berkeley Overmind13Search: path planningMinimax: targetingLearning: micro controlInference: tracking unitsScheduling: resource managementHierarchical controlhttp://overmind.eecs.berkeley.edu5Search for Pathing14[Pathing]Minimax for Targeting15[Targeting]6RL for Micro Control16[RL, Potential Fields]Inference / VPI / Scouting17[Scouting, Cloaking]7Starcraft AIs: AIIDE 2010 28 Teams: international entrants, universities, research labs…18AIIDE 2010 Competition19See http://eis.ucsc.edu/StarCraftAICompetition8Pacman: Beyond Simulation?Students at Colorado University: http://pacman.elstonj.com[DEMO]Bugman? AI = Animal Intelligence? Wim van Eck at Leiden University Pacman controlled by a human Ghosts controlled by crickets Vibrations drive crickets toward or away from Pacman’s locationhttp://pong.hku.nl/~wim/bugman.htm[DEMO]219Where to go next? Congratulations, you’ve seen the basics of modern AI … and done some amazing work putting it to use! How to continue: Machine learning: cs281a / cs281b (also a 194) Cognitive modeling: cog sci 131 Vision: cs280 Robotics: cs287 NLP: cs288 Decision making: cs289 … and more; ask if you’re interested Next term: cs280 (vision) cs281b (classificiation)22That’s It! Help us out with some course evaluations Have a good break, and always maximize your expected
View Full Document