The Pac-Man Projects, developed at UC Berkeley, aims to advance the field of artificial intelligence through the development and evaluation of intelligent agents in the context of the Pacman game. It serves as a playground for exploring different AI algorithms, including search algorithms, adversarial search, reinforcement learning, and probabilistic inference.

• 1-Search

  Implement and evaluate search algorithms to enable Pacman agents to navigate the game maze effectively. This includes depth-first search, breadth-first search, uniform-cost search, and A* search.

• 2-MultiAgent

  Develop intelligent agents that can compete against ghosts using adversarial search algorithms. Minimax and alpha-beta pruning techniques allow agents to make optimal decisions in a competitive environment.

• 3-ReinforcementLearning

  Employ Markov decision processes and reinforcement learning techniques, such as Value Iteration, Q-learning, and Approximate Q-learning, to enable Pacman agents to learn and improve their performance over time. Agents can adapt and make optimal decisions by maximizing cumulative rewards.

• 4-GhustBusters

  Utilize probabilistic models, such as Bayesian networks and Hidden Markov Models, to reason under uncertainty and make informed decisions in complex situations.