The assignments for this class consist of two based on the Wumpus World environment, where you develop code that governs the behavior of an agent exploring a grid-based world.
In the first assignment, the task of the agent is to systematically explore the environment based on search algorithms like breadth-first, depth-first and A*.
In the second assignment, your agent needs to cope with obstacles like bottomless pits, and a smelly creature, the Wumpus, that like to eat agents. So in addition to the systematic exploration, the agent also has to develop skills that let it avoid dangers while still trying to achieve its goal.
The third assignment offers a few options. Two of them are AI competitions: The Robocode tank battle game, where you can program an agent that performs better than other agents on a tank battlefield. The second competition is the prisoner’s dilemma, where your agent’s success depends on a strategy that has to take into account the behavior of an opponent. A non-competitive version of the third assignment is the exploration of environments that may be suitable as alternatives to the Wumpus World environment or Robocode.