Learning Modes for Sequential Decision Making Using Stochastic Search
The objective of this project is to design and implement a stochastic search algorithm to allow an agent to explore a given state space. The state space is given in the Planning Domain Definition Language (PDDL) format containing potential actions and their preconditions and effects. The agent attempts to create a plan for a randomly generated problem by stepping through the state space testing potential plans and updating reward values and parameters based on the validity of the test. Once the agent has solved the problem, the valid plan is returned.