Repeated Rpck-Paper-Scissors (RRPS)
Solving MRRPS problem using POMCP solving algorithm from pomdp-py library (2023)
---
This is a group project entitled 'Optimising Strategies in Repeated Rock Paper
Scissors: A POMDP Exploration' that I completed with my friends
as one of the assignments for the Intelligent Robotics module at the University of Birmingham
---
Abstract
This project attempts to explore game strategy optimization by investigating the potential of Partially Observable Markov Decision Processes (POMDPs) in a modified version of the classic Rock, Paper, Scissors (RPS) game. Inspired by an old online flash game, Club Penguin Card Jitsu, the project team introduces modifications to vanilla RPS, making the game partially observable and thus suitable to model as a POMDP problem.
Our project hypothesis states that a POMDP should be able to formulate decision-making strategies that can, on average, outplay basic opponents (random decision makers, etc.). Building on existing research on Repeated Rock, Paper, Scissors (RRPS) and recent studies exploring reinforcement learning, the project aims to adapt and extend existing methodologies to our modified gaming environment.
The code for this implementation can be retrieved at: GitHub Repository
