Organisation for the research project: Evaluating transfer learning for reinforcement learning agents across same-genre video games
Research project carried out in partnership between PulluP Entertainment and ISART Digital Paris
This study, explores the transfer of reinforcement learning-trained AI agents within the game engine, Unity.
Several agents were trained on two racing game prototypes that differed in physics and mechanics. The goal was to make the agents as generalists as possible. The cross-game transfer has been observed, particularly the retraining required in the process. This analysis allowed us to maximize the efficiency of our transfers.
The objective is to produce a comprehensive analysis of the transfer process, from the results to the parameters that impact its performance.
