Applying Reinforcement Learning to RTS Games
Allan Mørk Christensen, Martin Midtgaard, Jeppe Ravn Christiansen, Lars Vinther · 2010
Real-Time Strategy (RTS) games are challenging domains for AI, since it involves not only a large state space, but also dynamic actions that agents execute concurrently. This problem cannot be optimally solved through general Qlearning techniques, so we propose a solution using a Semi Markov Decision Process (SMDP). We present a time-based reward shaping technique, TRS, to speed up the learning process in reinforcement learning. Especially, we show that our technique preserves the solution optimality for some SMDP problems. We evaluate the performance of our method in the Spring game Balanced Annihilation, and provide some benchmarks showing the performance of our approach.