AMAF-CD: A Game-Changing Approach for Optimizing Squad Movement in StarCraft

In the realm of real-time strategy gaming, StarCraft has long been a testing ground for evaluating the capabilities of AI algorithms. This paper presents a significant advancement in AI strategy by proposing All-Moves-As-First considering duration (AMAF-CD) algorithm that incorporates the AMAF heuristic into the MCTS framework, exemplified by the StarAlgo project. Originating from the sphere of computer Go, the AMAF heuristic signifies a paradigm shift in decision-making strategies by treating all potential moves with equal promise. This approach effectively facilitates comprehensive exploration of decision spaces. By integrating systematic node expansion and dynamic scoring mechanisms, this strategy expedites AI learning process, particularly within intricate environments, such as StarCraft. The proposed AMAF-CD algorithm consistently outperforms existing state-of-the-art methods, achieving an overall win ratio of 58.6% in contrast to 48.2% for MCTS-CD and 24.6% for Negamax algorithms. The results demonstrate a significant shift, converting previously challenging scenarios, which historically resulted in losses, into decisive victories, highlighting the proposed algorithm’s effectiveness. Beyond gaming, this research offers insights into advanced AI heuristics and their practical applications in rapid and complex decision-making scenarios.