Improving Multi-Layer Decision Making in Simultaneous Strategy Games with Improved Command and Optimization Algorithm Method and Tactical Adaptive Control

Real-time strategy (RTS) games, with their complex dynamics and requirement for simultaneous decisions, provide a rich testbed for artificial intelligence research. This paper proposes a hybrid model to enhance intelligent agent performance in RTS environments, combining a hierarchical structure for strategic decision-making with adaptive algorithms for tactical control. At the strategic level, a hierarchical task network evaluates possible paths for base and army development, considering environmental state, resources, and objectives. At the tactical level, battle management and unit deployment leverage Puppet Search and Monte Carlo Tree Search algorithms with adaptive control. An improved command and optimization method, extending the BOSS algorithm, integrates strategic hierarchies, tactical adaptive algorithms, and reactive control (ABCD and UCT-CD) to support multi-layer decision-making. Low-level control employs ABCD and adaptive Monte Carlo algorithms to manage unit movements precisely. The proposed model was evaluated in StarCraft scenarios and compared with BWSAL, BOSS, and micro-based models. Results indicate superior performance in execution speed, accuracy, and tactical coordination, demonstrating suitability for real-time applications. Beyond gaming, these findings offer insights for designing multi-agent decision-support systems in military simulations, educational platforms, and intelligent gaming environments.