Improved Belgian AI Algorithm for Dynamic Management in Action Role-Playing Games

Artificial intelligence in games is one of the most challenging tasks in academia and industry. In action role-playing games, how to manage combat effectively is a key issue related to game development and the player’s experience. The Belgian artificial intelligence (BAI) algorithm is a classic but limited method that is widely used for combat management between the player and enemies. To address the poor adaptability of BAI, this paper proposes an improved Belgian artificial intelligence (IBAI) algorithm with dynamic difficulty adjustment (DDA) and implements two systems separately based on BAI and IBAI in Unreal Engine 4. Advantages on 12 parameters—10.086 mean total score greater, and 0.079 standard deviation smaller—demonstrate that the system based on IBAI has higher adaptability and a better player experience by comparing the two systems in different situations and inviting players to participate in gameplay experiences and questionnaires. The robust dynamic management mechanism of IBAI can help game designers and developers achieve the combat system of action role-playing games more efficiently, thus, shortening the development cycle and improving the player retention rate.