Inteligent control system with reinforcement learning for solving video game tasks

The object of research is intelligent control systems of virtual agents in video games. The purpose of the study is to improve the efficiency of solving the problem of building intelligent agents using neural networks and reinforcement learning strategies for playing video games. To achieve the goal, a neural network model based on a modified transformer and two fully connected neural networks is proposed to effectively solve reinforcement learning video game tasks. It has been demonstrated in the Battle City video game environment that careful design of the state functions can produce much better results without changes to the reinforcement learning algorithm, significantly speed up learning, and enable the agent to generalize and solve previously unknown levels. Ref. 25, pic. 5, tabl. 2