This project utilizes Multi-Agent Spatial Synthesis (MASS) as the environment for DQN agents, 3D lattice of configuration is taken as the observation, behaviors of the agents are taken as the action space, and satisfaction of spatial agents is taken as the reward.