Decentralized multi-agent path planning via immitation learning. This repository contains work done as part of a research project into multi-agent systems and decentralized path planning. We show that DMPP tasks can be solved effectively by simple (i.e. linear) models via immitation learning, whereas previous literature on the subject uses very complex ones.