This is the repository for the NeurIPS accepted paper:
A perspective on multi-agent communication for information fusion-> https://arxiv.org/abs/1911.03743
We have proposed a new algorithm for applying multi agent reinforcement learning to learn communication strategies in a goal driven fashion for sensor fusion. New multi agent sensor fusion environments as described in the paper are present in the Environments/ folder Application: Multi agent robots, Sensor fusion