Package Summary

oprc_env is a Haskell package which provides a simulator for Kevin Bradner's M.S. thesis at Case Western Reserve University, under the supervision of Prof. Wyatt Newman. The goal of the thesis is to develop a reinforcement learning algorithm which learns efficient policies for multi-agent robotic coverage of uncertain environments.

oprc_env provides a specific reinforcement learning environment in which to train such an algorithm. Briefly, this environment is a two dimensional region containing patches of land which must each be observed in adequate detail in order for the task to end in success. A team of drones must coordinate their paths and altitudes such that the environment is covered in as little time as possible.

The primary intended use of oprc_env is for development of reinforcement learning agents, and there is a particular emphasis on compatibility with the OpenAI gym framework for reinforcement learning. To this end, there is another page (work in progress) on this wiki which specifies the API exposed by this package for communication between the environment and the agents which explore it.