Evaluating and teaching AI the nice way!
Interacting with the simulator is done through two simple ROS-Topics.
Customizable field size, number of traps, number of Wumpus, and availability of arrows for agents.
Generate random fields according to given parameters.
Its possible to spawn several agents, in case you have competing or cooperating AIs.
Be the Wumpus and hunt your agents with your own AI! A controlled wumpus can move and make the challenge for your AI even more complex.
You found an interesting situation for your AI to solve? Just store the situation including agent positions and challenge your AI over and over again.
If you use our simulator, we would be very happy to be cited in your work! Ideally, you use the following bibtex entry:
@software{JakobOpferEtAl.2018 author={Jakob;Stefan and Opfer;Stephan and Groh;Olga and Copei;Sebastian and Lichtenberg;Kevin}, title={Wumpus Simulator}, howpublished={Available on \url{www.github.com/dasys-lab/wumpus-simulator}}, version={0.1.0} date={2018-17-08} }