Final report ”Establishing bonds for the advancement of the Rescue League”

The RoboCup Rescue Simulation League aims to benchmark the intelligence of software agents and robots on their capabilities to make autonomously the right decisions in a disaster response scenario. The UvA Rescue and Distribot team have come together to work cooperatively to support this aim in two ways. One aspect is to advance the popularity of this League by presenting the background of the scientific challenges outside the direct RoboCup community. Another aspect is to evaluate the effect of several state-of-the-art machine learning techniques, such as reinforcement learning, evolutionary computing and Markov decision processes.