2019 – TDP – Receptivity – RoboCupSoccer – Simulation – 2D

Receptivity is a new team for Robocup 2019 in the 2D sim- ulation league. It is based on agent2d-3.1.1 and the recently released Gliders2d. The strategic focus of the team is on improving the evaluation function with machine learning and reinforcement learning techniques. In particular, ne tuning of the agent decision can be made by identify- ing situations where the intended action was not successfully completed and instead led to negative outcomes such as loss of ball possession. This paper outlines the process of learning and evaluation for decision ne tuning.