Deep Learning in Evolutionary Robotics
The project aims at connecting the promising results achieved recently in Deep Learning with the complementary view on AI given by the theory of embodiment in Evolutionary Robotics. The project's goals are to establish a theoretical DL framework for ER, to study the requirements of objective quality metrics (benchmarks etc., particularly to measure "complexity") in this specific context, and to eventually evolve behaviors of higher complexity than the state of the art in ER.
von: 1 Januar 2017
bis: 30 März 2018