A SkillsCast for this session is not available.
With the widespread use of 3D printing and laser cutting technology, the price of manufacturing dropped considerably in the past years. Naturally, this development extends also to the manufacturing of robots which are becoming easier to build, prototype and access for everyone from year to year. This talk presents a series of experiments in which we applied reinforcement learning on robots made from paper to learn efficient locomotion strategies on granular media. We will cover the manufacturing process, why this type of robots is a good application for reinforcement learning algorithms and vice versa, some pitfalls of reinforcement learning on real robots, and what the presented robot has to do with land mine clearing and sea turtles.
YOU MAY ALSO LIKE:
Paper Robots and Reinforcement Learning: From the Lab to the Desert
Kevin S. Luck is a PhD candidate at the Interactive Robotics Lab in Arizona ,USA, and received his B.Sc. and M.Sc. from the Technische Universitaet Darmstadt, Germany. His main research interests are reinforcement learning and robotics, with a focus on sample efficient learning in the real world. In the past, Kevin has worked as a research intern at Facebook AI Research in Menlo Park, US, at DeepMind in London, UK, and at the Intelligent Robotics Lab at Aalto University in Finland.