Mark Pendrith conducted research in reinforement learning. In 1996, he and Malcolm Ryan presented a paper at the International Conference on Machine Learning on a method that allows a six-legged robot to learn to walk. They also had another paper in the 1998 ICML.