Policy gradient reinforcement learning for fast quadrupedal locomotion | IEEE Conference Publication | IEEE Xplore