Deriving Rewards for Reinforcement Learning from Symbolic Behaviour Descriptions of Bipedal Walking | IEEE Conference Publication | IEEE Xplore