Neural reinforcement learning to swing-up and balance a real pole | IEEE Conference Publication | IEEE Xplore