Towards Third-Person Visual Imitation Learning Using Generative Adversarial Networks | IEEE Conference Publication | IEEE Xplore