VP-GO: A ‘Light’ Action-Conditioned Visual Prediction Model for Grasping Objects | IEEE Conference Publication | IEEE Xplore