Composable Action-Conditioned Predictors: Flexible Off-Policy Learning for Robot Navigation

Open in new window