R-AIF: Solving Sparse-Reward Robotic Tasks from Pixels with Active Inference and World Models

Open in new window