Supplementary material

The videos reported here show the progression of the learned policies at different episodes. We have selected three goals. The videos show the outcome of the experiments for each of the considered agents:

  • BL: Baseline (TD3)
  • BL+DM: Baseline + Difficulty Manager
  • BL+EP: Baseline + Episodic Noise
  • OURS: Baseline + Difficulty Manager + Episodic noise

Notice that the goal is represented with a green arrow (the orientation of the robot is indicated by the red axis).

Episode 135

Episode 605