Science Robotics

Supplementary Materials

The PDF file includes:

  • Section S1. Nomenclature
  • Section S2. Implementation details
  • Section S3. Foot trajectory generator
  • Section S4. Reward function for teacher policy training
  • Section S5. Parameterized terrains
  • Section S6. Qualitative evaluation of the adaptive terrain curriculum
  • Section S7. Reconstruction of the privileged information in different situations
  • Section S8. Recurrent neural network student policy
  • Section S9. Ablation of the latent representation loss for student training
  • Fig. S1. Illustration of the adaptive curriculum.
  • Fig. S2. Reconstructed privileged information in different situations.
  • Fig. S3. Comparison of neural network architectures for the proprioceptive controller.
  • Table S1. Computation time for training.
  • Table S2. Parameter spaces C for simulated terrains.
  • Table S3. Hyperparameters for automatic terrain curriculum.
  • Table S4. State representation for proprioceptive controller and the privileged information.
  • Table S5. Neural network architectures.
  • Table S6. Network parameter settings and the training time for student policies.
  • Table S7. Hyperparameters for teacher policy training.
  • Table S8. Hyperparameters for student policy training.
  • Table S9. Hyperparameters for decoder training.
  • Algorithm S1. Teacher training with automatic terrain curriculum.

Download PDF

Other Supplementary Material for this manuscript includes the following:

  • Movie S1 (.mp4 format). Deployment in a forest.
  • Movie S2 (.mp4 format). Locomotion over unstable debris.
  • Movie S3 (.mp4 format). Step experiment.
  • Movie S4 (.mp4 format). Payload experiment.
  • Movie S5 (.mp4 format). Foot slippage experiment.

Files in this Data Supplement: