Table 2 List of hyperparameters used in each experiment.
From: Exploration-based model learning with self-attention for risk-sensitive robot control
Parameter | Simulation | Soft robot | Manipulator | Quadruped |
|---|---|---|---|---|
Initial safety factor λ0 | 2.0 | 2.0 | 3.5 | 2.0 |
Number of center (Eq. (4)) | 25 | 30 | 30 | 30 |
Learning rate | 5e−3 | 5e−3 | 5e−3 | 3e−3 |
Batch size | 32 | 6 | 32 | 6 |
Attention matrix size | 20 | 15 | 10 | 20 |
Leveraging factor γ | −3 | −3 | −3 | −3 |
Average period of ϵ (Eq. (27)) | 1500 ms | 10 s | – | 10 s |
Sliding feedback gain | 0.35 | 0.5 | 0.5 | 0.35 |
Buffer size | 6000 | 6000 | 2000 | 5000 |
