Table 2 Chosen parameters for the RL algorithms.
From: Robot movement planning for obstacle avoidance using reinforcement learning
Parameter | Value |
|---|---|
Update frequency | Every 50 steps |
Batch size | 256 |
Replay buffer size | \(1\times 10^{5}\) |
Discount factor (\(\gamma\)) | 0.98 |
Learning rate | \(1\times 10^{-3}\) |
TD3 soft-update rate | \(5\times 10^{-3}\) |
