SO-ARM: Training RL tasks in Isaac Lab