Reinforcement learning with prior policy guidance for motion planning of dual-arm free-floating space robot

Reinforcement learning methods as a promising technique have achieved superior results in the motion planning of free-floating space robots. However, due to increase dimension and intensification system dynamics coupling, dual-arm robots remains an open challenge. In particular, current study cannot handle task capturing non-cooperative object lack pose constraint end-effectors. To address problem, we propose novel algorithm, EfficientLPT, facilitate RL-based improve accuracy efficiently. Our core contributions are constructing mixed policy with prior knowledge guidance introducing ‖⋅‖∞ build more reasonable reward function. Furthermore, our method successfully captures rotating different spinning speeds.