[2409.16208] Context-Based Meta Reinforcement Learning for Robust and Adaptable Peg-in-Hole Assembly Tasks