Boosting Efficient Reinforcement Learning for Vision-and-Language Navigation With Open-Sourced LLM | IEEE Journals & Magazine | IEEE Xplore