Integrating Reinforcement Learning with Foundation Models for Autonomous Robotics: Methods and Perspectives