R2I integrates state space models to enhance long-term memory and credit assignment in model-based reinforcement learning, achieving superior performance across diverse tasks.
