IPD: Boosting Sequential Policy with Imaginary Planning Distillation in Offline Reinforcement Learning
This work addresses the problem of improving the efficacy of decision transformer-based sequential policies in offline reinforcement learning, which are constrained by static datasets and architectural limitations, for researchers and practitioners in RL.
The authors introduce Imaginary Planning Distillation (IPD), a framework that integrates offline planning into data generation, supervised training, and online inference for sequential policies in offline reinforcement learning. IPD learns a world model and value function to identify suboptimal trajectories, augmenting them with imagined optimal rollouts via Model Predictive Control. This enriched dataset is then used to train a Transformer-based policy, achieving significant performance improvements over state-of-the-art methods on the D4RL benchmark.
Decision transformer based sequential policies have emerged as a powerful paradigm in offline reinforcement learning (RL), yet their efficacy remains constrained by the quality of static datasets and inherent architectural limitations. Specifically, these models often struggle to effectively integrate suboptimal experiences and fail to explicitly plan for an optimal policy. To bridge this gap, we propose \textbf{Imaginary Planning Distillation (IPD)}, a novel framework that seamlessly incorporates offline planning into data generation, supervised training, and online inference. Our framework first learns a world model equipped with uncertainty measures and a quasi-optimal value function from the offline data. These components are utilized to identify suboptimal trajectories and augment them with reliable, imagined optimal rollouts generated via Model Predictive Control (MPC). A Transformer-based sequential policy is then trained on this enriched dataset, complemented by a value-guided objective that promotes the distillation of the optimal policy. By replacing the conventional, manually-tuned return-to-go with the learned quasi-optimal value function, IPD improves both decision-making stability and performance during inference. Empirical evaluations on the D4RL benchmark demonstrate that IPD significantly outperforms several state-of-the-art value-based and transformer-based offline RL methods across diverse tasks.