E$^2$DT: Efficient and Effective Decision Transformer with Experience-Aware Sampling for Robotic Manipulation

TL;DR

E$^2$DT introduces an experience-aware sampling framework for Decision Transformers in robotic manipulation, improving sample efficiency and exploration by prioritizing diverse, high-quality experiences.

Contribution

The paper proposes a novel experience-aware sampling method using a joint kernel to enhance Decision Transformer training for robotic tasks.

Findings

E$^2$DT outperforms prior methods on robotic manipulation benchmarks.

The approach improves sample efficiency and exploration in RL.

It demonstrates robustness in both simulation and real-robot experiments.

Abstract

In reinforcement learning (RL) for robotic manipulation, the Decision Transformer (DT) has emerged as an effective framework for addressing long-horizon tasks. However, DT's performance depends heavily on the coverage of collected experiences. Without an active exploration mechanism, standard DT relies on uniform replay, which leads to poor sample efficiency, limited exploration, and reduced overall effectiveness. At the same time, while excessive exploration can help avoid local optima, it often delays policy convergence and leads to degraded efficiency. To address these limitations, we propose E$^2$DT, a DT-guided k-Determinantal Point Process sampling framework that enables the model to actively shape its own experience selection. Our framework is experience-aware, allowing E$^2$DT to be both efficient, by prioritizing sampling quality, such as high-return, high-uncertainty, and underrepresented trajectories, and effective, by ensuring diversity across trajectory windows to preserve policy optimality. Specifically, DT's internal latent embeddings measure diversity across trajectory windows, while quality is quantified through a composite metric that integrates return-to-go (RTG) quantiles, predictive uncertainty, and stage coverage based on inverse frequency. These two dimensions are integrated into a novel quality-diversity joint kernel that prioritizes the most informative experiences, thereby enabling learning that is both efficient and effective. We evaluate E$^2$DT on challenging robotic manipulation benchmarks in both simulation and real-robot settings. Results show that it consistently outperforms prior methods. These findings demonstrate that coupling policy learning with experience-aware sampling provides a principled path toward robust long-horizon robotic learning.