RDD: Retrieval-Based Demonstration Decomposer for Planner Alignment in Long-Horizon Tasks
This addresses the issue of degraded performance in complex manipulation tasks due to misaligned sub-task decompositions, though it is incremental as it builds on existing hierarchical frameworks.
The paper tackles the problem of aligning sub-task decompositions in hierarchical vision-language-action frameworks for long-horizon tasks by proposing RDD, which automatically decomposes demonstrations to match low-level policy training data, outperforming state-of-the-art methods in simulation and real-world tasks.
To tackle long-horizon tasks, recent hierarchical vision-language-action (VLAs) frameworks employ vision-language model (VLM)-based planners to decompose complex manipulation tasks into simpler sub-tasks that low-level visuomotor policies can easily handle. Typically, the VLM planner is finetuned to learn to decompose a target task. This finetuning requires target task demonstrations segmented into sub-tasks by either human annotation or heuristic rules. However, the heuristic subtasks can deviate significantly from the training data of the visuomotor policy, which degrades task performance. To address these issues, we propose a Retrieval-based Demonstration Decomposer (RDD) that automatically decomposes demonstrations into sub-tasks by aligning the visual features of the decomposed sub-task intervals with those from the training data of the low-level visuomotor policies. Our method outperforms the state-of-the-art sub-task decomposer on both simulation and real-world tasks, demonstrating robustness across diverse settings. Code and more results are available at rdd-neurips.github.io.