Non-Markovian Long-Horizon Robot Manipulation via Keyframe Chaining
This addresses a key limitation in robot manipulation for tasks requiring long-term memory, though it appears incremental as it builds on existing VLA frameworks with specific enhancements.
The paper tackles the problem of Vision-Language-Action models struggling with long-horizon robot manipulation tasks due to Non-Markovian dependencies, introducing Keyframe-Chaining VLA to extract and link key historical frames, which achieves superior performance on a suite of four Non-Markovian tasks.
Existing Vision-Language-Action (VLA) models often struggle to generalize to long-horizon tasks due to their heavy reliance on immediate observations. While recent studies incorporate retrieval mechanisms or extend context windows to handle procedural tasks, they often struggle to capture Non-Markovian dependencies, where optimal actions rely solely on specific past states rather than the current observation. To address this, we introduce Keyframe-Chaining VLA, a framework that extracts and links key historical frames to model long-horizon dependencies. Specifically, we propose an automatic keyframe selector that learns a discriminative embedding space, effectively identifying distinct state transitions. To capture task-critical information, we design a progress-aware query mechanism that dynamically retrieves historical frames based on their temporal relevance to the current execution phase. These selected keyframes are integrated into the VLA as interleaved visual tokens, explicitly grounding the policy in the long-horizon temporal context. Finally, we introduce a suite of four Non-Markovian manipulation tasks built upon the ManiSkill simulator to measure task success rates. Experimental results demonstrate that our method achieves superior performance, effectively tackling robot manipulation tasks characterized by long-horizon temporal dependencies. Code is available at https://github.com/cytoplastm/KC-VLA.