Search papers, labs, and topics across Lattice.
This paper introduces ChunkFlow, a novel training-and-execution framework designed to enhance the performance of vision-language action models that utilize chunked action heads. By addressing boundary jitter through a seam-aware approach that incorporates continuity losses and structured chunk partitioning, the method significantly improves temporal coherence and task success rates. Experimental results demonstrate that ChunkFlow achieves a superior success-stability trade-off during low-latency inference across various benchmarks, including real robotic applications.
Boundary jitter in chunked action heads can be effectively mitigated, leading to improved task success rates and temporal coherence in vision-language models.
Vision-language action (VLA) models increasingly adopt chunked action heads to satisfy real-time constraints; however, this introduces boundary jitter: overlapping regions between consecutive chunks often yield inconsistent predictions, degrading temporal coherence and the task success rate. Existing methods, such as inference-time blending, merely reweight mismatched proposals without correcting underlying errors, leading to residual accumulation under biased or noisy histories. We propose ChunkFlow, a seam-aware training-and-execution framework for chunked policies that aligns chunk structure with boundary execution. It partitions each chunk into frozen, editable, and future zones, applies deterministic overlap blending at execution, and trains raw predictions with seam and first- and second-order continuity losses. History corruption and scheduled sampling improve robustness to executed-history errors, while an AWAC fine-tuning stage adapts the policy without removing these structural regularizers. Under mild smoothness assumptions, pre-blending seam discrepancies provably decay with increasing overlap. Experiments on CALVIN, LIBERO, and real robots show an improved success-stability trade-off with low-latency inference. Project page: https://cytoderm-ai.github.io/chunkflow.
