Newest papers

PACT-VLA: Speculative Action-Chunk Verification with Pre-Commit Observations

Yixiang Zhu, Ruilong Wu, Yonghao Chen, Jiaxiang Zou, Yusong Hu, Xinyu Chen · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28019 · cited by 0 works (OpenAlex) · open-access PDF

Indonesian

Kebijakan Vision-language-action-batayACT (VLA) kebijakan umumnya didistribusikan dengan inferensi asynchronous di pengaturan dunia nyata, di mana titik aksi berikutnya dihasilkan sementara robot masih melakukan tugas sebelumnya. Pada saat chunk baru dikompromikan, robot dan adegan telah berubah, sehingga chunk dikondisikan pada pengamatan konstan dan kesuksesan tugas berkurang tajam saat keterlambatan inferensi ini tumbuh. Kin menghasilkan keterlambatan inferensi yang lebih banyak dapat memungkinkan pengamatan yang lebih baru mengevaluasi pilihan akhir, tetapi metode multi-kandidat yang ada baik menempatkan kandidat dengan pengamatan inferensi-start konstan atau berulang kali beralih antara kandidat selama eksekusi, yang dapat membuat urutan tindakan yang dijalankan tidak berkelanjutan. wawasan utama kami adalah bahwa keterlambatan inferensi dapat digunakan sebagai waktu keputusan. Sementara

Machine translation (Meta M2M100); the English original follows.

Abstract

Vision-language-action (VLA) policies are commonly deployed with asynchronous inference in real-world settings, where the next action chunk is generated while the robot is still executing the previous one. By the time the new chunk is committed, the robot and scene have changed, so the chunk is conditioned on a stale observation and task success drops sharply as this inference delay grows. Generating multiple candidate chunks could let newer observations inform the final choice, but existing multi-candidate methods either rank candidates with the stale inference-start observation or repeatedly switch between candidates during execution, which can make the executed action sequence discontinuous. Our key insight is that inference delay can be used as decision time. While candidate generation is still underway, candidate selection can reuse the VLA itself to choose the complete action chunk that best fits a later observation. We propose PACT-VLA, a training-free framework that generates candidates from the inference-start observation and selects one using this pre-commit observation. Its scoring function, FlowRes, measures the residual between each candidate's implied flow direction and the velocity predicted by the frozen policy under the pre-commit observation, requiring no separate verifier. A pipelined implementation overlaps observation processing with candidate generation and batches scoring, so selection adds no control-step delay on the evaluated hardware. At a four-step delay, PACT-VLA improves average success by 15.0 percentage points over RTC on Kinetix and by 5.2 points over single-candidate asynchronous execution with GR00T N1.7 on LIBERO, and it further improves existing delay-robust methods. On three real-robot tasks, it raises average success from 60% to 82% over naive asynchronous execution with the same frozen $π_{0.5}$ checkpoint.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Asynchronous communication, Task (project management), Inference, Selection (genetic algorithm), Key (lock)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange