Rondo: Segment-Level Continuous Batching for Multi-Tenant Flow-VLA Serving

Flow-based vision-language-action (VLA) models generate actions via iterative denoising. In multi-tenant serving, diverse robot tasks ranging from reactive obstacle avoidance to precision manipulation naturally require varying numbers of denoising steps. However, executing complete trajectories as indivisible units restricts batching efficiency across differing step counts, while per-step continuous batching incurs prohibitive kernel launch and scheduling overheads. We propose Rondo, a serving framework that partitions denoising trajectories into schedulable segments through a resumable execution abstraction. It combines completion-boundary segmentation with dynamic remaining-work bucketing to prevent segment fragmentation and preserve multi-step execution efficiency. Evaluated with ${\pi }_{0.5}$ on 8 Ascend 910C NPUs under theoretical load capacity stress testing, Rondo reduces mean serving latency by up to 75.2% relative to Native. In strict real-time closed-loop robot fleet manipulation across 16-32 concurrent robots, Rondo achieves up to 92.3% task success and delivers up to a $4\text{-}5\times$ improvement over existing serving baselines, sustaining robust physical execution at scale where baseline systems suffer catastrophic buffer starvation.