OrbitWeaver saves vision language work during satellite interruptions
Resuming Vision-Language Inference across Onboard Interruptions
Distributed, Parallel, and Cluster ComputingArtificial Intelligence
Summary
Satellites with cameras and AI often get interrupted by power or system resets, causing them to lose progress on understanding images and language. The team behind this paper created OrbitWeaver, a system that carefully saves and restores the AI’s work so it can pick up exactly where it left off after interruptions. This saves time and storage by avoiding repeating the same steps. They tested OrbitWeaver on real satellite hardware and even deployed it in space for three days, showing it keeps AI tasks running smoothly despite shutdowns.
What this means in practice
- •For satellite system engineers: Maintain progress of vision-language AI tasks onboard small satellites despite power or system interruptions.
- •For edge ai device developers: Enable AI models on battery-limited or thermally constrained devices to resume interrupted computations efficiently.
Authors
Xiaolong Huang, Yinyuan Zhang, Daliang Xu, Yun Ma, Qing Li, Mengwei Xu, Shangguang Wang, Gang Huang, Xuanzhe Liu
Abstract
Executing vision-language models (VLMs) aboard low-Earth orbit satellites enables flexible processing of observation data without first downlinking it to the ground. On commercial off-the-shelf payloads, however, a request may span multiple execution windows: power and thermal constraints, fault-triggered reboots, and competing payload demands can interrupt inference before it completes. Each shutdown discards volatile activations and KV caches, forcing an unprotected request to repeat work after restart. We present OrbitWeaver, a framework that preserves VLM inference progress across onboard interruptions through coordinated checkpointing and recovery. Its phase-aware checkpointing selects recovery boundaries according to the state lifetimes and computational characteristics of vision encoding, language-model prefill, and decoding, balancing storage traffic against recomputation. Its dependency-aware dynamic recovery schedules restoration from partially completed states across the layer and token-chunk dimensions. By overlapping checkpoint and weight reads with recomputation, it reduces restart latency; the recovery planner computes an optimal execution order under its scheduling model within 100 ms. Experiments on satellite payload hardware using on-orbit telemetry traces show up to a 4.07-fold end-to-end speedup and a 3.56-fold reduction in SSD write volume over the evaluated baselines. A 72-hour deployment on two low-Earth orbit satellites demonstrates that OrbitWeaver retains inference progress and resumes execution across repeated bus-enforced shutdowns.