ROSA: A Robotics Foundation Model Serving System for Robot Factories
2026-07-01 • Robotics
RoboticsDistributed, Parallel, and Cluster Computing
AI summaryⓘ
The authors created ROSA, a new system to help multiple robots in a factory share powerful GPUs over a network instead of each using its own small one. This setup improves how fast and efficiently robots can do their tasks, saves battery life, and keeps GPUs busy. ROSA also supports running different models for different tasks, handles failures, and schedules work to boost overall factory productivity rather than just focusing on one robot's speed. Their tests show that ROSA can make the factory up to 12 times more productive than older systems.
Robotics foundation modelsGPU poolingInference servingMulti-model pipelinesTask schedulingFactory productivityRay ServeDistributed orchestrationvLLMPyTorch
Authors
Wenqi Jiang, Jason Clemons, Rowland O'Flaherty, Hugo Hadfield, Alperen Degirmenci, Shuran Song, Yashraj Narang, Christos Kozyrakis
Abstract
Robotics foundation models (RFMs) are making general-purpose robots increasingly practical for factory deployments. While RFM serving systems are central to this vision, existing systems are largely shaped by a single-robot, single-model assumption: inference is treated as an edge-computing problem handled by an on-robot or dedicated nearby GPU, and the serving objective is to minimize the latency of a single action model. In this paper, we propose ROSA, an RFM serving system for robot factories designed around three key principles. First, ROSA adopts shared GPU-pool serving, allowing a fleet of robots to access powerful server-class GPUs over the network in order to improve inference performance, battery duration, and GPU utilization. Second, ROSA provides a robotics-aware programming abstraction and system design that supports multi-model pipelines, per-task performance requirements, and failure handling. Third, ROSA uses factory-objective-driven scheduling to maximize SLO-qualified factory productivity rather than minimizing individual request latency. We implement ROSA on top of Ray Serve for distributed orchestration, with vLLM, PyTorch, and JAX as model-serving backends, and evaluate it on both real robots and synthetic large-scale workloads. The results show that ROSA improves factory productivity by up to 12.06x over conventional dedicated serving systems.