EcoVLA: Energy-Efficient Device-Edge Co-Inference for Vision-Language-Action Models under Real-Time Constraints

2026-08-15

AI Summary

This study addresses the challenges of high inference costs and conflicting real-time energy efficiency requirements in deploying Vision-Language-Action (VLA) models for robotics. We propose an adaptive device-edge collaborative inference framework that establishes an architecture-agnostic unified design space with stage-level abstraction. By integrating a joint latency-energy prediction model, millisecond-scale scheduling algorithms, and lightweight tensor transmission mechanisms, the system achieves dynamically optimal inference decisions. Experimental results demonstrate that under 20Hz real-time constraints, our approach improves energy efficiency by up to 236%. Furthermore, it consistently satisfies Service Level Objectives despite network and workload fluctuations, effectively balancing real-time performance with energy consumption for practical robotic deployment.

Abstract

Vision-Language-Action (VLA) models have emerged as a promising foundation for Embodied AI, but their high inference cost poses significant challenges for deployment in robotic systems. In practice, on-device inference is constrained by limited compute capacity and energy budgets, struggling to simultaneously satisfy real-time control and energy efficiency requirements. Alternatively, offloading the inference workload to an edge server is susceptible to fluctuations in system conditions, introducing unpredictable latency risks. Device-edge co-inference offers a promising solution, but systematic research tailored to VLA models remains scarce, particularly a unified co-inference framework that jointly addresses real-time constraints and system-level energy efficiency. Thus, we propose EcoVLA, an adaptive device-edge co-inference framework for VLA models that maximizes system energy efficiency under real-time constraints. EcoVLA first introduces a unified stage-level abstraction over different VLA paradigms, establishing an architecture-agnostic co-inference design space. It then formulates a joint device-edge-network latency and energy prediction model to enable rapid runtime evaluation of candidate co-inference schemes. Building on this, EcoVLA continuously selects the energy-optimal scheme satisfying real-time constraints with millisecond-level overhead, adapting to runtime variations in network and system states. Furthermore, EcoVLA incorporates a lightweight transmission mechanism for inter-stage intermediate tensors to reduce the communication overhead incurred by cross-device collaboration. Experimental results across VLA models show that EcoVLA improves system energy efficiency by up to 236% over existing co-inference approaches under a 20 Hz action output frequency constraint, while consistently maintaining SLO satisfaction under dynamic network and edge workload conditions.

Problem

• Vision-Language-Action Models
• Device-Edge Co-Inference
• Energy Efficiency
• Real-Time Constraints
• Embodied AI

Innovation

• Device-Edge Co-Inference
• Vision-Language-Action Models
• Energy Efficiency
• Real-Time Constraints
• Adaptive Scheduling