Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot Deployment

TL;DR

This paper systematically analyzes vision-language-action models on various edge hardware, revealing bottlenecks and proposing methods to accelerate inference for real-time robot control with minimal accuracy loss.

Contribution

It introduces a cross-accelerator leaderboard, uncovers inference phase patterns, and proposes DP-Cache and V-AEFusion to significantly speed up models on edge devices.

Findings

Edge devices can be more cost- and energy-efficient than flagship GPUs.

Inference involves a compute-bound backbone and a memory-bound Action Expert.

Proposed methods achieve up to 2.9x GPU speedup and 6x edge NPU speedup.

Abstract

Vision-Language-Action (VLA) models are promising for generalist robot control, but on-robot deployment is bottlenecked by real-time inference under tight cost and energy budgets. Most prior evaluations rely on desktop-grade GPUs, obscuring the trade-offs and opportunities offered by heterogeneous edge accelerators (GPUs/XPUs/NPUs). We present a systematic analysis for low-cost VLA deployment via model-hardware co-characterization. First, we build a cross-accelerator leaderboard and evaluate model-hardware pairs under CET (Cost, Energy, Time), showing that right-sized edge devices can be more cost-/energy-efficient than flagship GPUs while meeting control-rate constraints. Second, using in-depth profiling, we uncover a consistent two-phase inference pattern: a compute-bound VLM backbone followed by a memory-bound Action Expert, which induces phase-dependent underutilization and hardware inefficiency. Finally, guided by these insights, we propose DP-Cache and V-AEFusion to reduce diffusion redundancy and enable asynchronous pipeline parallelism, achieving up to 2.9x speedup on GPUs and 6x on edge NPUs with only marginal success degradation. The example leaderboard website is available at: https://vla-leaderboard-01.vercel.app/.