SP-VLA: A Joint Model Scheduling and Token Pruning Approach for VLA Model Acceleration

TL;DR

This paper introduces SP-VLA, a unified framework that accelerates Vision-Language-Action models by jointly scheduling models and pruning tokens, addressing temporal and spatial redundancies for real-time applications.

Contribution

The paper proposes a novel joint model scheduling and token pruning approach tailored for VLA models, incorporating action-aware scheduling and dual-aware token pruning for efficiency.

Findings

Achieves 1.5× lossless acceleration in LIBERO

Achieves 2.4× lossless acceleration in SimplerEnv

Improves inference frequency and latency significantly

Abstract

Vision-Language-Action (VLA) models have attracted increasing attention for their strong control capabilities. However, their high computational cost and low execution frequency hinder their suitability for real-time tasks such as robotic manipulation and autonomous navigation. Existing VLA acceleration methods primarily focus on structural optimization, overlooking the fact that these models operate in sequential decision-making environments. As a result, temporal redundancy in sequential action generation and spatial redundancy in visual input remain unaddressed. To this end, we propose SP-VLA, a unified framework that accelerates VLA models by jointly scheduling models and pruning tokens. Specifically, we design an action-aware model scheduling mechanism that reduces temporal redundancy by dynamically switching between VLA model and a lightweight generator. Inspired by the human motion pattern of focusing on key decision points while relying on intuition for other actions, we categorize VLA actions into deliberative and intuitive, assigning the former to the VLA model and the latter to the lightweight generator, enabling frequency-adaptive execution through collaborative model scheduling. To address spatial redundancy, we further develop a spatio-semantic dual-aware token pruning method. Tokens are classified into spatial and semantic types and pruned based on their dual-aware importance to accelerate VLA inference. These two mechanisms work jointly to guide the VLA in focusing on critical actions and salient visual information, achieving effective acceleration while maintaining high accuracy. Extensive experiments show that our method achieves 1.5$\times$ lossless acceleration in LIBERO and 2.4$\times$ in SimplerEnv, with up to 6% average performance gain. Inference frequency and latency improve by 2.2$\times$ in SimplerEnv and 1.4$\times$ in LIBERO.