Environment-Aware Adaptive Pruning with Interleaved Inference Orchestration for Vision-Language-Action Models

TL;DR

EcoVLA tackles real-time constraints in Vision-Language-Action models by introducing environment-aware adaptive pruning (EAP) and interleaved inference orchestration (I2O). EAP leverages temporal environment context to update structured channel sparsity, while I2O hides pruning overhead by scheduling computations inside FLOPs bubbles across VLA's backbone and action head. The framework is training-free and plug-and-play, compatible with other accelerators, and demonstrates up to 1.60x–2.18x speedups with minimal accuracy loss in diverse simulators and real-robot experiments. This work offers a practical path to deploy fast, robust VLA systems in dynamic environments without expensive retraining.

Abstract

While Vision-Language-Action (VLA) models hold promise in embodied intelligence, their large parameter counts lead to substantial inference latency that hinders real-time manipulation, motivating parameter sparsification. However, as the environment evolves during VLA execution, the optimal sparsity patterns change accordingly. Static pruning lacks the adaptability required for environment dynamics, whereas fixed-interval dynamic layer pruning suffers from coarse granularity and high retraining overheads. To bridge this gap, we propose EcoVLA, a training-free, plug-and-play adaptive pruning framework that supports orthogonal combination with existing VLA acceleration methods. EcoVLA comprises two components: Environment-aware Adaptive Pruning (EAP) and Interleaved Inference Orchestration ($I^2O$). EAP is a lightweight adaptive channel pruning method that incorporates the temporal consistency of the physical environment to update sparsity patterns. $I^2O$ leverages the FLOPs bubbles inherent in VLA inference to schedule the pruning method in parallel, ensuring negligible impact on latency. Evaluated on diverse VLA models and benchmarks, EcoVLA delivers state-of-the-art performance, achieving up to 1.60$\times$ speedup with only a 0.4% drop in success rate, and further reaches 2.18$\times$ speedup with only a 0.5% degradation when combined with token pruning. We further validate the effectiveness of EcoVLA on real-world robots.

Figures (7)

• Figure 1: During VLA execution, channel importance scores vary dynamically as the environment evolves, causing the optimal sparsity pattern to shift accordingly.
• Figure 2: Overall pipeline of EcoVLA. (a) Environment-aware Adaptive Pruning (EAP): EAP is a lightweight, environment-aware method that identifies sparsity variations by perceiving real-time dynamics. Considering the temporal consistency of VLA execution in physical environments, EAP integrates instantaneous features with historical features to jointly compute the sparsity pattern. (b) Interleaved Inference Orchestration (I2O): I2O interleaves sparsity pattern computation into the inherent FLOPs bubbles within the VLA inference using a non-blocking parallel paradigm.
• Figure 3: Robot Manipulation on Kinova Gen3 Platform.
• Figure 4: Acceleration breakdown for dense and sparse inference.
• Figure 5: Trade-off between Success Rate and Latency.