VOTE: Vision-Language-Action Optimization with Trajectory Ensemble Voting
This work addresses efficiency and performance bottlenecks in VLA models for robotics, offering incremental improvements with practical deployability.
The paper tackles the issues of high inference latency and insufficient action utilization in Vision-Language-Action (VLA) models for robotic manipulation by developing a training framework to reduce action tokens and an inference optimization with ensemble voting, achieving higher success rates and 39× faster inference than OpenVLA.
Recent large-scale Vision Language Action (VLA) models have shown superior performance in robotic manipulation tasks guided by natural language. However, current VLA models suffer from two drawbacks: (i) generation of massive tokens leading to high inference latency and increased training cost, and (ii) insufficient utilization of generated actions resulting in potential performance loss. To address these issues, we develop a training framework to finetune VLA models for generating significantly fewer action tokens with high parallelism, effectively reducing inference latency and training cost. Furthermore, we introduce an inference optimization technique with a novel voting-based ensemble strategy to combine current and previous action predictions, improving the utilization of generated actions and overall performance. Our results demonstrate that we achieve superior performance compared with state-of-the-art VLA models, achieving significantly higher success rates and 39$\times$ faster inference than OpenVLA with 46 Hz throughput on edge platforms, demonstrating practical deployability. The code is available at https://github.com/LukeLIN-web/VOTE.