Quantization-Aware Imitation-Learning for Resource-Efficient Robotic Control

TL;DR

This paper introduces a quantization framework for imitation learning policies that significantly reduces computational costs on resource-limited hardware while maintaining accuracy, enabling efficient robotic control and autonomous driving.

Contribution

It presents a novel quantization-aware training method for IL-based policies, improving robustness and efficiency on low-bit hardware without sacrificing performance.

Findings

Achieves up to 2.5x speedup and 2.5x energy savings in robot manipulation tasks.

Attains up to 3.7x speedup and 3.1x energy savings in self-driving models.

Maintains high accuracy despite low-bit quantization.

Abstract

Deep neural network (DNN)-based policy models like vision-language-action (VLA) models are transformative in automating complex decision-making across applications by interpreting multi-modal data. However, scaling these models greatly increases computational costs, which presents challenges in fields like robot manipulation and autonomous driving that require quick, accurate responses. To address the need for deployment on resource-limited hardware, we propose a new quantization framework for IL-based policy models that fine-tunes parameters to enhance robustness against low-bit precision errors during training, thereby maintaining efficiency and reliability under constrained conditions. Our evaluations with representative robot manipulation for 4-bit weight-quantization on a real edge GPU demonstrate that our framework achieves up to 2.5x speedup and 2.5x energy savings while preserving accuracy. For 4-bit weight and activation quantized self-driving models, the framework achieves up to 3.7x speedup and 3.1x energy saving on a low-end GPU. These results highlight the practical potential of deploying IL-based policy models on resource-constrained devices.