ControlVLA: Few-shot Object-centric Adaptation for Pre-trained Vision-Language-Action Models

Puhao Li, Yingying Wu, Ziheng Xi, Wanlin Li, Yuzhe Huang, Zhiyuan Zhang, Yinghan Chen, Jianan Wang, Song-Chun Zhu, Tengyu Liu, Siyuan Huang

Keywords: Robotic manipulation, Imitation learning, Few-shot learning

TL;DR: We propose ControlVLA, a novel manipulation learning framework that fine-tunes tasks using only 10-20 demonstrations in real-world.

Abstract:

Learning real-world robotic manipulation is challenging, particularly when limited demonstrations are available. Existing methods for few-shot manipulation often rely on simulation-augmented data or pre-built modules like grasping and pose estimation, which struggle with sim-to-real gaps and lack extensibility. While large-scale imitation pre-training shows promise, adapting these general-purpose policies to specific tasks in data-scarce settings remains unexplored. To achieve this, we propose ControlVLA, a novel framework that bridges pre-trained VLA models with object-centric representations via a ControlNet-style architecture for efficient fine-tuning. Specifically, to introduce object-centric conditions without overwriting prior knowledge, ControlVLA zero-initializes a set of projection layers, allowing them to gradually adapt the pre-trained manipulation policies.
In real-world experiments across 6 diverse tasks, including pouring cubes and folding clothes, our method achieves a 76.7\% success rate while requiring only 10-20 demonstrations --- a significant improvement over traditional approaches that require more than 100 demonstrations to achieve comparable success.
Additional experiments highlight ControlVLA's extensibility to long-horizon tasks and robustness to unseen objects and backgrounds.

Submission Number: 379