Aguvis: Unified Pure Vision Agents for Autonomous GUI Interaction
Yiheng Xu, Zekun Wang, Junli Wang, Dunjie Lu, Tianbao Xie, Amrita Saha, Doyen Sahoo, Tao Yu, Caiming Xiong
2024-12-06
Summary
This paper introduces Aguvis, a new system designed to automate tasks in Graphical User Interfaces (GUIs) by using advanced visual understanding instead of relying on text-based methods.
What's the problem?
Automating tasks in GUIs is challenging because GUIs can be complex and vary greatly between different applications. Most existing methods use text descriptions to understand GUIs, which can limit their ability to adapt to different environments and make them less efficient. This reliance on text can lead to difficulties in generalizing across various platforms.
What's the solution?
Aguvis solves this problem by creating a framework that uses visual observations directly from the GUI. It combines image-based inputs with natural language instructions to interact with visual elements in a consistent way. The system includes explicit planning and reasoning capabilities, allowing it to navigate and perform tasks in complex digital environments autonomously. The researchers also built a large dataset of GUI interactions to train Aguvis effectively, focusing on both understanding the GUI and planning actions.
Why it matters?
This research is important because it represents a significant step forward in making automated GUI agents that can work independently across different platforms. By improving how these agents understand and interact with visual information, Aguvis can enhance the efficiency of software testing, user interface automation, and other applications where human-computer interaction is key. The open-sourcing of their datasets and models will also encourage further research and development in this area.
Abstract
Graphical User Interfaces (GUIs) are critical to human-computer interaction, yet automating GUI tasks remains challenging due to the complexity and variability of visual environments. Existing approaches often rely on textual representations of GUIs, which introduce limitations in generalization, efficiency, and scalability. In this paper, we introduce Aguvis, a unified pure vision-based framework for autonomous GUI agents that operates across various platforms. Our approach leverages image-based observations, and grounding instructions in natural language to visual elements, and employs a consistent action space to ensure cross-platform generalization. To address the limitations of previous work, we integrate explicit planning and reasoning within the model, enhancing its ability to autonomously navigate and interact with complex digital environments. We construct a large-scale dataset of GUI agent trajectories, incorporating multimodal reasoning and grounding, and employ a two-stage training pipeline that first focuses on general GUI grounding, followed by planning and reasoning. Through comprehensive experiments, we demonstrate that Aguvis surpasses previous state-of-the-art methods in both offline and real-world online scenarios, achieving, to our knowledge, the first fully autonomous pure vision GUI agent capable of performing tasks independently without collaboration with external closed-source models. We open-sourced all datasets, models, and training recipes to facilitate future research at https://aguvis-project.github.io/.