Iron: Intent-Aligned and Retrospective Dual Learning Framework for Enhancing Generalist Virtual Agents

2026-08-27

AI Summary

本文提出Iron框架，通过步进循环一致奖励和后见重生产机制解决数据标注成本高、动作意图不精准及失败轨迹利用低效问题，提升跨环境任务性能。

Abstract

Achieving virtual agents capable of automating tasks across diverse digital environments remains a pivotal challenge in Embodied AI. While Multimodal Large Language Models (MLLMs) offer enhanced visual perception and reasoning, their agentic deployment faces three challenges: costly data annotation, imprecise action-intent alignment, and inefficient exploration from discarded failed trajectories. To address these, we introduce Iron, an intent-aligned, self-improved, and annotation-efficient framework for training GUI agents. Iron employs a novel dual learning strategy that utilizes a stepwise cycle-consistent (SCC) reward to achieve fine-grained alignment between low-level actions and high-level intents, thereby improving instruction grounding and intent understanding. Concurrently, Iron introduces a hindsight reproduction mechanism to repurpose failed trajectories for training, improving both learning efficiency and task diversity. Extensive experiments demonstrate that Iron-trained generalist agents consistently improve performance on cross-environment and cross-device tasks, outperforming models trained with three times more data. Iron also achieves a substantial 25.06% relative improvement on unseen web tasks, with further gains observed on inherently complex tasks, demonstrating the feasibility of building more capable virtual agents.

Problem

• Embodied AI
• Multimodal Large Language Models
• data annotation
• action-intent alignment
• exploration efficiency

Innovation

• dual learning strategy
• stepwise cycle-consistent reward
• hindsight reproduction mechanism