A3: Android Agent Arena for Mobile GUI Agents with Essential-State Procedural Evaluation

TL;DR

This paper introduces Android Agent Arena (A3), a comprehensive benchmark and evaluation system for mobile GUI agents that assesses performance in dynamic, real-world online apps using an innovative essential-state procedural approach.

Contribution

The paper presents a novel essential-state based procedural evaluation system and benchmark for mobile GUI agents, addressing limitations of static assessments and enabling more realistic performance measurement.

Findings

A3 includes 100 tasks from 20 popular online apps across diverse categories.

The evaluation method uses MLLMs as reward models for task verification.

The system facilitates data collection and environment management for mobile GUI research.

Abstract

The advancement of Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) has catalyzed the development of mobile graphic user interface (GUI) AI agents, which is designed to autonomously perform tasks on mobile devices. However, a significant gap persists in mobile GUI agent evaluation, where existing benchmarks predominantly rely on either static frame assessments such as AndroidControl or offline static apps such as AndroidWorld and thus fail to capture agent performance in dynamic, real-world online mobile apps. To address this gap, we present Android Agent Arena (A3), a novel "essential-state" based procedural evaluation system for mobile GUI agents. A3 introduces a benchmark of 100 tasks derived from 20 widely-used, dynamic online apps across 20 categories from the Google Play Store, ensuring evaluation comprehension. A3 also presents a novel "essential-state" based procedural evaluation method that leverages MLLMs as reward models to progressively verify task completion and process achievement. This evaluation approach address the limitations of traditional function based evaluation methods on online dynamic apps. Furthermore, A3 includes a toolkit to streamline Android device interaction, reset online environment and apps and facilitate data collection from both human and agent demonstrations. The complete A3 system, including the benchmark and tools, will be publicly released to provide a robust foundation for future research and development in mobile GUI agents.