Android in the Wild: A Large-Scale Dataset for Android Device Control

Christopher Rawles , Alice Li , Daniel Rodriguez , Oriana Riva , Timothy Lillicrap

🏛 Institutions: Google Research , Google DeepMind
📅 Date: July 19, 2023
📑 Publisher: NeurIPS 2023 Datasets and Benchmarks Track
💻 Env: Mobile
🔑 Keywords: dataset benchmark AITW device control gesture-based actions

TLDR

Introduces Android in the Wild, a large-scale dataset of human-labeled Android device-control episodes with natural-language commands and touch actions. It became one of the central training and evaluation resources for mobile agents because it stresses robustness across apps, tasks, and gesture types.

Open paper Report issue