Open-source PhoneDriver uses Qwen3-VL to automate Android phone control via ADB
tom_doerr · x · 2026-08-12
The GitHub project PhoneDriver leverages Qwen3-VL vision-language models and ADB commands to automate Android device control. Users can describe tasks in natural language, and the system visually interprets screens and executes actions. It includes a Gradio web UI with real-time screenshots and logs. Requires Python 3.10+, USB debugging enabled, ADB, and a GPU with sufficient VRAM (tested on 24GB with Qwen3-VL-8B).
More from coding & agent
- Inside OpenAI's Software Engineering for the Agent Era — every · 2026-08-12
- SellerIVA MCP: Analyze Amazon Sales & Ads with Claude — Vivid-Door87 · 2026-08-12
- AI Coding Agents Cheat on Tests: How Developers Prevent Unattended Drift — dunstemplea · 2026-08-12
- Statewave: An Open-Source Memory Runtime to Solve AI Agent Trust Issues — Shruti_0810 · 2026-08-12
- AI Agent Runs Autonomously for 2 Hours to Discover and Fix Issues — sujingshen · 2026-08-12
- Generating Interactive HTML Charts On-the-Fly with Agents — capetorch · 2026-08-12