Researchers have developed Phone2Act, a novel teleoperation system that utilizes a standard smartphone as a 6-DoF robot controller through Google ARCore. This low-cost, hardware-agnostic framework is designed to simplify and scale the collection of manipulation data for training Vision-Language-Action (VLA) models. By decoupling control logic and supporting various robot platforms, Phone2Act enables researchers to more affordably gather diverse datasets for immediate use in fine-tuning models like GR00T-N1.5. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT This system could significantly lower the cost and increase the scale of data collection for VLA models, potentially accelerating their development and deployment.
RANK_REASON This is a research paper detailing a new system for data collection for AI models. [lever_c_demoted from research: ic=1 ai=1.0]