NTT Develops Physical AI Technology Enabling Robots to Predict and Prevent Failures
NTT has announced a new physical AI technology that allows robots to predict and evaluate potential outcomes, such as load collapse or object drops, across multiple actions before execution. This system integrates a proprietary "physics control adapter" into a video generation model, explicitly handling physical quantities and predicting state changes based on equations of motion. Initial trials in shelf-loading tasks showed an improved success rate from 73.1% to 92.3%, with initial applications envisioned for automated airport baggage handling.
Want more?
Open NewsSnap.ai for the full app experience, including audio, personalization, and more news tools.
Open NewsSnap.ai