NEURA Robotics unveils Neuron AI multimodal foundation model

NEURA Robotics has introduced Neuron AI, a multimodal foundation model designed for Physical AI. Utilizing a Perception-Language-Action (PLA) architecture, the model integrates 3D vision, tactile and force sensing, audio, robot status, and language into a single shared space to enable complex physical actions.

Unlike traditional camera-based systems, Neuron AI incorporates multisensory data and memory, allowing robots to understand physical nuances such as the sound or feel of a lock opening. The model was developed and trained in Metzingen, Germany, using a combination of real-world robot data, simulated data, and human demonstrations.

NEURA Robotics plans to deploy this technology across its robot embodiments, including its humanoid 4NE1, which is scheduled to demonstrate autonomous task completion at NVIDIA GTC Berlin. Customers can further customize the model for specific tasks via the Neuraverse platform.