started · updated
Panasonic HD accelerates physical AI and mobile world models
Panasonic Holdings is advancing its research into “physical AI,” a strategy that combines manufacturing expertise with artificial intelligence to support human workers. During its Technology Day, the company revealed plans to integrate AI with its existing technologies in materials, sensing, and control to create a cycle where AI learns from human experience and field data.
To support this, Panasonic established a new AI and Robotics R&D division in October 2026, employing approximately 250 researchers. This division focuses on both fundamental research, such as 3D spatial recognition, and applied AI to solve practical challenges in factories and logistics centers. Notably, the company has developed a 3D spatial recognition technology that reduces input information volume by up to ten times compared to conventional methods, a breakthrough achieved through collaboration with UC Berkeley and Tsinghua University.
Additionally, Panasonic has developed a “world model” designed to assist with tasks on mobile devices and web browsers. By predicting how screens will change during operations, the AI agent improved task success rates on Android devices from 46.9% to 54.3%. This technology aims to reduce the workload of field workers by automating complex tasks across various applications, such as equipment management and inventory tracking.
Entities
Panasonic Connect · Panasonic Holdings · Tsinghua University · UC Berkeley · UCLA