Impinj, an RFID company, said enterprise AI often fails on factory floors and in warehouses because it lacks reliable real-time data about physical items. Its VP Matt Branda argued that AI without such data is just guesswork. He introduced the idea of 'physical intelligence' as the data layer that tells AI what to act on.
To use AI widely in machines that move based on sensors, such as robots, drones, and cars, chips and entire systems must be redesigned. The article points out that problems with memory, power, specialized chip structures, and safety verification are all tangled together.
Home service platform Measo is unveiling 'Misomotion,' a service that collects videos of field workers' hand skills, and is entering the physical AI data market in earnest. It is a project that gathers on-site work footage so that physical AI (AI that has a physical body and moves in the real world, such as robots) can learn from it.
Moobin, an AI company that uses lidar sensors to turn human movement into 3D data, has raised a 10 billion won Series A investment. IMM Investment led the round, and existing investors joined the follow-on funding. Moobin plans to use the money to build an R&D center and grow its physical AI (AI that handles robots and other machines that move in the real world).