Robots Enter Factories Must First Clear Smoke, Dust and Network Hurdles
The 100th themed seminar and industry matchmaking meeting of Origin Space, held under the theme “Bringing AI into the Physical World,” took place at the West Taihu (Changzhou) Artificial Intelligence International Community. Sun Yanlong, chairman of Danghong Technology, was invited to join representatives from Tsinghua University, the Chinese Academy of Sciences and industry to discuss technical breakthroughs and industrial applications of physical AI.
In the discussion, Sun focused on a practical question: once robots enter real industrial settings, can they actually perform work, and can they do so accurately and quickly? Real factories are far more complex than laboratories. Smoke, dust, occlusion, strong light and unstable networks can all interfere with robot perception and judgment.
The event framed these constraints as prerequisites for deployment, not secondary details. For embodied AI to move from demonstrations to production lines, systems must combine robust sensing, edge computing and resilient communications.
The seminar also served as a matchmaking platform, connecting research teams with companies seeking deployable solutions. Participants agreed that physical AI must be validated under harsh conditions, with reliability and speed measured in actual workflows rather than controlled testbeds.