Atomrock is at Computex Taipei 2026 (June 2–5, Taipei Nangang Exhibition Center, Hall 1, 4F, Booth N1329), where this edge-AOI solution makes its public debut. Coinciding with the show, Taiwan’s industrial press has put the work in the spotlight.

The Journal of Industrial Mechatronics (機械工業雜誌), the technical publication of Taiwan’s Industrial Technology Research Institute (ITRI), led its Featured Reports on June 3, 2026 with a story on Atomrock’s edge AOI work: “Edge AOI Moves Into Wafer Wet-Process Tools: Compressing the Defect Response Cycle From an Hour to a Minute.”

By building edge vision AI into the spin chamber, the system images every wafer during processing and runs multi-model inference at the edge on NVIDIA Jetson Orin, catching surface residue, liquid pooling, bevel anomalies, wafer wobble, and dispenser issues such as nozzle-angle drift and suck-back. That brings the defect response cycle down from the usual 30 to 60 minutes to under one minute, and narrows the impact of a bad batch from dozens of wafers to the event itself. When a recipe or wafer layer changes, models can be re-trained on site within 24 to 48 hours without taking wafer images out of the fab, meeting customers’ data-sovereignty requirements. VisionSense tags each event by severity (critical / major / minor).

Read the full feature on the Journal of Industrial Mechatronics: automan.tw.