AI · Factory production line
AI Machine-Safety Interlock
Cameras watch a production line and, when a person is somewhere unsafe, the machine near them is stopped through the PLC. One batched detection pass covers 15 cameras and controls two machines independently. Mask zones separate "outside the plant wall" from "allowed to stand here", and the design is fail-safe: when in doubt, stop.
🔒 Demo video not available for privacy reasons.
The problem
People get hurt by machines in a predictable way: someone reaches in to clear a jam while the machine is running. Guards get removed, light curtains cannot tell a person from material, and CCTV only records what already happened.
How it works
- Cameras send live video of the machine area
- A YOLOv8 model finds people in each frame — people, not just movement
- Each detection is tested against the marked danger zones
- When a person is inside a zone, a stop signal is sent to the machine through the PLC
Design choices
- Detecting people instead of motion, so moving fabric and shadows do not stop the line
- Two kinds of zones: areas outside the plant wall are ignored, permitted standing areas are allowed
- One batched detection pass covers many cameras and controls two machines independently
- Fail-safe: stopping a machine without need costs time; not stopping it can cost a person