Engineering virtual prototype · Hybrid AI + Rule-Based architecture · interactive simulation
Animated dashed arrows = live data flow. Solid grey arrows = physical observation (sensor contact / camera view). Use the controls below to drive the simulation.
AIHandles perception: a CNN classifies each CCTV frame as OPEN/CLOSED with a confidence score; uncertain frames are never silently trusted. RULESHandle decisions: deterministic, auditable logic fuses the API train state, the sensor contact, and the AI output. The railway gate status is derived from the sensor (most reliable) cross-checked with train data; the alternate gate status comes from the vision model. Suggested hardware: industrial PC with GPU/NPU, PoE camera, isolated digital input for the contact sensor, UPS power, and a cloud-hosted dashboard behind TLS. Note: this is a monitoring concept, not a safety-rated interlock for the crossing itself.