When Intelligence Meets Safety.

Protecting a rider at 220km/h leaves no room for error, and no time for reactive systems. The Pufferfish Airbag System was built around a different premise: safety that anticipates. Six advanced sensors embedded in the racing suit work in tandem with GPS, a custom motherboard, and dedicated processing hardware to continuously monitor the rider's dynamics. A machine learning model runs on-device, trained to recognize specific accident scenarios (falls, slides, and high-sides) and distinguish them from normal aggressive riding. When a critical pattern is detected, the airbag deploys in milliseconds, before impact occurs. Ermit owned the full development stack: electronic design, mechanical engineering, firmware, and the complete software suite for data collection, analysis, and simulation. The system has been validated in the most demanding conditions possible, deployed across MotoGP, Moto2, Moto3, and SBK championships and is now being adapted for everyday road use. It's a concrete example of what becomes possible when hardware, real-time inference, and human safety are engineered together as a single system from day one.