← Back

Sensors and algorithm

Thousands of tests were done, reenacting cycling accidents using stunt riders and crash-test dummies, to collect the specific movement patterns of cyclists in accidents.

200 times

When activated, Hövding records the cyclist’s movements 200 times a second.

0,1 second

In the event of an accident, the cyclist’s abnormal movement is detected, and the airbag inflates in less than 0.1 second.

Hövding’s accident detection system consists of an algorithm based on artificial intelligence technology.

This algorithm is trained by providing it with data, corresponding to both accidents and non-accidents, to learn from.

The accident data set is created by collecting the movement patterns from staged accidents. Professional stunt performers simulated over 3,000 accidents of different types and intensity.

The non-accident data is collected by over 300 Hövding fans biking and generating over 2,000 hours of normal biking data.

These two sets of data create a solid base for Hövding’s accident detection algorithm.

The result is an independent system which can detect an accident and deploy the airbag in less than 0.1 seconds.

Hövding 3 stunt test #3207