Case studies

Crash detection system

Challenge

The goal of the project was to create a system that automatically detects that a vehicle was involved in an accident and automatically notifies an emergency center about possible dangerous situation.

Tool used

ML, SVM, Python, Scikit-learn

Solution

We have created an ML model based on support vector machines. The model analyses in real-time high- frequency (100 samples per second) data from an accelerometer sensor. To train the model we have used both existing public data sources (NHTSA crash database) as well as recorded our own data set coming from simulations. We have also applied various augmentations to increase the size of the training set. The final model was then implemented on mobile platforms.

All projects

See also

Leading Consulting Company in Japan

Driver’s Behaviour Analysis Platform