FURI | Spring 2018

Deep Predictive Models for Collision Risk Assessment in Autonomous Driving

Security icon, disabled. A blue padlock, locked.

The research objective is the implementation of a Collision Avoidance System for automobiles using deep neural networks. The researchers have been able to generate the data set in Webots, and use it to train/test the predictive model, thus obtaining higher levels of accuracy compared to the previous simulation environment (VREP). The next goal is to publish a data set paper and to increase the realism of the scenarios by adding abnormal driving behaviours.

Student researcher

Cesar Tamayo

Computer systems engineering

Hometown: Havana, Cuba

Graduation date: Spring 2020