INSAIT develops breakthrough system to train smarter autonomous vehicles

The Institute for Computer Science, Artificial Intelligence and Technology (INSAIT) at Sofia University ‘St. Kliment Ohridski’ has introduced DiffSim Trinity, a new algorithm designed to help self-driving cars make better decisions. Made up of three key innovations, it could change the way autonomous vehicles understand their surroundings, predict what might happen next, and choose the safest path.

A key challenge for autonomous cars today is not just carrying out commands, but also understanding the consequences of their actions — for example, how a slight turn or acceleration can alter the situation just seconds later. The method developed by INSAIT researchers enables a vehicle to react to a scenario and anticipate the consequences of its actions by analysing multiple potential outcomes. This allows it to choose the optimal trajectory.

Asen Nachkov, a PhD student at INSAIT and the project’s lead author, returned to Bulgaria three years ago after completing a master's degree at Imperial College London, in order to continue his research in Sofia. This work was carried out in collaboration with the University of Zurich and ETH Zurich.

The project has already attracted interest from leading technology companies in the autonomous driving sector.

Editor: Miglena Ivanova
Posted in English by E. Radkova