Networked intelligent vehicles with adaptive autonomy in realistic traffic scenarios

The goal of this research is to improve driver safety and overall traffic fluidity using innovative intelligent vehicle technology.

This project focuses on the development of novel algorithms able to manage a variable level of driving assistance in intelligent vehicles operating in realistic traffic scenarios. The level of assistance can vary depending on the surrounding driving context (e.g., traffic pattern, type of road, weather conditions) and the driver’s state (impaired or not). It can range from warning signals and basic recommendations to fully autonomous driving.

Scientific challenges are concerned with designing an efficient sensory system, performing situational risk, and planning the appropriate actions. In terms of algorithmic validation, we will work both with real vehicles, provided by Groupe PSA, and simulation tools at various abstraction levels.

Our approach differs from those currently pursued in the area of intelligent vehicles by emphasizing three aspects of the problem often investigated separately: first, we are interested in exploring an adaptive level of driving assistance; second, we will exploit car-to-car communication as an additional source of information for better assisting the driver; third, we will investigate the impact of such local vehicle capabilities on the overall performance of the traffic system and aim at minimizing the local complexity of the individual vehicle for achieving a given level of safety and fluidity at the global level.

This project will support PhD student Milos Vasic and will be carried out at Distributed Intelligent Systems and Algorithms Laboratory (DISAL) headed by Prof. Alcherio Martinoli. It is sponsored by Groupe PSA and is part of the long-term collaborative research agenda between Groupe PSA and EPFL.
Press release 4.10.2017

Principal investigator
Prof. Alcherio Martinoli
Project manager
Milos Vasic
Sponsor Groupe PSA
Period 2012-2017
Laboratory DISAL
Collaboration TRACE