Traffic safety benefits in a new SAFER connected project: Deep multimodal learning for automotive applications
This project aims to make driving safer by improving the perception systems in Autonomous Drive (AD) and Advanced Driver Assistance Systems (ADAS). The researchers are focusing on creating advanced sensor fusion methods to enhance the accuracy and robustness of these systems. The project specifically targets three key areas:
- developing fusion architectures for dynamic and static objects,
- exploring self-supervised learning for multimodal data,
- and improving the system's capability to handle rare events, objects and road users.
By implementing these techniques, we anticipate a significant improvement in the safety of vehicles equipped with ADAS/AD systems. This enhanced safety, in turn, can speed up the public acceptance of AD systems, contributing to a much safer transportation environment for all road users.
Zenseact, Volvo Cars, and Chalmers are partners in this recently launched Vinnova project, scheduled for completion in August 2027. The project has a budget of 24.5 MSEK, and it is expected that three PhD students will contribute to the research efforts.