PROSENSE: PROactive SENSing for autonomous driving

Important results from the project

PROSENSE is a Vinnova-funded research project carried out by Scania CV AB in collaboration with KTH. The aim has been to develop new capabilities in detection and prediction of moving objects (e.g. vehicles), especially in environments where sensors can be obscured by other objects. The project has primarily contributed to developing new competence and knowledge in the perception and prediction, i.e. using LLMs and VLMs, an area in which it is very important for Sweden to show its leadership.

Expected long term effects

The biggest long-term impact that PROSENSE will have is indirect. We believe that the knowledge of how to build, train and evaluate neural networks with sensor data from real vehicle environments will be crucial for the future development of autonomous vehicles, in short, practical data-driven development. Some of the methods and prototypes that emerged during the project may also prove to be important for future product development.

Approach and implementation

The initial plan was to have two PhD supervisors, each with an industrial PhD student, who would handle the project full-time. Additional engineering resources could be added for specific tasks (integration, production of datasets, etc.). During the course of the project, major changes occurred in the external world (AI), which affected choice of methods. There was also some rotation of staff and supervisors, but in the end we are very satisfied with the result and the collaboration with KTH.