Your browser doesn't support javascript. This means that the content or functionality of our website will be limited or unavailable. If you need more information about Vinnova, please contact us.

AutoDeep: Automatic Design of Safe, High-Performance and Compact Deep Learning Models for Autonomous Vehicles

Reference number
Coordinator Mälardalens Universitet - Akademin för innovation, design och teknik, Västerås
Funding from Vinnova SEK 6 145 879
Project duration September 2020 - May 2025
Status Completed
Venture Electronics, software and communication - FFI
Call Electronics, software and communication - FFI - 2019-12-10

Important results from the project

the project met its key goals by developing an automatic framework for compact, robust, and efficient deep learning models in safety-critical autonomous systems. It delivered validated solutions for trajectory prediction and lane detection, with strong industrial collaboration. The results led to multiple publications, student training, and spin-offs into other AI reliability and Edge-AI initiatives, strengthening Sweden’s position in embedded AI.

Expected long term effects

The project lays the foundation for robust, compact AI models in safety-critical applications, enabling real-time deployment in autonomous vehicles. Its methods can be reused across sectors like healthcare, robotics, and aerospace. It also contributes to long-term industrial competence, education, and the development of future AI certification and reliability frameworks in Sweden and beyond.

Approach and implementation

The project was structured into four work packages and carried out in close collaboration between MDU, Volvo CE, and Zenseact. Despite initial delays due to COVID-19 and visa issues, the project adapted well. WP1 followed the plan; WP2 was delayed but recovered through remote work and access to compute resources. All major goals were achieved, and the collaboration was effective, resulting in strong outcomes in research, demonstrations, and industrial impact.

The project description has been provided by the project members themselves and the text has not been looked at by our editors.

Last updated 16 September 2025

Reference number 2019-05881