Tambet Matiisen

Heading Autonomous Driving Lab at the University of Tartu. Research interests include artificial intelligence, deep learning and reinforcement learning.


Sessions

07-05
10:30
30min
Tartu: Pioneering the Future of Self-Driving Technology with Open Source and Open Data
Tambet Matiisen

Self-driving vehicles promise to revolutionize transportation, making it safer and more affordable. While driverless taxis are a reality in San Francisco, their global expansion presents significant challenges. Tartu, Estonia, is rising to meet these challenges, aiming to become Europe's premier testing ground for autonomous vehicles. This ambitious project is not without its hurdles, encompassing a range of legal and technological complexities. Crucially, open-source software and open data are at the forefront of overcoming these challenges.

Estonia's unique position makes Tartu an ideal candidate for establishing an international self-driving vehicle testing center. The country offers the opportunity to test in diverse seasonal conditions, a feature absent in regions like California. Estonia has also shown agility in adapting legislation to safely permit the testing of autonomous vehicles on its streets. Furthermore, the nation boasts a dynamic ecosystem of companies specializing in autonomous technology, including Starship, AuveTech, Clevon, and Milrem Robotics. The University of Tartu's Autonomous Driving Lab serves as a central hub for self-driving technology research and development.

Our vision for Tartu includes several key components:

  1. Designated testing zones for autonomous vehicles, encompassing both specialized closed areas and marked public city spaces.
  2. A comprehensive high-definition map of Tartu, featuring a detailed spatial point cloud and lane-level road network.
  3. A digital twin, or simulation, of Tartu, facilitating pre-arrival testing.
  4. Machine-readable traffic lights throughout Tartu, enhancing autonomous system safety beyond traditional light signals.

Open-source tools like QGIS, Shapely, Blender, and the CARLA autonomous driving simulator, alongside open datasets from the Estonian Land Board and the City of Tartu, have been instrumental in achieving the high precision required for our high-definition map and digital twin. These resources have been vital in realizing our vision with decimeter-level accuracy.

In this talk, I will provide an overview of how Tartu leverages open-source software and open data in developing the high-definition map and digital twin, key components in our journey to become a leader in self-driving technology.

Use cases & applications
GEOCAT (301)