Autonomous Driving Meetup #22 Munich GARCHING

22nd Autonomous Driving Meetup - hosted by driveblocks in Munich

Join us, enjoy the the talks and socializing!

Getting there: Take U6 to Garching hochbrück+ short walk

Please note that seats are limited. First come first serve :-)

___ Agenda ___

_ Introduction & Welcome

_ Transformer NNs & Explainable Sensor-fusion for Robust Perception
Alexander Wischnewski, driveblocks

_ Simulation for Semantic Self-labeling & Self-driving
Prof. Antonio Lopez, ICREA UAB CVC (more details below)

_ The superposition state of autonomous driving: simultaneously ready and not ready
Prof. Mirko Mählisch, UniBW

_ Socializing

___ Further Details ___

_ If you join the Meetup, you give declaration of consent that pictures will be taken
_ The talks are all held in English and will take ~25 minutes each + ~10 minutes Q&A
_ Sponsor: Warmest thanks to driveblocks for making this meetup possible
_ Snacks and drinks will be provided
_ Exact time slots for the presentations may change

---
Antonio Lopez - Bio and Abstract

Speaker:
Dr. Antonio M. López
ICREA Academia Professor & Principal Investigator
Computer Science Dpt. & Computer Vision Center (CVC)
Universitat Autònoma de Barcelona (UAB)
Short Bio:
Antonio has a long trajectory carrying research at the intersection of computer vision, simulation, machine learning, driver assistance, and autonomous driving. Antonio has been deeply involved in the creation of the SYNTHIA dataset and the CARLA open-source simulator, both for democratizing autonomous driving research. He is actively working hand-on-hand with industry partners to bring state-of-the-art techniques to the field of autonomous driving.
Title:
Simulation for semantic self-labeling and self-driving
Abstract:
Developing autonomous vehicles involves training and testing deep drivers by using supervised data collected on a diversity of driving episodes. We could say that data is the driver in autonomous driving. In this context, the talk focuses on the efforts conducted at CVC/UAB to minimize human data labeling. We introduce two recent works. On the one hand, we present, for the first time, a new synthetic but photo-realistic dataset. We will show how it can be combined with other synthetic datasets and semi-supervised learning techniques to bring state-of-the-art results in onboard semantic segmentation. On the other hand, we will show how a digital twin and CARLA simulator allowed us to perform autonomous driving at narrow mountain roads located at the Catalan Pyrenees.