Event

SAFER ONLINE SEMINAR: NEW SAFER COMPETENCE NETWORK; PERCEPTION, SENSING & COMMUNICATION

Date
4 March 2021 11:30-12:30
Place
ONLINE - MICROSOFT TEAMS

The SAFER Thursday seminar series continues with presentations connected to SAFER’s new competence network; Perception, Sensing & Communication. Competence network leader Lars Hammarstrand, Assistant Professor at the Signal Processing Group, Electrical Engineering at Chalmers University is hosting this seminar.
______________________________________________________________________________________

Semantics- and Uncertainty-aware Perception for Autonomous Driving 
Speaker: Eren Erdal Aksoy, Associate Senior Lecturer, School of Information Technology, Halmstad University

Scene understanding is an essential prerequisite for autonomous vehicles. Semantics helps gaining a rich understanding of the scene by assigning a meaningful class label for each individual sensory data point. Safety-critical systems, such as self-driving vehicles, however, require not only highly accurate but also reliable semantic predictions with a consistent measure of uncertainty. This is because the quantitative uncertainty measures can be propagated to the subsequent units, such as decision-making modules to lead to safe maneuver planning or emergency braking, which is of utmost importance in safety-critical systems. Therefore, semantic predictions integrated with reliable confidence estimates can significantly reinforce the concept of safe autonomy. 

In this talk, I will introduce our recent neural network architecture, named SalsaNext, which can achieve uncertainty-aware semantic segmentation of a full 3D LiDAR point cloud in real-time.  I will present various quantitative and qualitative experiments on the large-scale challenging Semantic-KITTI dataset showing that SalsaNext significantly outperforms other state-of-the-art networks in terms of pixel-wise segmentation accuracy while having much fewer parameters, thus requiring less computation time. I will end my talk by arguing how semantics can be employed to implement generic neural networks solving domain translation problems, e.g. by converting raw 3D LiDAR point clouds to panoramic color RGB images. Such domain translations play a crucial role in solving sensor failures in autonomous driving.  

Technologies for wireless V2X communication: update on recent development 
Speaker: Erik Ström, Professor, Electrical Engineering, Chalmers

There is a heated debate about the pros and cons for IEEE (802.11p/802.11bd/…) and 3GPP (4G/5G/…) standards for vehicle-to-anything (V2X) communications. In this talk, we give a status update on the current technologies and try to provide some unbiased observations on the technologies. In particular, the enhancements to 802.11p suggested in the upcoming 802.11bd standard will be discussed.

Welcome!
_____________________________________________________________________________________

This seminar is only open for SAFER partners. Missing your invitation? Please contact Mikael von Redlich.

Info

Contact
Mikael von Redlich
Email
redlich [at] chalmers.se
Category
Seminar