The concept of integrity was firstly formalized and developed for GNSS in the aviation domain for Safety-of-Life applications. It is defined as “a measure of trust which can be placed in the correctness of the information supplied by the total system”. For highly automated vehicles (e.g., SAE L3 and above), integrity is crucial for safe navigation. The integrity requirement concerns the system’s capability to provide statistical guarantees that a computed piece of information is not misleading and more particularly that it will not mislead the system into a hazardous situation. That is, the system needs to model and bound the uncertainty of its estimates and its validity in terms of the operational domain and conditions. For AVs to drive safely in varying environments, they need to gather and evaluate localization information from different sensors, such as LiDAR, IMU, GNSS, and camera. The navigation of land vehicles is more complex than aviation applications. AVs require accurate positioning to operate, e.g., decimeter-level for highways and centimeter-level for urban environments. The randomness encountered in road networks like changes in the infrastructure, weather conditions, sensor masking and occlusions make high-integrity localization more difficult. Moreover, interference, such as occlusions, by neighboring road users like pedestrians, cyclists, and other vehicles also needs to be continuously considered during the localization. Currently, there is an extensive use of position estimates in passenger vehicles, beyond driver guidance, the introduction of actuating driving assistance systems has intensified the reliance on localization systems.
The goal of the i.c.sens iLoc workshop is to emphasize the importance of integrity in AVs and to address the scientific challenges of managing localization integrity for vehicle navigation in complex traffic environments including its use as part of perception tasks.
2nd Edition of iLoc workshop, Bilbao, Spain 2023
1st Edition of iLoc workshop, Aachen, Germany 2022
Integrity concepts and methods for multi-sensor system including GNSS, LiDAR, IMU, camera, and HD map
Modelling uncertainty for localization, mapping, and perception in dynamic environments
Standards and requirements for integrity, continuiuty, and accuracy in autonomous driving
08:30 - 08:50
Opening Workshop Navigating Uncertainty
08:50 - 09:20
Invited Talk by Guohao Zhang
Assisting GNSS Fault Detection and Exclusion in Urban Areas by Deep Learning
09:20 - 09:40
PL-RAS: A Robust Localization System with Real Time Protection Level Calculation and Adaptive Kernel for Enhanced Integrity
(Elias Maharmeh, Fazwi Nashashibi, Zayed Alsayed)
09:40 - 10:00
Towards Provably Reliable Uncertainty Quantification
for Automotive State Estimation
(Penggao Yan, Li-Ta Hsu)
10:00 - 10:30
Invited Talk by Ilaria Martini
Robust and resilient PNT for autonomous driving
11:00 - 11:30
Invited Talk by Gerhard Dorn
Lessons-learned from embedding occupancy grids with Julia
11:30 - 12:00
Invited Talk by Abhinav Valada
Learning for Open-World Autonomy
12:00 - 12:20
MUVO: A Multimodal Generative World Model for Autonomous Driving
with Geometric Representations
(Daniel Bogdoll, Yitian Yang, Tim Joseph, Melih Yazgan, J. Marius Zöllner)
13:30 - 14:00
Invited Talk by Thomas Griebel
Modeling and Encompassing Uncertainty: A Unified Self-Assessment Framework for AD Stacks Using Subjective Logic
14:00 - 14:20
Lidar Pole Detection Training using Maps for Localization
(Maxime Noizet, Philippe Xu, Philippe Bonnifait)
14:20 - 14:40
Traffic Participant Behavior Prediction based on a
Dynamic Graph Neural Network
(Ning Qian, Yiming Xu, Monika Sester)
14:40 - 15:10
Invited Talk by Igor Gilitschenski
Do Androids Dream of Electric Sheep? A Generative Paradigm
for Dataset Design
15:10 - 15:30
Adaptive Urban Navigation Using Graph-Based Markov Decision Chains
(Tiago Caldeira, Majid Khonji, Jorge Dias, Pedro U. Lima)
16:00 - 16:30
Invited Talk by Shane Murray
Radar-based Occupancy Grids for Autonomous Driving
16:30 - 16:50
Wrap-up