Workshop on Online Map Validation and Road Model Creation.

A workshop being held in conjunction with the
IEEE Intelligent Vehicles Symposium - IV2021, July 11th, 2021, Nagoya, Japan + Zoom.
Free registration to the workshop!
Workshop's Zoom link

Call for Papers

Description

Scope


The goal of this workshop is to form a platform for exchanging ideas, fostering research and connecting both industry and academia in the domain of offline map and online road model creation and validation for autonomous driving.

This workshop will provide the opportunity to discuss about scenarios (highways and urban environments), ways of data collection (i.e. single vehicle and vehicle fleets) as well as data processing (in-vehicle and backend).

To this end we welcome contributions related to the following topics:

Validated Road Models

Creation of validated road models both at runtime and offline, both from single vehicles as well as from a fleet of vehicles.

Map Data Life-Cycle Management

Life-cycle management of map data, i.e. initial creation, change detection, update, verification and deployment.

Mapping Fraud Detection and Prevention

Mapping fraud detection and prevention, i.e. manipulation of traffic signs, fleet data spoofing.

Accepted papers will be published in the conference proceedings and indexed by IEEE Xplore!

Invited Speakers

We are proud to welcome the following invited speakers.

From HD Maps to No Maps for Autonomous Driving
Prof. Dr. Wolfram Burgard
Vice President & Professor
Automated Driving at Toyota Research Institute and University of Freiburg
Monitoring of Perception Systems for Certifiable Autonomous Vehicles
Prof. Luca Carlone
Assistant Professor
Massachusetts Institute of Technology
HD maps – Mobile mapping and modelling
Dr. Dejan Vasic
General Manager
DataDEV GmbH
Data-Efficient Perception for Autonomous Driving
Dr. Xiaodong Yang
Principal Scientist & Head of Research
QCraft

Call for Papers

We are soliciting high quality papers covering, but not limited to, the following topics:


We accept the following forms of contributions:

  • Research papers
  • Data sets

Data set submissions shall include (1) one data set with labelling, (2) a supporting paper (to clarify the importance of the data set, a description on how to use it, and what features of map validation and road model approaches can be best evaluated with the proposed data set), and (3) the code that exemplifies how to access the data. Authors submitting a data set will be asked to submit and share their data via the IEEE DataPort.

Paper length should not exceed 6 pages (two additional pages allowed with a fee) according to the IEEE IV 2021 Guidelines.

Each paper will undergo a peer-reviewing process by at least three independent reviewers. Contributions will be reviewed according to relevance, originality and novel ideas, technical soundness and quality of presentation.

Due to the pandemic situation we expect the workshop to be held either purely virtually or in a hybrid format with the possibility of in-person attendance.

Submission Deadline May 10th, 2021
Submission through PaperPlaza
Workshop code: 3n685

Notification Date May 15th, 2021

Camera-ready Deadline May 31st, 2021

Rationale

Safety and comfort of autonomous vehicles (AVs) critically depend on the availability of accurate environment models. At runtime, data coming from maps and from sensors are fused together with the goal of providing a unique and consistent view of the environment to software components responsible for maneuver and trajectory planning. One integral part of this view is the road model, which is a geometric and semantic description of the road around the AV, composed of lanes, traffic lights/signs, traffic rules and right of way. From the perspective of a fusion algorithm, maps are sensors with an almost unlimited field of view (FOV) and with a large, potentially unknown, perception delay. The unlimited FOV provided by maps allow AVs to speculatively plan upcoming maneuvers and trajectories on those areas not observed by other sensors yet. In urban scenarios this is extremely important, since traffic can strongly occlude the actual FOV of other sensors. Still, the unknown delay affecting map data may yield to conclusions about the environment model based on outdated information.
For example, geometry of roads can change due to construction work. Thus it is imperative to either verify the validity of map data before fusing them with sensor data, or to build an online road model from in-vehicle sensors alone. Although many approaches for online validation of map exist, e.g, based on Simultaneous localization and mapping (SLAM), very few are suited at validating maps with the stringent accuracy requirements typically found in AV applications. The desired accuracy of maps used by AVs is typically in the range of a few centimeters. Also, algorithms responsible for map validation, need to be able to validate an area of a map, before the AV enters that area. Finally, the large amount of data contained in the high-definition (HD) maps used by AVs, pose a variety of problems in satisfying performance requirements at runtime. Additionally, approaches for online road model creation typically assume highway scenarios and are thus not applicable to urban environments.

Call for Papers

Agenda

Below the workshop agenda, which will be held online on July 11, 2021 starting at:
21:00 JPT (UTC+9) - 14:00 CEST (UTC+2) - 8:00 EDT (UTC-4) - 5:00 PDT (UTC-7).
All the times in the agenda below are in Nagoya time zone, JPT (UTC+9).

Begin End Title Speaker/Author Type
21:00 - 21:10 Welcome Session Angelo Cenedese, Umar Zakir Abdul Hamid, Luca Parolini, Christopher Plachetka, Sebastian Schneider, Qing Rao, Oliver Wasenmüller, Roberto G. Valenti, Jenny Yuan
21:10 - 21:50 From HD Maps to No Maps for Autonomous Driving [Slides] Prof. Dr. Wolfram Burgard Invited Talk
21:50 - 22:10 HD Map Error Detection Using Smoothing and Multiple Drives [Slides] Welte et al. Paper
22:10 - 22:50 Monitoring of Perception Systems for Certifiable Autonomous Vehicles [Slides] Prof. Dr. Luca Carlone Invited Talk
22:50 - 23:10 Online and Adaptive Parking Availability Mapping - An Uncertainty-Aware Active Sensing Approach for Connected Vehicles [Slides] Varotto, Cenedese Paper
23:10 - 23:25 Break
23:25 - 00:05 HD maps – Mobile mapping and modelling [Slides] Prof. Dr. Dejan Vasic Invited Talk
00:05 - 00:25 Towards Knowledge-Based Road Modeling for Automated Vehicles - Analysis and Concept for Incorporating Prior Knowledge [Slides] Fricke et al. Paper
00:25 - 01:05 Data-Efficient Perception for Autonomous Driving [Slides] Dr. Xiaodong Yang Invited Talk
01:05 - 01:25 Use of Probabilistic Graphical Methods for Online Map Validation [Slides] Fabris et. al. Paper
01:25 - 01:35 Closing Remarks Angelo Cenedese, Umar Zakir Abdul Hamid, Luca Parolini, Christopher Plachetka, Sebastian Schneider, Qing Rao, Oliver Wasenmüller, Roberto G. Valenti, Jenny Yuan

Organizers

We do our best to make this workshop a success.

Angelo Cenedese received the M.Sc. and the Ph.D. degrees from the University of Padova, Italy, where he is currently an Associate Professor with the Department of Information Engineering. He is founder and leader of the SPARCS (SPace-Aerial-gRound Control Systems) research group. He has held several visiting positions at the UKAEA-JET laboratories in the Culham Research Centre (UK), the UCLA Vision Lab (CA-USA), the F4E European Agency (Spain). His research interests include system modeling, control theory and its applications, multiagent systems, and mobile robotics, including autonomous vehicle systems. On these subjects, he has published more than 150 papers and holds three patents.
Angelo Cenedese

University of Padova
A PhD holder, Umar Zakir Abdul Hamid has been working in the autonomous vehicle field since 2014 with various teams in different countries (Malaysia, Singapore, Japan, Finland). He is now the Team Lead of the Autonomous Vehicle Planning and Control Algorithm Development at Sensible 4 Oy, Finland - leading a team of 12 engineers. Umar Zakir is an active member of several Society of Automotive Engineers (SAE) Task Forces, focusing on the autonomous vehicle active safety topics. He is also one of the recipients for the Finnish Engineering Award 2020 for his contributions to the development of all-weather autonomous driving solution with Sensible 4.
Umar Zakir Abdul Hamid

Sensible 4
Luca Paroloni received the B.Sc. degree in information engineering and the M.Sc. degree in automation engineering from the University of Padova, Padova, Italy, in 2004 and 2006, respectively and the Ph.D. degree in electrical and computer engineering from Carnegie Mellon University, Pittsburgh, PA, in 2012. He is a specialist S/W developer at BMW AG, Munich, Germany, which he joined in 2016. His research interests include functional development for autonomous vehicles, with a special focus on map and localization problems, functional safety, and functional degradation.
Luca Parolini

BMW
Christopher Plachetka received his Master degree in 2017 from Technical University of Braunschweig, Germany, with a major focus on automotive systems and pattern recognition. In his master thesis, he focused on deep learning-based object detection and automatic labeling of datasets, which lead to the publication of the TUBS Road User Dataset during his time as scientific assistant in 2018. Since 2019, Christopher works at Volkswagen AG, Wolfsburg, Germany, as a PhD candidate. His research interests include deep learning-based deviation detection between sensor and map data, LiDAR-based object detection, and road model creation.
Christopher Plachetka

Volkswagen
Qing Rao received his Bachelor's degree in 2010 from Shanghai Jiao Tong University and his Master's degree in 2012 from TU Munich, with a major focus on computer vision and robotics. In 2019, he received his Ph.D. degree from TU Munich with the dissertation entitled Merging the Virtual and Real in a Car: In-Vehicle Augmented Reality. Since mid-2017, Qing works at BMW AG, Munich, Germany, as a machine learning expert in the area of autonomous driving. His current research interests include road model generation, 3D object detection, and active learning.
Qing Rao

BMW
Sebastian Schneider received his Diploma degree in 2006 from Technical University of Darmstadt, Germany, with a major focus on computer vision and robotics. Since 2014, Sebastian works at BMW AG, Munich, Germany, as a sensor fusion expert in the area of driver assistance systems and autonomous driving. As such he has contributed to the design of the sensor setup as well as the sensor fusion architecture of upcoming level 4 autonomous vehicles. His current research interests include localization, map and road model validation and reinforcement learning.
Sebastian Schneider

BMW
Oliver Wasenmüller is full Professor at the Mannheim University for Applied Science. His research is in the intersection of Computer Vision and Artificial Intelligence with a focus on automotive. Previously he was a team leader for "machine vision and autonomous vehicles" at the German Research Center for Artificial Intelligence (DFKI). He is both speaker and reviewer in many scientific conferences in this field and co-organizes the ACM Computer Science in Cars Symposium (CSCS) as well as the IEEE CVPR workshop SAIAD.
Oliver Wasenmüller

Mannheim University for Applied Science
Roberto G. Valenti is currently a Senior Research Scientist at MathWorks where he is responsible for autonomous driving, robotics, and deep learning. His research interests include sensing for navigation, sensor fusion, autonomous vehicles (self-driving cars, unnamed aerial vehicles), inertial navigation and orientation estimation, control, computer vision, and deep learning. Previously, he worked as a Research and Development Engineer within the Autonomous Driving team at Nvidia. He obtained a Ph.D. in Electrical Engineering at the City University of New York, The City College, NY, USA where he focused his research on state estimation and control for autonomous navigation of micro aerial vehicles. Dr. Valenti received his M.Sc. in Electronics Engineering from the University of Catania, Italy. He is a member of IEEE and RAS.
Roberto G. Valenti

MathWorks
Jenny Yuan holds a master degree from Technical University of Munich in Germany. Her major research direction is related to deep-learning and image processing in the field of computer vision, such as object detection and classification.
Jenny Yuan

BMW