Industry as well as academia have made great advances working towards an overall vision of fully autonomous driving. Despite the success stories, great challenges still lie ahead of us to make this grand vision come true. On the one hand, future systems have to be yet more capable to perceive, reason and act in complex real world scenarios. On the other hand, these future systems have to comply with our expectations for robustness, security and safety. ACM, as the world’s largest computing society, addresses these challenges with the ACM Computer Science in Cars Symposium. This conference provides a platform for industry and academia to exchange ideas and meet these future challenges jointly. The focus of the 2018 conference lies on AI & Security for Autonomous Vehicles. Contributions centered on these topics are invited.
Registration is now open here.
Tickets are limited.
Submission of contributions are invited in (but not limited to) the follow key areas:
- Artificial Intelligence in Autonomous Systems: Sensing, perception & interaction are key challenges — inside and outside the vehicle. Despite the great progress, complex real-world data still poses great challenges towards reliable recognition and analysis in a large range of operation conditions. Latest Machine Learning and in particular Deep Learning techniques have resulted in high performance approaches that have shown impressive results on real-world data. Yet these techniques lack core requirements like interpretability.
- Automotive Security for Autonomous driving: Autonomous cars will increase the attack surface of a car as they not only make decisions based on sensor information but also use information transmitted by other cars and infrastructure. Connected autonomous cars, together with the infrastructure and the backend systems of the OEM, constitute an extremely complex system, a so- called Automotive Cyber System. Ensuring the security of this system poses challenges for automotive software development, secure Car-to-x communication, security testing, as well as system and security engineering. Moreover, security of sensed information becomes another important aspect in a machine learning environment. Privacy enhancing technologies are another issue in automotive security, enforced by legislation, e.g., the EU General Data Protection Regulation. For widespread deployment in real-world conditions, guarantees on robustness and resilience to malicious attacks are key issues.
- Evaluation & Testing: In order to deploy systems for autonomous and/or assisted driving in the real-world, testing and evaluation is key. Giving realistic and sound estimates – even in rare corner cases – is challenging. A combination of analytic as well as empirical methods is required.
Important dates & logistics:
- Full paper submission deadline: May 28th 2018
- Extended abstract submission deadline: July 16th 2018
- Notification of acceptance (full papers): July 12th 2018
- Notification of acceptance (extended abstracts): July 31st 2018
- Camera ready full papers due: July 31st 2018
- Symposium: September 13th+14th 2018
- Location: Technikum, Munich, Germany (proximity of ECCV’18 main venue)
- Organizers: German Chapter of the ACM
The symposium will take place in Munich, Germany — in proximity to the European Conference on Computer Vision as a satellite event.
Raquel Urtasun, Uber ATG, Vector Institute, University of Toronto
Christoph Stiller, Karlsruhe Institute of Technology (KIT)
Thomas Wollinger, ESCRYPT GmbH, Germany
Jack Weast, Intel
September 13th: Security
11:15 Keynote Talk: Responsibility Sensitive Safety
Jack Weast, Chief Systems Architect for Automated Driving Solutions, Intel
12:15 Spotlight presentations of extended abstracts in Security
14:00 Oral Session: Security Evaluation of Autonomous Vehicles
- Attacker Model for Connected and Autonomous Vehicles
Jean-Philippe Monteuuis, Jonathan Petit, Jun Zhang, Houda Labiod, Stefano Mafrica, Alain Servel
- CyberSecurity Evaluation of Automotive E/E Architectures
Martin Ring, Davor Frkat, Martin Schmiedecker
15:00 Coffee break
15:30 Oral Session: Security for CAN
- CAN Obfuscation by Randomization (CANORa) – A technology to prevent large scale malware attacks on autonomous vehicles
Tobias Madl, Jasmin Brückmann, Hans-Joachim Hof
- Towards Viable Intrusion Detection Methods For The Automotive Controller Area Network
Andrew Tomlinson, Jeremy Bryans, Siraj Shaikh
- A Survey on Media Access Solutions for Controller Area Network Penetration Testing
Enrico Pozzobon, Nils H Weiss, Sebastian Renner, Rudolf Hackenberg
16:30 Keynote talk: The art of cyber security orchestration
Dr. Thomas Wollinger since 2007 Managing Director ESCRYPT GmbH.
17:30 Drinks & Reception
September 14th: AI
9:00 Keynote Talk: The Role of Artificial Intelligence in Automated Driving
Christoph Stiller, Institut für Mess- und Regelungstechnik KIT – Karlsruher Institut für Technologie, Germany
10:00 Oral Session: AI Safety
- AutoRVO: Local Navigation with Dynamic Constraints in Dense Heterogeneous Traffic
Yuexin Ma, Dinesh Manocha, Wenping Wang
- Sequential Attacks on Agents for Long-Term Adversarial Goals
Edgar Tretschk, Seong Joon Oh, Mario Fritz
10:45 Coffee Break
11:15 Oral Session: AI Sensing
- Real Time Single Image Dehazing and Soil Removal Using CNNs
Wajahat Akhtar, Sergio Roa Ovalle
- Classification of LIDAR Sensor Contaminations with Deep Neural Networks
Jyothish Karakkaparambil James, Vladislav Golyanik, Georg Puhlfürst
12:00 Spotlight presentations of extended abstracts in AI
12:30 Lunch & Posters
14:00 Keynote Talk: Raquel Urtasun, Uber ATG, Vector Institute, University of Toronto
15:00 Coffee Break
15:30 Panel Discussion
16:45 Closing Remarks
- Taking advantage of sensor modality specific properties in Automated Driving
- A Cognitive Assistant for Route Selection Using Knowledge Heuristics
Dan Cunnington, Geeth de Mel
- Long-Term On-Board Prediction of People in Traffic Scenes under Uncertainty
Apratim Bhattacharyya, Mario Fritz, Bernt Schiele
- Threading frameworks analysis with respect of production requirements, such as FuSa/ISO 26262
Maxym Dmytrychenko, Ilya Burylov
- Long-range obstacle detection from a monocular camera
Muhammad Abdul Haseeb, Danijela Ristic-Durrant, Axel Gräser
- Automated Scene Flow Data Generation for Training and Verification
Oliver Wasenmüller, René Schuster, Didier Stricker, Karl Leiss, Jürgen Pfister, Oleksandra Ganus, Julian Tatsch, Artem Savkin, Nikolas Brasch
- Dense Scene Flow from Stereo Disparity and Optical Flow
René Schuster, Oliver Wasenmüller, Didier Stricker
- Put the enterprise virtualization to automotive
- Risk Averse Robust Adversarial Reinforcement Learning
Xinlei Pan, John Canny
Confirmed Program Committee
- Submission: We are inviting industrial and academic participation in the event. We are looking for high-quality, original contributions to our peer reviewed “Full Paper” track with oral and poster presentations. The research papers must be formatted according to the acm-sigconf-authordraft template, which can be obtained from http://www.acm.org/publications/article-templates/proceedings-template.html. Page limit is 8 pages with an additional 9th page only containing references. Accepted papers will be published as a conference publication in the ACM Digital Library. Contributions have to be submitted in the “Full Paper” track by the deadline specified below at https://cmt3.research.microsoft.com/CSCS2018/
- Review Process: The review process is double blind, that is, authors do not know the names of the reviewers of their papers, and reviewers do not know the authors’ names. Avoid providing information in the submission that may identify the authors in the acknowledgments where possible (e.g., company, co-workers and grant IDs). Avoid providing links to websites that identify the authors.
- Submission: We are inviting submissions for the “Extended abstracts” tracks with poster presentation — with online publication (this does not count as references publication) — in the following 5 categories: demo, exhibitions, discussion papers, PhD position paper, and significant, already published work. The extended abstracts must be formatted according to the acm-sigconf-authordraft template which can be obtained from http://www.acm.org/publications/article-templates/proceedings-template.html Page limit is 2 pages with an additional 3rd page only containing references. Contributions have to be submitted in the “Extended Abstract” track by the deadline specified below at https://cmt3.research.microsoft.com/CSCS2018/
Submissions for “significant, already published work” are also acceptable in their original form (not subject to formatting or page constraints).
- Review Process: The review process is light-weight and single blind, that is, the authors, do not know the reviewers’ names, but the submission does not have to be anonymized