Simulator Showdown. Pitch Your Virtual Ride.
With autonomous driving on the horizon, new research challenges appeared and subsequently, new methods and research instruments became necessary. To adapt to these emerging research questions, driving simulators, the cornerstone of automotive human factors research, have been tweaked, modified or developed from scratch. This one-day workshop invites academics and practitioners to report, demo or discuss their solutions for simulating the next wave of automotive interaction research. The goal of this workshop is twofold: (1) we provide a forum for researchers concerned with simulator software and discuss opportunities for a future collaboration platform for sharing and co-develop simulator software. (2) We collect and discus the needs, expectations and solutions of the automotive UI community to articulating a road map for the developing future driving simulators setups.
Driving simulators play a critical role in investigating driver behavior in a safe and controlled environment. Without doubt, driving simulators are one of the most important instruments for human factors researchers within the automotive domain ranging from simple desktop simulators to multidimensional, highly-immerse simulator environments. However, with recent advancements in autonomous driving the research questions and subsequently the requirements of an suitable driving simulator are changing rapidly. To adapt to the challenges and needs, researchers in academia and industry are required to develop new simulation environments or hack existing solution to guide their research.
In the following we introduce a few of the organizers' developments on driving simulator technologies.
This project enable the experience of realistic felt motion inside of a simulated environment. By using a real vehicle driving on an virtual course we can generate a wide variety of scenarios. The motion and sound perceived by the participants are real and just the visual modality is replaced by a normal Virtual Reality headset. This realistic motion is especially useful for either sublet scenarios (e.g. driving style preference, motion cues) or emergency scenarios (e.g. take over and crash avoidance).
Developed to enable rapid prototyping of vehicle cockpit configurations facilitating the integration of multi-modal sensors with Web-based User Interfaces to create event-reactive UIs. Skyline has been also applied to induce and study passenger emotional reactions under automated driving conditions.
The miniSim driving simulator has powerful scenario control and data acquisition capabilities based on 20 years of research and development at the University of Iowa's National Advanced Driving Simulator (NADS). The miniSim team currently provides user support for over 60 sites, some with multiple simulators. Typical applications include human factors, driver state modeling, distraction, and impairment research.
The simulation environment enables fast creation of different scenarios including different types of vehicles, motorbikes, bicycles and pedestrians. It supports data acquisition from multiple biometrical sensors and eye trackers. With proprietary camera-based head tracking system, it enables automatic mapping of gaze position to screen coordinate system and consequentially automatic gaze analysis.
CARLA was developed to study End to End development of Automated Driving algorithms from Perception to Actuation in realistic Urban Scenarios. CARLA has become a standard tool for Automated Driving curriculums in Computer Science and Automotive Engineering schools. It provides multi-sensor input and flexible scenario creation. Also recently we added pedestrian control and safety restrictions for AV planning.
Maps and data
To study new ways of interaction with the physical world surrounding the car and to conduct research in autonomous driving scenarios, the use of High Definition maps and the integration of georeferenced data coming from the real world (POIs, Venues, etc) into simulators are key elements. By matching the virtual simulation with ground-truth data and integrating multimodal sensors in the simulation environment, it is possible to develop context-aware UIs, end-toend simulations, and to test natural user interfaces against realistic conditions.
Call for Participation
Participants interested in presenting their simulator setups, tools or challenges are invited to submit a 2-4 pages position paper (ACM SIGCHI Extended Abstract template) that specifies one or all of the following aspects:
- Tell us what you have, use or build and show us what it can do!
- Share what tools or modifications you developed that makes your system special!
- What research questions do you address with your simulator and what are the expected challenges in term of technologies or content?
- Tell us where and what we need to download to demo your system!
Please send the position paper to firstname.lastname@example.org.
Roughly the half of all accepted full-papers of the last conferences (AutoUI '16 '17 '18) explicitly mentioned to conduct a simulator study in their abstracts. Despite trending towards autonomous driving topics, simulation technologies are and will be a critical component of any automotive-related interaction research. To do so, researchers often struggle to adapt their simulator software to the emerging themes such as autonomous driving, passenger entertainment or interaction with multiple road users. The modifications necessary, can be simple changes of code but also substantial technically advanced extensions that are worthy to be shared with a broader community.
Furthermore, for automotive UI researchers, there is only little opportunity to share and collaborate on their simulator systems. The systems are often outdated, not properly maintained or simply technical incompatible to be productively shared and supported by a broader community. With this workshop we want to provide a first overview of approaches and tool-kits used in automotive UI research, reveal tools and hacks dedicated to facilitate novel research questions and finally aim to discuss the relevancy of a mutual platform for a closer collaboration.
The workshops central objective is to gain an overview of simulation approaches and technologies used in current automotive UI research. As an interactive workshop participants and organizers are asked to pitch or demo one of their simulation solutions. Topics and questions of potential interests include:
- What simulator platforms are out there? What are the pros and cons?
- Which simulator platform or modification are best suited for addressing which research questions?
- What is the role hybrid simulators and what can we learn from them?
- Show us how you changed, modified or hacked an simulator platform to make it ready for autonomous driving research?
- Design for immersion: How to transform simulation technology into a realistic scenario?
- How to make systems modular, shareable, accessible and reusable for the entire research community?
- What extensions and standards should be developed?
- Present your "tiny-tool" that make life easy running simulator studies: from data collection to analysis and system development!
We aim for two tangible outcomes. By connecting researchers and collecting simulator projects and toolkits we articulate a road map for developing a platform that allows automotive UI researchers to share and collaborate on simulator projects. Secondly, we plan a mutual publication as a special issue on simulators for the autonomous world. This publication contains not only the organizers' simulator approaches but also invite the participant to contribute with their system. Besides a detailed overview of relevant simulator project, we discuss a taxonomy for simulator technologies and how their address the emerging research questions on the autonomous world.
Both outcomes are intended to establish a closer collaboration among the automotive UI researchers and their approaches on simulation technologies and promote an open research and development culture that benefit the next wave of simulator technologies.
- Sven Krome Uber ATG, Pittsburgh, PA, USA
- Eric Deng Uber ATG, San Francisco, CA, USA
- David Goedicke Cornell Tech, New York, NY, USA
- Wendy Ju Cornell Tech, New York, NY, USA
- Ignacio Alvarez Intel Labs, Portland, OR, USA
- Jaka Sodnik University of Ljubljana, Slovenia
- Andrew Veit National Advanced Driving Simulator, Iowa City, IA, USA
- Francesco Grani HERE Technologies, Berlin, Germany