The former basketball court on the campus of the University of Stuttgart in Vaihingen has been declared a test track without further ado. The three SimTech doctoral researchers Robin Strässer, David Meister, and Marc Seidel as well as Felix Brändle, who are all pursuing their doctoral degree studies at the Institute for Systems Theory and Automatic Control (IST) under the supervision of Prof. Dr.-Ing. Frank Allgöwer, have set up a laptop and brought along a prototype of an autonomously driving e-scooter. This is a normal e-scooter, the kind you can buy or rent and can now be found in almost every city. But this e-scooter has a special feature.
A casing is attached to the bar above the front wheel, behind which a number of technical components are hidden: a steering motor so that the scooter can drive curves autonomously, a reaction wheel to keep it in balance and prevent tipping over while driving, a GPS receiver and other sensors, small computers that evaluate everything, electrical components for the power supply and, of course, cables.
The casing of the e-scooter is the work of the four doctoral researchers and a team of students who are working on how the e-scooter can avoid collisions. During a test drive around the basketball court, Robin Strässer demonstrates the development of the e-scooter and crosses its path: The e-scooter is slowing down autonomously and comes to a halt in front of him. To enable the scooter to detect obstacles, the four early career researchers attached three ultrasonic sensors to the front of the casing that emit and reflect signals and can therefore measure distances.
Reliable data despite sensor failure and measurement errors
"However, the data measured by the ultrasonic sensors is sometimes faulty," explains Robin Strässer. "That's why we use three independent sensors. In this way, we can ensure that we’ve got correct data, at least in total, and can measure the real distance, even if individual data is incorrect." For this purpose, the signals from the ultrasonic sensors are filtered. Filtering means: Several measurements are combined and the filtered data is smoothed over a certain period of time. This means that the measured "outlier" data is recognized and won’t be taken into account when the actual distance is calculated. In this way, the early career researchers ensure that the obstacles are reliably detected and that the e-scooter stops safely and autonomously.
The project has been researched at the IST since 2019. The four young scientists are already the third generation of doctoral researchers to work on the e-scooter alongside their doctoral degree studies - voluntarily, as the e-scooter project is separate from their respective doctoral projects. "My doctoral project is very theoretical. That's why the cool thing about the e-scooter is that we can put theory into practice and immediately see how the physical system behaves," says Robin Strässer. He also enjoys the opportunity to gain different insights by coordinating the various projects in this relatively large team.
In addition to the four doctoral researchers, there are twelve other people working on the project. These are students who help to develop new functionalities for the prototype or who are implementing a sub-project of the autonomously driving e-scooter in the context of their Bachelor's or Master's thesis. Various projects from different disciplines such as computer science, control engineering, and mechanics have to be brought together so that the overall system works. “My studies of Simulation Technology helped me a lot, because content from many disciplines is also covered there," adds Robin Strässer.
The goal is a sustainable campus
The autonomous e-scooter is a sub-project of MobiLab, with the aim of achieving emission-free mobility on the campus of the University of Stuttgart. "Our e-scooter is now the third prototype since the start of the project," says Marc Seidel. The first prototype was already able to stabilize itself and drive straight ahead at a slow speed. The second prototype focused on developments relating to driving curves. The third prototype is aimed at driving to the next location autonomously while stopping if an obstacle gets in its way or a person crosses its path, even at higher speeds. The e-scooter can be driven by one person at a speed of 20 kilometers per hour, manually controlled like standard e-scooters. When it drives autonomously, it travels at walking speed, i.e., at around five kilometers per hour.
Using simulations, the four doctoral researchers determined how the ultrasonic sensors need to be arranged in order to cover a specific field of vision in front of the e-scooter with the employed signal detection, so that no obstacles remain in a blind spot. "Our aim now is to increase the field of vision of the sensors even further," Robin Strässer explains. "Because when there is a large curve radius or when the speed is faster, it’s important to have a wider field of vision so that the e-scooter can still detect an obstacle in the curve and stop safely." Additional sensors would therefore have to be installed.
However, the autonomous e-scooters will not only make the campus of the University of Stuttgart more sustainable. The problem of e-scooters, which are on the road in large numbers in many cities, blocking sidewalks or lying around somewhere, could be solved if they were able to distribute autonomously or drive to the charging station on their own. This would not only reduce the emissions of the vehicles that have to collect the e-scooters, but also the number of e-scooters.
The autonomous e-scooter is being developed at the Institute for Systems Theory and Automatic Control under the name eStarling.io and under the leadership of SimTech's Principal Investigator Prof. Dr.-Ing. Frank Allgöwer. The project is part of the Mobility Living Lab of the University of Stuttgart.
Read more
Strässer, R., Seidel, M., Brändle, F., Meister, D., Soloperto, R., Hambach Ferrer, D., & Allgöwer, F. (2024). Collision Avoidance Safety Filter for an Autonomous E-Scooter using Ultrasonic Sensors. Accepted for presentation at the 17th IFAC Symposium on Control in Transportation Systems (CTS 2024), Preprint: arxiv:2403.15116.
About the doctoral researchers
David Meister (left) studied mechanical engineering at Technical University of Darmstadt (TU Darmstadt), Marc Seidel (2nd from left) and Felix Brändle (3rd from left) studied Engineering Cybernetics at the University of Stuttgart, and Robin Strässer (right) studied Simulation Technology at the University of Stuttgart.
All four are now pursuing their doctoral degree studies at the Institute for Systems Theory and Automatic Control at the University of Stuttgart - David, Marc, and Robin in SimTech projects. The research on the e-scooter is separate from the topics of their doctoral projects, which have a theoretical focus, but offers an exciting balance as practical research.