The STELLA Efficient Mobility group explores issues in intelligent transportation. Our work centers around the energy-efficient control and design of vehicles, and their relationship with their passengers. Our focus is on bringing state-of-the-art machine learning solutions to traditional transportation problems.
Our experiments are conducted using high-performance electrically assisted velomobiles. These light-weight, aerodynamic vehicles act as a test-bed for ideas about the larger transportation picture. With a weak motor compared to the its weight, the pull of gravity caused by the slope of the road has an outsized influence on the motion of the vehicle. Controllers designed to take advantage of, and compensate for, this effect can yield large savings in energy consumption, and be applied to vehicles which are similarly effected, including some of our most energy hungry vehicles: trains, trucks, and buses.
Creating these controllers requires a deep understanding of how the vehicle behaves in the real world. Designed primarily by hobbyists and small companies, velomobiles lack the sophisticated aerodynamic and motor models produced for traditional vehicles by generations of engineers. We take this as an opportunity to explore innovative techniques for creating these models by taking advantage of new computational tools and advances in machine learning. We are most interested in the use of real world data, rather than data collected in highly controlled environments such as wind tunnels. We believe that modeling techniques which can take into account how vehicles behave as we drive them, not just as the could be driven, will have widespread application.
Modeling the vehicle itself is only one aspect of our research. By collecting data from riders as they ride we are building models of how our bodies react to training, in both the long and short term. This allows us to not only support riders with electrical-motor assistance to avoid exhaustion on individual trips, but for long-term training plans to be developed so that riders get the most out of their efforts. Coupled with intelligent motor controllers, a day’s training can be
customized, even without changing routes.
The Stella project is composed of researchers from a variety of backgrounds and disciplines at the Bonn-Rhein-Sieg University of Applied Sciences. Our foremost concern is to bring our work into reality and affect change through innovation. We are always eager to cooperate with those in industry, government, and like-minded academic institutions.
Project manager at H-BRS
- Helge Spieker, Alexander Hagg, Adam Gaier, Stefanie Meilinger, Alexander Asteroth. Multi-stage evolution of single- and multi-objective MCLP, Successive placement of charging stations. Soft Computing Journal, Springer, 2016, to appear
- Matthias Füller, Ashok Meenakshi Sundaram, Melanie Ludwig, Alexander Asteroth and Erwin Prassler. Modeling and Predicting the Human Heart Rate During Running Exercise, in Information and Communication Technologies for Ageing Well and e-Health, pages 106-125, Springer, 2015
- Alexander Asteroth and Alexander Hagg. How to successfully apply genetic algorithms in practice: representation and parametrization. In International Symposium on Innovations in Intelligent Systems and Applications, 2015.
- David Schaefer, Alexander Asteroth, and Melanie Ludwig. Training plan evolution based on training models. In International Symposium on Innovations in Intelligent Systems and Applications, 2015.
- Adam Gaier and Alexander Asteroth. Evolution of optimal control for energy-efficient transport. In Intelligent Vehicles Symposium, Proceedings, pages 1121–1126. IEEE, 2014.