idw – Informationsdienst Wissenschaft
Nachrichten, Termine, Experten
Nachrichten, Termine, Experten
The prototype of an intelligent sensor system capable of automatically detecting cyclists was developed as part of the BikeDetect research project. To identify cyclists, the team used lidar systems, 3D cameras and thermal imaging. Field tests in urban traffic conditions demonstrated that the system is feasible.
An intelligent assistance system for cars and lorries that detects cyclists, measures the distance between vehicle and bike and emits a warning signal in dangerous situations can make cycling on roads safer. BikeDetect, a research project led by Professor Dr Jorge Marx Gómez, Chair of Business Information Systems at the University of Oldenburg, Germany, has laid the foundations for such a system. In a feasibility study carried out over 18 months, his team developed a prototype and then tested it in urban traffic in the North German city of Osnabrück. The researchers recently presented their findings.
The main objective was to create a sensor technology prototype that can reliably detect cyclists in road traffic using cameras as well as low-cost, non-camera-based sensors. The team tested ultrasonic, radar and optical sensors for distance measurements and LiDAR systems, 3D cameras and thermal imaging for detecting people on bikes. Based on these data it then trained several AI models to identify cyclists in the vicinity of a vehicle. “We were able to demonstrate that, in principle, this sensor system can detect cyclists, even though we could only collect limited amounts of data in this feasibility study,” explained operative project leader Johannes Schering from the University of Oldenburg.
The team first carried out tests in the lab and a car park to develop a test setup. Then in October it performed a two-day field test in real-world traffic conditions during which data was collected to train the AI systems. For this, the team mounted a metal arm fitted with selected sensors on the passenger side of a car and then drove the vehicle along a 22-kilometre route around the city of Osnabrück multiple times.
The project results showed that, on the basis of video data, AI models were able to detect cyclists in traffic with a fair degree of reliability. “The systems were particularly prone to errors when cyclists were in a group or obscured by trees, or when the distance between car and bike was either very large or very small,” Schering explained. The researchers also demonstrated that the reliability of the AI models can be enhanced by incorporating additional data such as that from thermal sensors into the training process. Among the distance-measuring sensors, the team identified a 360-degree LiDAR system, which uses laser beams to scan its surroundings, and a radar sensor as having the greatest potential. The data from the ultrasonic sensor was not usable.
The team also found that environmental conditions such as the weather and light levels were crucial for the system’s reliability. In addition, factors such as whether the test vehicle was moving or standing at a traffic light also affected the quality of the results. “Future driver assistance systems should use various AI models, each adapted to specific environmental conditions or traffic situations,” Schering emphasised. The data also provided initial indications of potential problem areas in the city of Osnabrück, including locations where cars and bicycles frequently overtake each other or where the prescribed distances are difficult to maintain due to the traffic layout.
Alongside the University of Oldenburg, iotec GmbH, a company based in Osnabrück, and the City of Osnabrück participated in BikeDetect as associated partners. The project was funded by the Federal Ministry of Transport (BMV) as part of its mFUND innovation initiative. This programme was launched in 2016 and supports research and development projects focused on data-driven digital innovations for the mobility of the future. It also supports participants by holding events to promote active networking between stakeholders from politics, business, public administration and research, as well as by providing access to open data.
Prof. Dr. Jorge Marx Gómez, Tel.: 0441/798-4470, E-Mail: jorge.marx.gomez@uol.de
https://uol.de/vlba/
https://bike-detect.vlba.net/
As part of the BikeDetect project, the team carried out field tests using a prototype sensor system.
Copyright: BikeDetect / Johannes Schering
Merkmale dieser Pressemitteilung:
Journalisten, Lehrer/Schüler, Studierende, Wirtschaftsvertreter, Wissenschaftler, jedermann
Elektrotechnik, Informationstechnik, Maschinenbau, Umwelt / Ökologie, Verkehr / Transport
überregional
Forschungsergebnisse, Forschungsprojekte
Englisch
As part of the BikeDetect project, the team carried out field tests using a prototype sensor system.
Copyright: BikeDetect / Johannes Schering
Sie können Suchbegriffe mit und, oder und / oder nicht verknüpfen, z. B. Philo nicht logie.
Verknüpfungen können Sie mit Klammern voneinander trennen, z. B. (Philo nicht logie) oder (Psycho und logie).
Zusammenhängende Worte werden als Wortgruppe gesucht, wenn Sie sie in Anführungsstriche setzen, z. B. „Bundesrepublik Deutschland“.
Die Erweiterte Suche können Sie auch nutzen, ohne Suchbegriffe einzugeben. Sie orientiert sich dann an den Kriterien, die Sie ausgewählt haben (z. B. nach dem Land oder dem Sachgebiet).
Haben Sie in einer Kategorie kein Kriterium ausgewählt, wird die gesamte Kategorie durchsucht (z.B. alle Sachgebiete oder alle Länder).
