Autonomous Driving in the Commercial Vehicle Sector for Trucks

BMWi-Project: Identification of Dynamics- and Safety-Relevant Trailer Conditions for Automated Trucks (IdenT)

Autonomous driving of trucks is one of the current topics in the commercial vehicle industry. Particularly because the logistics industry has long relied on digitized efficiency, this area holds special opportunities. Truck trailers have received little attention in automated driving to date, even though they largely determine driving dynamics and truck reliability. Their relevance will increase as autonomous driving advances and expectations of supply chain efficiency rise.

The research project IdenT (Identification of Dynamics- and Safety-Relevant Trailer States for Automated Trucks) focuses on the trailer and develops solutions in different areas. The goal of the project, in which we as ITWM are involved, is the construction and testing of an IdenT system, which consists of an intelligent trailer sensor network, a cloud-based data platform, and methods for the on- and offline processing of the data.

This involves sensor data fusion via digital twins to determine the state of components relevant to driving dynamics and safety, trailer dynamics, and the environment (e.g., road conditions). We develop methods for information fusion of heterogeneous data and apply machine learning methods as well as classical simulation models. The determined information will be provided to the tractor and will be used to support autonomous driving. The system is to be designed and built in the project and tested in test drives.

BPW: Blick unters Fahrzeug
© BPW Bergische Achsen
Ein von BPW zur Verfügung gestelltes Versuchsfahrzeug ist bereits im Einsatz: Der Auflieger dient dazu, das System im realen Fahrbetrieb zu erproben. Hier der Blick unters Fahrzeugs.

Smart Sensors, Machine Learning and Digital Twins

How can smart trailers contribute to trucks being autonomous and at the same time safe and economical on the road? These are the questions we are addressing together with seven partner companies from research and industry. An important component and central sub-goal is the development of a real-time capable digital online twin. This is integrated into the trailer data network and consists of a reduced, real-time trailer model. The online twin processes the received sensor data directly via proven methods of sensor data fusion and machine learning. In addition, as a further important component of the project, the construction and integration of a physical, non-real-time capable digital offline twin will be realized. This consists of a multi-body simulation model (MBS model) that represents components relevant to dynamics, safety and wear as precisely as possible.

Driving Dynamics and Data Analysis

The first substantive goal is for the trailer to provide information relevant to driving dynamics for the tractor. For example, how heavy it is at any given moment, where the center of gravity of the load is, how the axle loads are distributed. This data is not yet available today – but analyzing it is relevant when it comes to making autonomous driving safer. The data collected is to be used in a targeted manner with the IdenT system to be developed: Sensor data will be analyzed and processed in real-time by a mathematical model of the semitrailer – in other words, a digital online twin. The truck also sends information collected online during the journey via a cloud infrastructure to an offline twin, which uses more detailed vehicle models to calculate component wear, for example, and reports back to the online twin. A test vehicle provided by lead project partner BPW Bergische Achsen is already in use: The semitrailer is being used to test the system in real driving conditions and is supplying the first data.

Testtrailer von BPW: Daten nutzen
© BPW Bergische Achsen
Der Testtrailer von BPW ist mit intelligenter Sensorik ausgestattet: Sie erfasst wertvolle Daten und ermittelt fahrdynamische Zustände während der Fahrt auf einer Recheneinheit im Trailer. Diese Daten werden dann von den Forschenden gezielt abgerufen und genutzt.

Our Main Tasks and Mathematical Expertise at ITWM

We are mainly involved in two fields of work

  • the online detection of road conditions, including the associated selection of critical sections
  • the construction, road excitation, and operation of the offline twin.

A key component in the first area is the procedure of detecting road conditions for particularly damaging sections based on condition variables measured on the trailer. For such sections, we use customized algorithms and stochastic road models to determine so-called roughness indicators. A particular challenge here is the necessary computational efficiency of the approach, which must be implemented online, on the trailer.

We draw on extensive expertise and years of application experience in the fields of systems theory, optimal control, and machine learning, and apply procedures and methods from these areas to efficiently estimate road profiles or road roughness during trailer operation. This roughness information – in the form of scalar index numbers for individual road sections – is then used as a classification for trailer loads and is intended to trigger a more detailed simulation the offline twin for particularly damaging sections.

Offline and Online Twins in Interaction

For the operation of the offline twin, on the one hand sufficiently detailed road excitations are necessary, and on the other hand consistent tire models as well. We contribute our expertise in this area to the creation, adaptation, and optimization of the offline twin and also provide the tire model CDTire developed at ITWM. A main focus of our work continues to be the determination of a suitable road profile for the excitation of the offline twin – here, much higher requirements have to be met with respect to accuracy as it is necessary for the online estimation of road roughness described above.

However, the computation at this point does not have to be done online anymore, but can be done together with the planned cloud operation of the offline twin with a limited time offset, which allows more complex computational methods here. We rely on methods of nonlinear optimal control and system inversion to determine a detailed road profile based on trailer measurements or on variables generated by the online twin and in interaction with the offline twin itself.

Project Partner:

  • BPW Bergische Achsen (consortial)
  • Fraunhofer ITWM
  • Fraunhofer LBF
  • Institute of Mechatronic Systems of the University Hannover
  • OKIT GmbH
  • Industrial Science GmbH
  • ts3 the smart system solution GmbH
  • Viscoda GmbH
Gruppenbild zum Projekt IdenT
© BPW Bergische Achsen
Gruppenbild zum Projekt IdenT auf dem Werksgelände von BPW.

Project Duration:

February 2020 to 2023

The project, which is funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) and has a total volume of 4.7 million euros, is scheduled to run for three years and will help to actively drive forward the transformation of transportation with innovative solutions.