In this contribution, we present and discuss a method for solving inverse tracking problems in vehicle engineering. The goal is to compute input signals for dynamic systems in order to reproduce measured output reference data. Especially, in the black-box situation, where only input-output behaviour of the considered system can be observed, an iterative approach is applied to solve the inverse problem: the iterative learning control (ILC) method. Hereby, first a linear surrogate model in the frequency domain is estimated using spectral methods. Then, a Newton-like iteration procedure is used to compute input updates via the inverse of the surrogate model in each iteration step. The method is applied and demonstrated in a framework for identifying road profiles based on measurement data of a three axle trailer. The identified road profile can be used as excitation and input quantity for a detailed multi-body system simulation.

Traffic jams are a regular annoyance on today’s roads, resulting in longer travel times, higher air pollution and more stressed traffic participants. It has previously been shown that traffic jams do not only occur following a dedicated event, such as road works or accidents, but may also form randomly if traffic density and average velocity rise over a critical threshold.A typical proving ground scenario to study human and automated vehicle behavior in these spontaneous traffic jams is a single-lane, closed ring road, fully populated by vehicles. The simplicity of this scenario allows for the detailed study of traffic jam dissipation techniques and related driving behavior, and for the comparison of human drivers with automated vehicles. In an interactive driving simulator study, thirty human drivers and several automated intelligent controller variants were tasked with dissipating an occurring traffic jam. It demonstrated markedly different strategies employed by human drivers and identified successful and maladaptive approaches. By comparison, automated controllers were more consistently able to resolve traffic jams, allowing for relatively high average velocities during the obstruction. However, the total time the intelligent controllers needed to stabilize the traffic situation was only average compared to the human drivers.In a future faced with mixed traffic involving both automated and human controlled vehicles in complex traffic scenarios, understanding the diverse approaches to the dissipation of traffic jams is crucial for evolving traffic management systems.

Simulation and simulation-assisted methods have a large impact for product development in many industrial fields. Multibody dynamics (MBD) simulation, for instance, is an established method used for durability analysis and energy efficiency calculations in vehicle engineering, based on its ability to provide accurate prediction of interaction forces. In the domain of off-road vehicles and heavy machinery, such considerations need to be expanded by a method to model the soil and its interaction with the vehicle, i.e., the soil–tool interaction, as the resulting forces may drastically influence the durability and energy efficiency of the machine. Therefore, the model must be chosen carefully to maintain the high accuracy in the force prediction for the soil–tool interaction, as required for a reliable and robust product development. In this contribution, we present a workflow to tackle this topic, based on a cosimulation scheme between an MBD-based vehicle model and a particle simulation realized in the Discrete Element Method (DEM). The simulated soil is parametrized and validated by matching simulation results from a virtual experiment with measurement data from real-world soil laboratory experiments as the triaxial compression test. Using this process, the applicability and performance of the numerical methods can be determined.

A current trend in the commercial vehicle industry are autonomous trucks and tractor-semitrailers which will lead to an increasing automation of functions and autonomous transport processes in the future. In this contribution, we present the IdenT system concept, which has been developed for tractor-semitrailers within the research project of the same name, consisting of an intelligent trailer sensor network, a cloud-based data platform and methods for on- and offline data processing. In this work, we focus on the offline process, by presenting its architecture and main functions. As fundamental elements of the offline process, the digital twin of the trailer, i.e., a detailed multi-body model with current inputs and parameters acquired via the system’s online setup, and the methodology to identify the road profile are presented in detail. We show and discuss results of the offline process for some demonstration cases that illustrate how the simulation of the trailer digital twin with identified road profiles can provide a valuable identification of the trailer’s real system dynamics. Moreover, some expected limits of the approach and results of some representative signals are discussed for the demonstration cases.

This paper introduces and analyses a continuous optimization approach to solve optimal control problems involving ordinary differential equations (ODEs) and tracking type objectives. Our aim is to determine control or input functions, and potentially uncertain model parameters, for a dynamical system described by an ODE. We establish the mathematical framework and define the optimal control problem with a tracking functional, incorporating regularization terms and box-constraints for model parameters and input functions. Treating the problem as an infinite-dimensional optimization problem, we employ a Gauss-Newton method within a suitable function space framework. This leads to an iterative process where, at each step, we solve a linearization of the problem by considering a linear surrogate model around the current solution estimate. The resulting linear auxiliary problem resembles a linear-quadratic ODE optimal tracking control problem, which we tackle using either a gradient descent method in function spaces or a Riccati-based approach. Finally, we present and analyze the efficacy of our method through numerical experiments.

The Discrete Element Method (DEM) is broadly used for soil modeling, especially if a realistic prediction of interaction forces with solid materials, i.e. tools is required. While recent enhancements of computing power allow for faster computing times in many fields, DEM-based calculations are still far from real-time, typically by a factor of 100 or more. This is a bottle neck within the design and development processes of agricultural and construction machinery, which are relying on the interaction forces with soils. Finding a decent surrogate model, combining higher computing speeds without loosing accuracy in the prediction of soil-tool interaction forces, would be highly beneficial. Here, we discuss an approach based on recurrent neural networks with the potential of combining real-time capability with accurate soil-tool interaction force prediction.

Automated vehicles have been proposed as a way to influence traffic flow to avoid congestion and maintain a smooth traffic flow. Experiments have shown that congestion formation can be reproduced in an artificial ring road scenario. We design a model predictive controller for the ring road system assuming heterogeneous drivers and an automated vehicle for which congestion resolution and convergence to a reference speed can be shown. The driver's model captures human driving responses. A stabilizing predictive controller framework is employed under the use of a local controllability assumption connected to a local linear quadratic regulator. A case study shows the efficacy of the proposed controller and provides numerical values for the required prediction horizon, highlighting congestion resolution as well as theoretical conservativeness.

Soil models coupled with multibody systems are wellestablished in the development process of construction machinery to predict reaction forces in relevant application maneuvers. In order to accommodate the large variety of soils, it is crucial to choose a suitable model complexity and corresponding identifiable parameters to describe the respective soil characteristic. Within this contribution, we present the parametrization of a soil model based on the Discrete Element Method and discuss efficient and robust methods with the help of direct optimization approaches.

We study different approaches to use real-time communication between vehicles, in order to improve and to optimize traffic flow in the future. A leading example in this contribution is a virtual version of the prominent ring road experiment in which realistic, human-like driving generates stop-and-go waves. To simulate human driving behavior, we consider microscopic traffic models in which single cars and their longitudinal dynamics are modeled via coupled systems of ordinary differential equations. Whereas most cars are set up to behave like human drivers, we assume that one car has an additional intelligent controller that obtains real-time information from other vehicles. Based on this example, we analyze different control methods including a nonlinear model predictive control (MPC) approach with the overall goal to improve traffic flow for all vehicles in the considered system. We show that this nonlinear controller may outperform other control approaches for the ring road scenario but intensive computational effort may prevent it from being real-time capable. We therefore propose an imitation learning approach to substitute the MPC controller. Numerical results show that, with this approach, we maintain the high performance of the nonlinear MPC controller, even in set-ups that differ from the original training scenarios, and also drastically reduce the computing time for online application.

In diesem Beitrag präsentieren wir das sog. IdenT-System, welches im Rahmen des gleichnamigen Forschungsprojekts für LKW-Trailer entwickelt wird. Wir betrachten eine spezifische Teilkomponente, nämlich die Online-Identifikation von Straßenprofilen und -rauigkeiten basierend auf Trailer-Messungen, einfachen Modellen und maßgeschneiderten mathematischen Verfahren. Die Rauigkeit des befahrenen Straßensegments wird mithilfe des sog. IRIs (International Roughness Index) aus dem Profil abgeleitet und dient u.a. der Detektion besonders beanspruchender und kritischer Straßensegmente. Für solche wird im IdenT-System u.a. ein Signal ausgelöst, das den auf der Cloud-Plattform befindlichen Offline-Zwilling startet, der für solche Straßenabschnitte höher aufgelöste Informationen generiert. Die basierend auf den identifizierten Straßeneigenschaften gewonnenen Informationen dienen der Überwachung des Trailerzustands und können zukünftig einen Beitrag dazu leisten, Sattelzüge sicherer zu betreiben.

We study approaches to use (real-time) data, communicated between cars and infrastructure, to improve and to optimize traffic flow in the future and, thereby, to support holistic, efficient and sustainable mobility solutions. To set up virtual traffic environments ranging from artificial scenarios up to complex real world road networks, we use microscopic traffic models and traffic simulation software SUMO. In particular, we apply a reinforcement learning approach, in order to teach controllers (agents) to guide certain vehicles or to control infrastructural guidance systems, such as traffic lights. With real-time information obtained from other vehicles, the agent iteratively learns to improve the traffic flow by repetitive observation and algorithmic optimization. For the RL approach, we consider different control policies including widely used neural nets but also Linear Models and Radial Basis Function Networks. Finally, we compare our RL controller with other control approaches and analyse the robustness of the RL traffic light controller, especially under extreme scenarios.

In modern cars, a huge number of different cables can be found, they are typically combined in hoses and bundles in various different ways. For virtual product development and simulation-based design, it is necessary to know the characteristic physical parameters, like the effective bending or torsion stiffness, of these cable systems. In early stages of the development process as well as for highly customized individual cable configurations, measuring effective stiffness properties is, however, often very challenging. In this contribution, we show results from our current research activities aiming at data-based modeling and estimating effective stiffness parameters for cable bundles. On the basis of an available data set consisting of measured stiffness values for varying cable bundles, the overall goal is to identify a model out of this data, that predicts bundle stiffness values with bundle characteristics as inputs that can be specified without complex measurement efforts. We outline our approach to solve this nonlinear identification task with Gaussian Process (GP) regression. Besides a short introduction to the industrial application area, we demonstrate and illustrate the applicability and prediction quality of Gaussian process regression for this task.

Durability engineering for vehicles is about relating the real operational loading and fatigue environment to the actual capability and strength capacity of the product and its parts. The statistical modelling of the usage loading distribution has always been a big challenge; the latter is necessary to derive statistically validated design targets. In the past, dedicated durability measurement campaigns were the only available data source to obtain durability loads. Nowadays, this has changed due to the increasing availability of new complementary data sources. We show how such data like field monitoring data and geographic data, along with the usage of new data analytics and simulation methods can be used to derive customer and usage related durability loads and targets.

Intelligent traffic control is a key tool to achieve and to realize resource-efficient and sustainable mobility solutions. In this contribution, we study a promising data-based control approach, reinforcement learning (RL), and its applicability to traffic flow problems in a virtual environment. We model different traffic networks using the microscopic traffic simulation software SUMO. RL-methods are used to teach controllers, so called RL agents, to guide certain vehicles or to control a traffic light system. The agents obtain real-time information from other vehicles and learn to improve the traffic flow by repetitive observation and algorithmic optimization. As controller models, we consider both simple linear models and non-linear radial basis function networks. The latter allow to include prior knowledge from the training data and a two-step training procedure leading to an efficient controller training.

In this contribution, we start with a policy-based Reinforcement Learning ansatz using neural networks. The underlying Markov Decision Process consists of a transition probability representing the dynamical system and a policy realized by a neural network mapping the current state to parameters of a distribution. Therefrom, the next control can be sampled. In this setting, the neural network is replaced by an ODE, which is based on a recently discussed interpretation of neural networks. The resulting infinite optimization problem is transformed into an optimization problem similar to the well-known optimal control problems. Afterwards, the necessary optimality conditions are established and from this a new numerical algorithm is derived. The operating principle is shown with two examples. It is applied to a simple example, where a moving point is steered through an obstacle course to a desired end position in a 2D plane. The second example shows the applicability to more complex problems. There, the aim is to control the finger tip of a human arm model with five degrees of freedom and 29 Hill's muscle models to a desired end position.