Volume 70, Issue 3

Data fusion is one of the key ingredients of Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS). The article provides an introduction into the field by highlighting different possible data fusion categorizations and architectures alongside pros and cons. Benefits, limitations, and pitfalls of exemplary data fusion systems are presented, and the consequences of specific design choices discussed.

This work provides methodological approaches to solve convex optimization problems arising in multi-agent systems which can be reformulated in terms of a so called N -cluster game. We consider different settings of information available to each agent in the system. First, we present a centralized algorithm, which requires a central coordinator having full access to information about agents’ actions and gradients of their cost functions, to demonstrate how the standard gradient descent method can be applied to achieve an optimal output in N -cluster games. After that we relax the full information setting and assume that only partial information is available to each agent. Focus lies on the following two cases. In the first case, the agents have access to their gradient functions and are allowed to exchange information with their local neighbors over a communication graph that connects the whole system. In the second case, the agents do not know the functional form of their objectives/gradients and can only access the current values of their objective functions at some query point. Moreover, the agents are allowed to communicate only with their local neighbors within the cluster to which they belong. For both settings we present the convergent optimization procedures and analyse their efficiency in simulations.

In this paper, a new approach for the optimal synchronization of homogeneous multi-agent systems is presented. In contrast to existing approaches, we offer an intuitive cost for the transition of each agent to a common synchronization trajectory, which facilitates tuning of the cost weights. The introduced LQ optimal tracking problem is solved by reducing it to a LQ regulator problem. By choosing a linear parameterization of the synchronization trajectory, we enable the distributed implementation of the optimal synchronization controller. We give examples how such a parameterization can be chosen and how it influences the required communication between the agents. Furthermore, we derive an optimal communication topology that minimizes the weighted sum of all agents’ transition costs. Finally, we demonstrate our results in a simulation example.

In previous works we discovered that rule-based systems severely suffer in performance when increasing the number of rules. In order to increase the amount of possible boolean relations while keeping the number of rules fixed, we employ ideas from well known Spatial Transformer Systems and Self-Attention Networks: here, our learned rules are not static but are dynamically adjusted to fit the input data by training a separate rule-prediction system, which is predicting parameter matrices used in Neural Logic Rule Layers. We show, that these networks, termed Adaptive Neural Logic Rule Layers, outperform their static counterpart both in terms of final performance, as well as training stability and excitability during early stages of training.

Zusammenfassung In diesem Beitrag wird ein Design für eine visuelle Passermarke zur Posenschätzung von Kameras vorgestellt, die sogenannte Dynamische Passermarke . Die Passermarke wird auf einem Anzeigegerät dargestellt und passt ihr Aussehen an die räumlichen und zeitlichen Anforderungen der visuellen Wahrnehmungsaufgabe an. Dies ist vor allem für mobile Roboter von Vorteil, die eine Kamera als Sensor verwenden. Es wird eine Regelung entworfen, die das Aussehen der Passermarke in Abhängigkeit der aktuellen Pose der Kamera verändert, um den Erfassungsbereich der Kamera für die Posenschätzung zu vergrößern und eine höhere Genauigkeit der Posenschätzung im Vergleich zu herkömmlichen passiven Passermarken zu erreichen.

In this paper, we discuss the systematic design of gain scheduling controllers for nonlinear systems. We discuss control signal blending induced windup phenomena and employ an extended Model-Recovery Anti-Windup (MRAW) scheme to mitigate them. Combining control signal blending with MRAW allows to build gain scheduling controllers without imposing the common slow variation assumption regarding the scheduling vectors. In addition, the approach offers a higher flexibility in the choice of the sub-controllers compared to classical approaches and convex optimization allows for a very efficient design of the anti-windup networks.

Zusammenfassung Die steigende Komplexität der Randbedingungen für den Betrieb von Windenergieanlagen stellt deren herkömmliche Betriebsstrategie in Frage. Erfolgt die Vergütung über Erlöse an der Strombörse, ist es sinnvoll, beim Betrieb auf die Schwankungen der Preise zu reagieren. Wie sollten derartige Reaktionen gestaltet werden? Welche Auswirkungen hat dies auf die Rentabilität der Windparkprojekte aus Betreibersicht? Mit Hilfe einer Fallstudie bewertet der vorliegende Beitrag anhand ökonomischer Kriterien die gezielte Zu- und Abschaltung der Anlagen in Abhängigkeit von Windverhältnissen und Strompreis. Die Ergebnisse der Fallstudie zeigen, dass es sinnvoll sein kann, die Anlagen in wirtschaftlich ungünstigen Situation abzuschalten. Entgangene niedrige Erlöse bei schadensintensiven Windverhältnissen werden durch die Erlöse über eine deutlich verlängerte Lebensdauer der Anlagen wiedergewonnen und die Retabilität des Projektes steigt erheblich.

Input shaping is a well-established approach for suppressing oscillations in systems with flexible modes. When applying input shaping, it is common to assume the system to be at rest initially, i. e., the initial conditions of the oscillatory part of the system have to be at zero. In this paper, we propose a method that allows relaxing the aforementioned assumption for a large class of input signals. The approach relies on the standard input shaper structure but re-parameterizes the shaper such that non-zero initial conditions are cancelled out by the inputs, resulting in zero residual oscillation. Natural physical limitations of the concept are discussed and the application to overhead cranes is presented. The method is validated both in simulation as well as in experiments using a small scale crane. In addition, a simplified trajectory generator is presented that tailors a standard seven-segment jerk-limited motion profile such that zero residual vibrations are achieved without any dedicated input shaping.