Jens Ziegler’s research while affiliated with Technische Universität Dortmund and other places

... Das Bild der Chemie mit ihren vielen zeitgleich stattfindenden Reaktionen dient der ersten Gruppe von Systemen als Metapher für parallele Algorithmen und großen dezentralen Datenstrukturen ( [27,114,120]). ...

Reference:

Genetische Programmierung einer algorithmischen Chemie

... Bei einfachen geometrischen Objekten sind die Trägheitsmomente aus Tabellenwerken entnehmbar, bei komplexen Körpern müssen sieüber ein Volumenintegral berechnet werden. Die Angabe der Hülle eines Körpers kann beispielsweiseüber eine VRML 1 -Datei erfolgen[14].Abhängig von der aktuellen Konfiguration werden die Bewegungsgleichungen (Differentialgleichungen) generiert, die anschließend im Zeitbereich gelöst werden. Dies erfolgt durch numerische Integration der Gleichungssysteme. ...

Reference:

Evolution von Laufrobotersteuerungen mit Genetischer Programmierung

... However, this universality often proves to be a double edged sword, as will be described below. Nevertheless ANN have successfully been applied to many metamodeling problems (Xiao et al. 2003, Hillbrand and Karagiannis 2002, Dahm and Ziegler 2002, Panayiotou et al. 2000 ) ranging from economic validation of capital projects (Chaveesuk and Smith 2003), through process optimisation (Chambers and Mount-Campbell 2002 ), to microwave circuit modeling (Devabhaktuni et al. 2001, Zhang et al. 2003). An ANN consists of a number of, usually adaptive, artificial neurons that are interconnected to form a network. ...

Reference:

Grid Enabled Sequential Design and Adaptive Metamodeling

... One intensively researched area is surrogate-assisted design optimization, such as turbine blades [9,23,84,85], airfoils [27,86], forging [87], vehicle crash tests [88], multi-processor systems-on-chip design [89] and injection systems [90]. Other applications include drug design [2], protein design [5], hydroinformatics [91] and evolutionary robotics [92]. We must note that not many substantial successful applications of meta-model based evolutionary optimization have been reported, which, however does not necessarily mean no such work has been done. ...

Reference:

Surrogate-assisted evolutionary computation: Recent advances and future challenges

... In addition, SAEAs have been applied to problems other than CEPs, including interactive optimization problems, dynamic optimization [5], network architecture search [9] and others. SAEAs have also been successfully adopted to solve expensive real-world optimization problems from different domains, including engineering design [2], health industry [10] and interactive design [11,12]. ...

Reference:

Surrogate Models in Evolutionary Single-Objective Optimization: A New Taxonomy and Experimental Study

... Gu and Hu [40] employed a combination of reinforcement learning [10] and evolutionary computation to learn refinements of the initially handcrafted parameters of their fuzzy logic controller architecture. Dahm and Ziegler [26] from the German Team envisioned how evolutionary computation could improve localization. Genetic programming allowed them to increase the robots' walking speed to over 20 cm/s. ...

Reference:

Machine Learning With AIBO Robots in the Four-Legged League of RoboCup

... The fundamental goal of the research program behind the present work is to develop computational systems that are living technology in the sense of Bedau et al. (2010)hardware-software artificial chemistries in digital machinery that will perform useful work for society. To some, the reach from software-based artificial chemistry to real-world utility may seem hopelessly long, and indeed this author is aware of only scattered examples-with Ziegler and Banzhaf (2001) offering one clear example. ...

Reference:

Digital protocells with dynamic size, position, and topology

... Their work was one of the first artificial chemistry models based on constructive processes, and it focused on the production of new entities (construction) over time rather than motion through predetermined state spaces (dynamics). 4,5 The model, dubbed AlChemy (for algorithmic chemistry), was based on the construction and composition of λ expressions-objects whose internal structure determines their relation and interaction with each other. 6 The original analyses of AlChemy became foundational studies in understanding the origins of self-organized complexity. ...

Reference:

Self-organization in computation and chemistry: Return to AlChemy

... This kind of approach was observed to work well for complex environments such as those in robotic soccer where multiple restrictions, not present in robotic navigation [6], can be found. Bearing this in mind, in this paper we propose an approach which combines constrained evolutionary computation methods with the Extended Kalman Filter, and a new search space definition can be effective to determine the position of a robot within the context of a robotic soccer game. ...

Reference:

Evolutionary constrained self-localization for autonomous agents

... Seok et al describe a technique to perform calibration of sensors using Genetic Programming on evolvable hardware [7]. Ziegler and Banzhaf proposed to use evolutionary techniques to develop a sensory nose for a robot [9]. The Mate system proposed by Levis is a tiny virtual machine for sensor networks [2]. ...

Reference:

Genetic Programming in Wireless Sensor Networks