The chair of Cognitive Systems
We do research in the following fields:
Machine Learning methods
Our original research about artificial neural network simulation, which has begun with the development of the Stuttgart Neural Network Simulator (SNNS) and continued with its successor JavaNNS, is still applied in some industry projects. In recent years, in many of our applications artificial neural networks have been replaced by kernel-based machine learning algorithms, especially in chemoinformatics and bioinformatics. These have been used for kernel based ADME prediction, kernel based virtual screening with molecular flexibility, data mining on chemical compounds, applicability domain estimation and combustion engine state prediction.
In the EvA2 project (evolutionary algorithms) a heuristic optimization system was developed, which contains a large number of variants of genetic Genetic Algorithms (GA), Evolution Strategies (ES), Differential Evolution (DE), Genetic Programming (GP), Model-Assisted ES, Multi-Objective ES, Niching ES, Particle Swarm Optimization (PSO), Ants algorithms, and many more.
EvA2 is used for optimization in dynamic environments, for optimization in geoscience, and in a number of industrial cooperation projects.
Autonomous Mobile Robots
Our robotics research aims at robot vision, sensor fusion, navigation, path planning and control of mobile robots. We operate 3 robotics labs with a total of some 35 robots: in our large robots lab we run 3 MetraLabs SCITOS G5 and 2 RWI B21 robots (Robin and Colin) equipped with cameras, 2D laser scanners, UHF RFID readers and sonar sensors. Swarm behaviors and heterogeneous robot teams are investigated in our small robots lab with 5 omnidirectional service robots (former RoboCup robots) and 12 small c't-Bots. Two types of outdoor robots, one RWI ATRV-Jr and a team of 8 outdoor robot buggies custom-built from 1:8 model monster trucks, are used for outdoor robotics projects. In our flying robots lab we research into aerial robotics with 3 Asctec quadrocopters.
(This still needs to be updated. See the research projects pages)
A series of application areas for bioinformatics is tested here together with partners from chemistry and biochemistry as well as biology. The BMBF research projects automated combinatorial chemistry and search and optimization of lead compounds are counted among them as well as the analysis of planar biological neural networks by means of artificial neural networks. Concerning the project Bioinform@tik teaching material is developed here for the new study course Bioinformatics.