Universitšt Bonn: Autonomous Intelligent SystemsInstitute for Computer Science VI: Autonomous Intelligent Systems

Research Topics of Nils Goerke

Nils Goerke's focus of research is to design and investigate a framework that fosters the development of behaviour. Based of several years of experience in teaching and working with neural networks an neural computation approaches including cognitive architecture for robot control his most recent research topics are in neural computation, artificial life systems, nonlinear dynamical systems and cognitive architectures for robot control.

Neural Networks and Neural Computation

Attractive Vector Fields

Clustering with Multi SOMs and Multi Neural Gas Networks

Artificial Life Systems

Evolving Robot Controllers

Experiments on Swarming Behaviour

Cognitive Architectures for Robot Control

Growing up Robots

One of the most challenging features of living artifacts: The ability to grow.

One of the most interesting features of growing artifacts: The special capability to grow up.

Growing up robots means to have a guided ontogenesis of an artifact from an infantile stage to a fully matured entity. Start with an architecture that is pre-structured (genotype) to enable basic abilities right from the beginning. According to a schedule of structural development the modules are trained using diverse paradigms for learning and adaptation. Training the artifact with respect to interaction with the environment is generating individual behaviour (phenotype).
Thus the artifact evolves from an infantile stage to a fully matured entity: The artifact grows up.

The EMOBOT Approach

The EMOBOT concept is an approach to model and generate behaviour for autonomous vehicles (robots) similar to behaviours that we know from living entities. The name EMOBOT consists of the part EMO as a hint to EMOtions that shall/can govern the behaviour of the system, and BOT to indicate an artificial system like a roBOT.
The EMOBOT concept is one approach among others trying to achieve a sophisticated and more flexible control policy for robots. The way the EMOBOT is designed will make the behaviour to become more felxible and more interesting.
The EMOBOT concept is based on the Systemic Architecture approach for designing robot controllers with an Internal Value System (IVS) that serves as basis for controlling the actions. The psychological concepts Drives and Emotions are the role model that govern the EMOBOT behaviour. First implementations of the EMOBOT concept have shown promising results.

Nonlinear Dynamical Systems

Coupled Nonlinear Oscillators