Peter Pd Dr.-Ing. habil. Otto

Neuro-Fuzzy kolmogorov's network for time series prediction and pattern classification

In the paper, a novel Neuro-Fuzzy Kolmogorovs Network (NFKN) is considered. The NFKN is based on and is the development of the previously proposed neural and fuzzy systems using the famous Kolmogorovs superposition theorem (KST). The network consists of two layers of neo-fuzzy neurons (NFNs) and is linear in both the hidden and output layer parameters, so it can be trained with very fast and simple procedures: the gradient-descent based learning rule for the hidden layer, and the recursive least squares algorithm for the output layer. The validity of theoretical results and the advantages of the NFKN are confirmed by experiments.

Universal Approximator Employing Neo-Fuzzy Neurons

A novel fuzzy neural network, called Fuzzy Kolmogorov’s Network (FKN), is considered. The network consists of two layers of neo-fuzzy neurons (NFNs) and is linear in both the hidden and output layer parameters, so it can be trained with very fast and computationally efficient procedures. Two-level structure of the rule base helps the FKN avoid the combinatorial explosion in the number of rules, while the antecedent fuzzy sets completely cover the input hyperbox. The number of rules in the FKN depends linearly on the dimensionality of input space. The validity of theoretical results and the advantages of the FKN are confirmed by a comparison with other techniques in benchmark problems and a real-world problem of electrical load forecasting.

Cooperative line of sight target tracking for heterogeneous unmanned marine vehicle teams

In this paper we present the principle of Cooperative Line Of Sight Target Tracking (CLOSTT) for Heterogeneous Unmanned Marine Vehicle Teams. Thereby CLOSTT is part of a control architecture developed to coordinate existing single heterogeneous autonomous marine vehicles as a team. Within this control architecture CLOSTT separately offers a solution to the task of following a moving underwater target with a team of unmanned marine vehicles. We describe an algorithm for target tracking by a team of autonomous marine vehicles.A general control architecture for such a team is described.Focus is put on tracking an underwater target by a team of autonomous marine vehicles.The tracking algorithm was validated in real sea trials.In the trials, an acoustic communication link was in the loop.

Kalman filter based team navigation for Multiple Unmanned Marine Vehicles

In applications employing multiple unmanned marine vehicles (MUMVs), the navigation has a very great importance to guarantee formation preservation and collision avoidance. While single vehicles usually base their navigation on absolute measurements (GPS, inertial navigation) to determine their position relative to the world, it may be reasonable to perform a relative navigation within vehicle teams. In this paper, we propose relative team navigation based on Kalman Filters to enable a velocity controller to establish a close formation under the typical marine constraints (narrow band width communication with low reliability). We will simulate a team of three marine vehicles and compare the results of different strategies for team navigation.

Machine learning of expert decision or system behaviour

Abstract A fuzzy-modeling method for the emulation of expert decision behavior or for static as well as dynamic systems is presented. The input – output dataset of the system – or expert behavior is changed using fuzzy-sets into examples in linguistic form. These resulting examples build the fundament of the machine learning process for rule production (ID3). The fuzzy sets are optimized in order to minimize the mean square error between the model and the system output.

Path Planning for Cooperative Line Of Sight Target Tracking of Heterogeneous Unmanned Marine Vehicle Teams

Abstract In this paper we suggest a procedure to perform path planning for cooperative target pursuit, performed by a heterogeneous team of unmanned marine vehicles. This work was performed in the framework of the research project GREX (IST-Project-No. 035223) which aims for the realization of several mission scenarios including cooperative marine vehicles. We will describe the algorithms for the team leader to calculate new paths for all team members online and demonstrate the results in a computer simulation.

A FUZZY-BASED MANOEUVRE MANAGEMENT SYSTEM FOR AN AUTONOMOUS UNDERWATER VEHICLE

Abstract The problem domain in this work is a three-dimensional simulation of an underwater vehicle (AUV) that must navigate through obstacles towards a stationary goal point. The AUV has a limited set of sensors, including sonar, and can set its speed and direction each decision cycle. We wish to learn a strategy that is expressed as a set of reactive rules, (i.e. stimulus-response rules) that map sensor readings to actions to be performed at each decision time step. Note that the system does not learn a specific path, but a set of rules that reactively decide a move at each time step allowing the vehicle to reach its goal and avoid the obstacles.