Main contents

Systems and algorithms

Research in theory and algorithmic development mainly supports the other fields, but certain independent studies have also been done especially concerning multi-modelling techniques and new state concept called "functional state". This is a formalism helping to model complex dynamical processes, like biotechnical ones, with a hybrid two layer model. A recent result which has drawn international interest is joining the Wiener-type non-linear dynamical model with Neural Network approach.

Currently the modelling techniques and algorithms are tested with data obtained from industrial enzyme production and chromatographic separation process used in food industry, Nonlinear Wiener-NN models for dynamic systems and Monitoring and validation of control. The research is part of TEKES national programme on "Adaptive and Intelligent Systems Applications".

In another project belonging to the same programme applicability of soft computing methods, such as fuzzy logic, neural nets and genetic algorithms have been tested to solve complex industrial design and production control problems (a case study with ABB), Sales support system and manufacturing scheduling.

Data fusion by using non-linear filtering techniques have been applied specifically to navigation of mobile robots when data from different sources, like inertia navigation system and GPS are fused to make the result more accurate than either of them can give separately. Navigation in unstructured environments, like a forest area, by using perception data, like lasar (laser radar) data, instead of GPS is another problem area presently under studying. In the near future there will be probably more activity in theoretical navigation research because of an industrial partner of the laboratory has made initiative to establish a special research group in this area.