Demonstration of a robot that autonomouslz discovers the laws of its environment. The robot autonomouslz plans and executes experiments in its environment, and builds a theory of it environment from measured data using machine learning programs. This demo was awarded first prize as the best exhibit in the exhibition FET 2009. Many media world wide reported on this prize, including Financial Times and BBC2 TV which also showed part of the demo.
F.08 Development and manufacture of a prototype
COBISS.SI-ID: 6909268Members of the program team organised the traditional 23rd workshop on Qualitative Reasoning, which in 2009 took place in Ljubljana from 22nd to 24th June.
B.01 Organiser of a scientific meeting
COBISS.SI-ID: 23328551For the D. discoideum research community, we have developed a web-based analytical interface for gene expression and function analysis (www.ailab.si/dictyExpress) [F.06]. It offers novel ways to visually interact with the data and to perform the analysis on selected genes. It is used daily by ~15% of D. discoideum researchers around the world. The interface is included in the reference database of this organism (www.dictyBase.org) [F.16], which receives 60000 hits monthly.
F.06 Development of a new product
COBISS.SI-ID: 7219028The patent is related to the intelligent entry control, based on a flexible architecture enabling integration of an arbitrary number of sensors and an arbitrary number of software methods at an arbitrary number of levels of software systems. Before our research, there was no publication or patent presenting a universal method to combine the sensors and methods in a systematic way. Our approach is based on a device for combining an arbitrary number of sensors in a cascade way, and an algorithm that enables integration of several SW methods based on the sensors.
F.33 Slovenian patent
COBISS.SI-ID: 22402855In electrical discharge machining (EDM), a suitable machining regime has to be set in the gap between the workpiece and the electrode to ensure high material removal rate at low electrode wear. For complex surfaces this involves on-line setting of process parameter values which is difficult to perform on standard machines. Using machine learning, we built a model that accurately determines the machining regime based on the attributes of the voltage and current signals. The approach was verified in laboratory EDM process control, and is suitable for implementation on industrial EDM machines.
F.10 Improvements to an existing technological process or technology
COBISS.SI-ID: 10882587