title: Robot learning in dynamic environments
reg no: ETF5613
project type: Estonian Science Foundation research grant
subject: 1.1-1.5. Exact Sciences
1.10-1.16. Bio-Geo Sciences
2.9. System Engineering and Computer Technology
status: completed
institution: TTU Faculty of Mechanical Engineering
head of project: Maarja Kruusmaa
duration: 01.01.2003 - 31.12.2005
description: The aim of the project is to develop learning and classification methods for dynamic environemnts. The equipment using this methods (autonomous and semi-autonomous robots) should be able to process information and make decisions in cases when:
- the environemnt is difficult to be modelled
- the environemnt is dynamic and changes in the environemnt are not predictable
- information is noisy, incomplete and incorrect
The methods for learning and classification will be developed for two applications:
1. for mobile robot path planning in dynamic environemnts.
the outcome of this implementation is
- a method for mobile robot path planning in partially unknown dynamic environemnts that minimize the total travel time and energy consumption as well as reduce risk.
- an analysis of the method that proves mathematcally its efficiency, scalability and generalizability.
2. image recognition for classification of benthic vegetation and benthic morphology in Baltic Sea.
the outcome of this implementation is:
-image processing algorithm for classification of benthic vegetation and benthic morphology from underwater video images.
- a prototype of a semi-autonomous robot for seabed monitoring in Baltic Sea

project group
no name institution position  
1.Kristo HeeroDept. of Mathematics and Informatics, Tartu UniverPh.D. student 
2.Maarja KruusmaaTTU Faculty of Mechanical Engineeringseniour researcher 
3.Madis ListakInstitute of Mechatronics, Tallinn Technical UnivePh.D. student 
4.Georg MartinEstonian Marine Instituteseniour researcher 
5.Jan WillemsonDept. of Mathematics and Informatics, Tartu UniverLecturer/seniour reseracher