10.05.2014 Views

Annual Report 2002 - Örebro universitet

Annual Report 2002 - Örebro universitet

Annual Report 2002 - Örebro universitet

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

60 AASS – Center for Applied Autonomous Sensor Systems<br />

Project: Simultaneous Localization and Map Building<br />

Project leader: Tom Duckett<br />

Project staff: Tom Duckett<br />

Funding: KK and Faculty funding.<br />

Cooperation: Artificial Intelligence Group, Manchester University, UK, with Dr. Stephen<br />

Marsland and Dr. Jonathan Shapiro.<br />

Synopsis:<br />

This project addresses the problem of navigation by mobile robots operating in large, real<br />

world environments, which have not been modified for the purpose of robot navigation. Maps<br />

are essential for mobile robots navigating in complex environments, being needed for selflocalisation,<br />

path planning and interaction with humans. To navigate in unknown<br />

environments, a self-governing robot is faced with a fundamental dilemma: to explore and<br />

learn maps of uncharted territory, the robot needs to know its location, but in order to know<br />

its location, the robot needs a map. Most existing solutions to the problem of simultaneous<br />

localisation and mapping (SLAM) work by decoupling the localisation and mapping<br />

processes, assuming that the data association problem (i.e., landmark identification) is already<br />

solved when measurements are updated into the map. However, this assumption is bound to<br />

fail eventually for complex environments. This project investigates robust SLAM algorithms<br />

which take into account both the measurement uncertainty (e.g., due to sensor noise) and<br />

identification uncertainty (e.g., due to perceptual aliasing, environmental dynamics, sensor<br />

noise, etc.).<br />

Results in <strong>2002</strong>:<br />

- Article on map learning published in Autonomous Robots Journal, based on work carried<br />

out in cooperation with Manchester University.<br />

- Development of a new approach for SLAM based on genetic algorithms, paper submitted<br />

to IEEE ICRA’03.<br />

Future developments:<br />

- Planned research visit by Udo Frese, a Ph.D. student from DLR (German Aerospace<br />

Center), Germany as a Marie Curie fellow in January-March 2003.<br />

- Development of new SLAM algorithms based on techniques for statistical analysis and<br />

global optimization.<br />

An example of a learned map

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!