|
Home
|
|
Theresa Rienmüller: Self-Localization for soccer robots of the middle size league |
 | Theresa Rienmüller finished her Master's Thesis (download ) Self-localization, that is the provisioning of an accurate estimation of a robot's own position and heading in a known model of the environment, is a major research task in mobile
robotics as it is crucial to provide a robot with autonomous capabilities. |
Many efficient algorithms have been developed in the last years specially focusing on probabilistic methods.
One of the most commonly used techniques is the Monte Carlo Localization that is
based on probabilistic models and state transition probabilities. In this thesis this method
was adapted to the robots of the Mostly Harmless RoboCup Middle Size Team, leading
to an efficient and robust algorithm that is able to determine the absolute position and
orientation of a soccer robot on the field and tracking its movement during a game. The
algorithm is based on two sensors, an omnidirectional camera and an odometry sensor.
The white field markings were used as features for the localization and global localization
was possible at almost every position on the field. Occasionally occuring position losses
are in the majority of cases due to strong changes in the illumination of the field. |
|
|
Newsletter
Mostly Harmless Newsletter:
|
