MUREA Self-localization     To MRT scientific achievements     To MRT home page

 

MUREA (MUlti-Resolution Evidence Accumulation) is a mobile robot localization method for known 2D environments. The localization space is divided into subregions (cells) and then it applies an evidence accumulation method, where each perception votes those cells that are compatible with the map of the environment. In order to reduce the complexity of working with a fine grid, we have adopted a multi-resolution scheme. We start applying evidence accumulation on a coarse grid, with few large cells. Then we select and refine (i.e., divide into smaller cells) those cells that have collected highest votes.

The main characteristics of MUREA are: the capability of the system to integrate sensor data coming from different sensors the capability of the system to accept both raw sensor data as well as independently generated localization estimates the capability of the system to be both a global or a local localization system, depending on the availability of a global estimate the capability of the system to give out a (less) accurate estimate whenever asked to do so (e.g. before its regular completion), which could be called any-time localization

We have tested MUREA in the RoboCup domain, with robots equipped with an omnidirectional vision sensor. We made several tests placing the robot in different positions of the field and changing the amount of perceptions acquired. The average localization errors were less than 10 cm for the position and about 1° for the orientation. Recently, we have elaborated a framework to define a set of weights which takes into account the different amount of information and the reliability provided by each class of perceptions. So, a perception contributes to the voting phase according to its reliability and its capability of excluding alternatives. Using these weights we obtain that the cell with the highest vote is also the cells that holds the highest probability of containing the actual robot pose. Principal investigators M. Matteucci, F. Marchese, M. Restelli, D.Sorrenti. Research contributors A. Bonarini, ... Related Papers Restelli, M., Sorrenti D. G. , Marchese F. M. (2002) MUREA: a MUlti-Resolution Evidence Accumulation method for robot localization in known environments, IROS 2002, Lausanne, Switzerland, 415-420. Restelli, M., Sorrenti D. G. , Marchese F. M. (2002) Evidence Accumulation Method for Mobile Robot Localization, Proceedings AI*IA 2002, Siena, Italy Sorrenti D. G. , Restelli, M., Marchese F. M. (2002) A robot localization method based on Evidence Accumulation and Multi-Resolution, RoboCup 2002, Fukuoka, Japan, 2002

To MRT home page

bonarini@elet.polimi.it