|
|
 |
| |
|
|
|
Improved Approach to Global Localization based on Odometry Error Modeling for Mobile Robots |
|
|
|
PP: 563-576 |
|
|
|
Author(s) |
|
|
|
Jingdong Yang,
Jinghui Yang,
|
|
|
|
Abstract |
|
|
| It is often the odometry accumulative error without bound after long-range movement that decreases the precision of
global localization for wheeled mobile robots. Therefore, an efficient approach to odometry error modeling is proposed regarding
gentle drive type mobile robots. The approximate functional expressions, between process input of odometry and non-systematic error
as well as systematic error, are derived based on odometry error propagation law. Further, the odometry error model is applied to
the global localization to compensate the accumulative error during long-time navigation. In addition, Because a lot of candidate
poses of robots are generated in the process of monocular visual localization, particle swarm optimization is applied to acquire the
optimal pose for mobile robots during global localization. The experiments denote that in spite of sacrificing a little computation
time, the proposed method decreases odometry accumulative errors, and improves the global localization precision during autonomous
navigation efficiently. |
|
|
|
|
|
|
|
