Abstract
Odometry using incremental wheel encoder sensors provides the relative position of mobile robots. This relative position is fundamental information for pose estimation by various sensors for EKF Localization, Monte Carlo Localization etc. Odometry is also used as unique information for localization of environmental conditions when absolute measurement systems are not available. However, odometry suffers from the accumulation of kinematic modeling errors of the wheel as the robot's travel distance increases. Therefore, systematic odometry errors need to be calibrated. Principal systematic error sources are unequal wheel diameters and uncertainty of the effective wheelbase. The UMBmark method is a practical and useful calibration scheme for systematic odometry errors of two-wheel differential mobile robots. However, the approximation errors of the calibration equations and the coupled effect between the two systematic error sources affect the performance of the kinematic parameter estimation. In this paper, we proposed a new calibration scheme whose calibration equations have less approximation errors. This new scheme uses the orientation errors of the robot's final pose in the test track. This scheme also considers the coupled effect between wheel diameter error and wheelbase error. Numerical simulations and experimental results verified that the proposed scheme accurately estimated the kinematic error parameters and improved the accuracy of odometry calibration significantly.
Original language | English |
---|---|
Pages (from-to) | 134-142 |
Number of pages | 9 |
Journal | International Journal of Advanced Robotic Systems |
Volume | 8 |
Issue number | 6 |
DOIs | |
Publication status | Published - 2011 Dec |
Keywords
- Calibration
- Mobile robots
- Odometry
- Pose estimation
- Systematic errors
ASJC Scopus subject areas
- Software
- Computer Science Applications
- Artificial Intelligence