Home
About Us
People
Visitors
Groups
Projects
Publications
Software
Courses
Laboratory
Seminars
Students
Matlab
Alumni
Links
 
[Advanced Control Systems Group] [Autonomous Mobile Robotics Group]





Mobile Robots Localization and Map Building


Odometry calibration

Odometry Calibration


Odometry uses known mobile robot start pose and measured displacement in a particular time step to estimate current mobile robot pose. Itís a simple and computationally inexpensive method for relative localization but prone to systematic and nonsystematic errors. Error growth is quadratic with traversed distance.

 

Influence of systematic errors on the mobile robot trajectory

 

To reduce influence of systematic errors and decrease the pose error growth rate odometry calibration can be done. We are using an optimization-based approach to find optimal values of parameters of a 2- or 3-calibration parameters expanded mobile robot kinematics model. Optimization is done using several exciting trajectories that reveal influence of type A (rotational) and type B (translational) odometry errors. Exciting trajectories are so a straight-line trajectory, open trajectory with a constant curvature and turn in place trajectories. UMB mark trajectory is used for validation. Used optimization criterions are:

 

MIN (Exact_orientation - Estimated_orientation)

 

 
 

 

 

 

 

 

 


 

Experimental setup for automatic odometry calibration

 

For alleviation and acceleration of the odometry calibration process an automatised odometry calibration setup with appropriate application setup has been developed. It consist of a server computer equipped with a camera for mobile robot pose measurement and a client computer placed in the mobile robot which odometric system is being calibrated.

 

 

Graphical interface of the automatic odometry calibration application

 

 

†††

 

 

 

For more details see our paper "Simple off-line Odometry Calibration of Differential Drive Mobile Robots" ( PDF ).

 

Home
About Us
People
Visitors
Groups
Projects
Publications
Software
Courses
Laboratory
Seminars
Students
Matlab
Alumni
Links