Pose Estimation By Multisensor Data Fusion Of Wheel Encoders, Gyroscope, Accelerometer And Electronic Compass

In this paper we describe a method for combining data from multiple onboard sensors to determine a mobile robot pose. An error model for a gyroscope, a dual-axial accelerometer and wheel encoders are derived for estimating the mobile robot's pose. A tri-axial magnetometer measures the magnetic field strength which is used as a criterion for acceptance of electronic compass readings to correct the azimuth of the mobile robot's orientation. The errors in each sensor are estimated mutually rather than independently considering each sensor error model. The final estimated pose is used to navigate, guide and control a mobile robot in an indoor environment.