Lifting Judgment with a Boosting Algorithm for Robust Dead Reckoning When Carrying a Robot

Abstract

A lifting judgment for a robust dead reckoning system is proposed for a wheeled mobile robot when the robot is carried by the person. In this system the position of the robot on the ground is estimated by an odometry, whereas that in the air is estimated by an inertial sensor. The estimation mode is switched by the lifting judgment. Neither enough accuracy nor reliability is actually obtained though it is thought that the lifting judgment can be decided by only height information. This is attributed to the fact that there are cumulative errors due to the bias drift in the height calculation from the integration of the value of the acceleration sensor. Therefore, a PSD distance sensor was installed for only the lifting judgment in the previous implementation. This paper proposes a lifting judgment method with a boosting algorithm by using the information on the internal sensors such as accelerometer, gyro sensor, and wheel encoder without the PSD distance sensor. Furthermore, experimental results using an actual mobile robot are given to demonstrate the effectiveness of the present method.