Mobile robot ego-motion estimation by proprioceptive sensor fusion

For any mobile robot it is a major issue that of estimating its position into the working environment. Although this task is partly carried out through external sensors, incrementally computing the ego-motion of the robot using proprioceptive sensors still is a fundamental step to obtain an estimation of the robot displacement. In this work we deal with the sensor fusion problem for the case of a mobile robot equipped with an odometer and an inertial sensor (a gyroscope). We address this problem rigorously through its formulation as a probabilistic estimation problem, developing an efficient solution in the form of an Extended Kalman Filter (EKF), which can be easily implemented in the low-level firmware of a real mobile robot. Experimental results reveal a qualitative improvement in the robot pose estimation for our sensor fusion system when compared with odometry only, which is the most wide spread technique in commercial robots.