Study on multi-sensor data fusion for the wheeled mobile robot

Abstract

The usual method for estimating the posture of a wheeled mobile robot is dead-reckoning algorithm. However, it has the problem of gradual error accumulation due to slippage of wheels and measurement noise. To enhance the positioning precision for mobile robots, the information fusion method using Extended Kalman Filter algorithm is investigated in which multi-sensor data are provided by internal sensors such as odometers and external sensor such as laser scanner. Practical path-tracking experiment shows that the estimated posture by this system is precise to be useful.