Range sensor data filtering for mobile robot localization

Estimating the configuration is one of the key issues for successful mobile robot navigation. Dead-reckoning is simple and fast but it can lead to inaccurate or unreliable estimates. Range sensors are widely used in known environment for localization, but it may not always work due to sensor range limit. Optical flow sensor such as the optical mouse gives accurate relative movement but it requires strict vertical placement. This paper introduces new way of estimating robot configuration by combining the consecutive range sensor data and optical flow information in known environment. Feature extraction from range sensor information and estimation of the relative movement using optical flow sensors are investigated. The proposed method is implemented and tested with a mobile robot