A new certainty grid based mapping and navigation system for an autonomous mobile robot

A complete Bayesian derivation for a sensor data updating formula for certainty grids is given, and its validity confirmed in a simulation. In the real world this formula alone is less valid because of the specular reflection effect, which seriously detracts from the map quality. A simple and practical approach, which greatly enhances the map quality in spite of some loss of range information, is proposed to compensate for this. A new path planning method based on the enhanced map is implemented on a mobile robot, and every aspect of the new algorithm is investigated through a series of experiments. The results show the validity and usefulness of the proposed approaches, i.e. updating formulae, specular reflection correction and path planning methods.