Enhanced GMapping Algorithm for Static Glass Detection

Detecting glass targets and obstacles has significant practical importance in modern society. However, the state-of-art GMapping algorithms suffer from inaccuracies in identifying and mapping their positions when used to deal with highly transparent and reflective glass obstacles under the conditions with only laser sensors. This finally results in incomplete maps and potential navigation errors in robot SLAM applications. To overcome the limitation, the paper presents an enhanced GMapping algorithm by incorporating a glass detection method to fuse laser and ultrasonic data. By exploiting the disparity in distances measured by these two sensors, the method effectively detects glass and refines the algorithm by substituting erroneous distance data with accurate ones. The indoor experimental results demonstrate the improvement of mapping accuracy and reduction of navigation errors owing to the application of the proposed sensor fusion method.