Improvement of multisensor fusion in speed limit determination by quantifying navigation reliability

Speed limit determination is a complex task that may be solved by fusing data from GIS (Geographical Information System) and camera sensor. Among the existing data fusion models the Dempster-Shafer Belief Theory is found to be the most appropriate in this application. A confidence measure weights each source output, namely speed limit present on road sign and driving situation. Using the discounting scheme of Dempster-Shafer, we propose a new way of computing the navigation confidence measure by taking into account the reliability of the GIS. Preliminary tests showed that our method achieves promising results and solves conflicts between vision- and navigation-based system.