Volunteered Geographic Information (VGI) has been recently used to generate low cost, up-to-date, and reliable road maps. Previous methods are periodically applied to update the road maps in a particular time interval and scan the whole study area instead of considering the changed parts of road networks. This paper presents a multi-agent system that dynamically explores the newly collected trajectory data and updates the Google Earth’s road map geometry. In this regard, the developed agents use morphological operation in a raster space, generated from newly added trajectory data to extract road map. Kernel Density Estimation (KDE) method was exploited to transform vector space of trajectory data to raster. Moreover, the road geometry extraction algorithm is applied on the changed parts, reported by newly added trajectory data, instead of the whole study area. In this regard, the work space is divided into regions of equal size. To evaluate the functionality of the proposed algorithm, the result was compared to the Google Earth’s base-map and F-score, Missing and Spurious parameters were 0.57, 0.40 and 0.37, respectively.