Assessing the Impact of Fleet Size on Crowdsourced Mapping Using a Dissimilarity Measure
This study addresses the unclear relationship between fleet size and mapping quality in crowdsourced mapping, as well as the high cost and limited scalability of traditional surveying methods. To this end, it proposes a multi-vehicle cooperative perception simulation framework based on the GOSPAM metric. By integrating localization errors with detection performance (false positives and false negatives), and incorporating spatial clustering, semantic filtering, and real-world trajectory simulation, the framework effectively quantifies individual fleet contributions. Experimental evaluations across fleet sizes ranging from 5 to 50 vehicles demonstrate that GOSPAM precisely assesses early vehicle contributions and accuracy gains in large-scale fleets. Consequently, this work provides a scalable, quantitative evaluation approach for maintaining crowdsourced maps.