Clustering of Naturalistic Driving Encounters Using Unsupervised Learning

Deep understanding of driving encounters could help self-driving cars make appropriate decisions when driving in complex settings with surrounding vehicles engaged. This paper develops an unsupervised classifier to group naturalistic driving encounters into several distinguishable clusters by combining an auto-encoder with a k-means clustering (AE-kMC). In order to show the effectiveness of our developed classifier, the data of 10000 naturalistic driving encounters collected by the University of Michigan, Ann Arbor in the past five years were tested using the two proposed methods above. We compare our developed method with the k-means clustering methods. The comparison experiment results demonstrate that our developed AE-kMC outperforms the k-means clustering method.