Dynamic clustering analysis for driving styles identification

For intelligent driving systems, the ability to recognize different styles of surrounding vehicles is crucial in determining safest, yet more efficient decisions especially context mixed environment. Knowing for instance if vehicle adjacent lane aggressive or cautious can greatly assist decision making ego terms whether and when it appropriate make particular manoeuvres (e.g. change). In addition, behave differently under environments, identification highly challenging. To this end, paper we propose a dynamic clustering based profiling approach where clusters vary response changing better capture patterns understand style switch behaviours complicated patterns, position-dependent structure developed driver assigned cluster sequence rather than single cluster. best our knowledge, first research its kind on styles. The usefulness proposed method demonstrated real-world trajectory dataset results show that switches complex be captured. potential applications systems are also discussed.