Using Graph-Theoretic Machine Learning to Predict Human Driver Behavior

Studies have shown that autonomous vehicles (AVs) behave conservatively in a traffic environment composed of human drivers and do not adapt to local conditions socio-cultural norms. It is known socially aware AVs can be designed if there exists mechanism understand the behaviors drivers. We present an approach leverages machine learning predict, This similar how humans implicitly interpret on road, by only observing trajectories their vehicles. use graph-theoretic tools extract driver behavior features from obtain computational mapping between extracted trajectory vehicle behaviors. Compared prior approaches this domain, we prove our method robust, general, extendable broad-ranging applications such as navigation. evaluate real-world datasets captured U.S., India, China, Singapore, well simulation.