Probabilistic driver modeling to characterize human behavior for semiautonomous framework

In this paper, we explore a probabilistic driver model that predicts human behavior based on the external environment, the state of the driver, and previously observed trajectories for an individual driver. The overall goal of this driver model is to create an individualized prediction scheme over a long time horizon in various situations. The novel feature of this particular model is that it uses empirical driver data to create a set of potential future trajectories. A clustering algorithm was developed to identify scenarios and collect the associated driver behavior in a controllable form. This model can then be used to determine when a safe controller should intervene to keep the vehicle in a safe region of the current environment. Experimental results show that this method is has an accuracy of up to 90%, showing that human drivers tend to drive in a reproducible manner. Here, this model is extended to a probabilistic method to examine the reliability of the model, characterize human behavior, and quantify the dissimilarities of driver behavior in different scenarios. In addition, a semiautonomous framework is proposed that encapsulates the driver behavior and vehicle dynamics to examine threat in a given situation.