A Data-driven, Falsification-based Model of Human Driver Behavior

TL;DR

This paper introduces a data-driven framework using parametric signal temporal logic to distinguish human driver trajectories from non-human ones, aiding in behavior analysis and control synthesis.

Contribution

It presents a novel method to construct a boundary for human driver behavior using STL and falsification, improving behavior classification and understanding.

Findings

Successfully differentiates human and non-human trajectories

Reduces uncertainty in driver behavior modeling

Demonstrated on real-world intersection scenario

Abstract

We propose a novel framework to differentiate between vehicle trajectories originating from human and non-human drivers by constructing a data-driven boundary using parametric signal temporal logic (STL). Such construction allows us to evaluate the trajectories, detect rare-events, and reduce the uncertainty of driver behaviors when it assumes the form of a disturbance in control synthesis and evaluation problems. We train a classifier that separates admissible (i.e. human) examples - which arise from real-world demonstrations - and inadmissible (i.e. non-human) examples that are generated by falsifying specifications synthesized from the same real-world driving data. Proceeding in this fashion allows for finding a reasonable boundary of human behaviors exhibited in real-world driving records. The framework is demonstrated using a case study involving a human-driven vehicle approaching a signalized intersection.