Safe Human-Inspired Mesoscopic Hybrid Automaton for Autonomous Vehicles

TL;DR

This paper introduces a mesoscopic hybrid automaton model for autonomous vehicles that mimics human driving behavior in car-following and lane-changing scenarios, integrating microscopic and macroscopic data for improved control.

Contribution

It presents a novel human-inspired hybrid automaton model that incorporates macroscopic parameters into microscopic vehicle control for adaptive cruise control.

Findings

Simulation demonstrates advantages of the mesoscopic model.

The model effectively imitates human driver behavior.

Feasibility of communication network is analyzed.

Abstract

In this paper a mesoscopic hybrid model, i.e. a microscopic hybrid model that takes into account macroscopic parameters, is introduced for designing a human-inspired Adaptive Cruise Control. A control law is proposed with the design goal of replacing and imitating the behaviour of a human driver in a car-following situation where lane changes are possible. First, a microscopic hybrid automaton model is presented, based on human psycho-physical behavior, for both longitudinal and lateral vehicle control. Then a rule for changing time headway on the basis of macroscopic quantities is used to describe the interaction among next vehicles and their impact on driver performance. Simulation results show the advantages of the mesoscopic model. A feasibility analysis of the needed communication network is also presented.