Research

Lead Researcher(s)

Huei Peng
Roger L McCarthy Professor of Mechanical Engineering, Professor of Mechanical Engineering, College of Engineering and Director of Mcity

Project Team

Minghan Zhu

Project Abstract

The concept of roadmanship has been studied and quantified in previous Mcity affiliated projects. This project is the first attempt to focus on near-term ADAS applications, i.e., to study the key challenges and plausibility of using onboard low-cost sensors (monocular cameras). The data observed and learned outcome will then be used to adapt the ADAS behavior to conform to the local/temporal driving situation. The key example we will study is headway policy for L1 ACC/L2 Highway Pilot. It is well known that human drivers adapt to regional driving cultures and temporal events, and their headway policy may vary significantly, e.g., during rush hours.

Project Outcome

The basic research premise of this project is an intelligent ADAS system should demonstrate better roadmanship and adjust its headway policy to better conform to the current situation or regional culture. A key difference of the related computer vision (CV) solution requirement is the CV must detect a pair of other vehicles and determine their relative positions accurately. Most current CV solutions focus on detecting individual road users and assessing the range between those objects and the ego vehicle. A direct perception solution to detect a pair of vehicles will provide more accurate measurements, compared with indirect solutions when the errors can compound and become larger.