Development of Machine-Learning Models for Autonomous Vehicle Decisions on Weaving Sections of Freeway Ramps

Center for Connected and Automated Transportation/Research/Research from the University of Michigan/Development of Machine-Learning Models for Autonomous Vehicle Decisions on Weaving Sections of Freeway Ramps

Development of Machine-Learning Models for Autonomous Vehicle Decisions on Weaving Sections of Freeway Ramps

Complete
Brian Lin Headshot. The link directs to their bio page.
Brian Lin
The University of Michigan Transportation Research Institute Logo. The link directs to the funded research led by this institution.

Principal Investigator(s):

Brian Lin, Assistant Research Scientist – The University of Michigan Transportation Research Institute

Project Abstract:
To date no systems can recommend when and how lane changes should be made in weaving sections with limited length to ensure that traffic stays safely and smoothly separated. This study aims to (1) investigate drivers’ decision and speed control before changing lanes into/out of the weaving section, (2) develop the lane change decision and maneuver algorithms for automated vehicles (AVs), (3) apply the algorithms to AVs, and (4) validate the algorithms on Mcity Test Facility. Two types of model/algorithm will be created to (1) identify the surrounding vehicle characteristics, and (2) classify drivers’ decision to change lanes and model the lane change maneuvers in the weaving section. The validation taking place on Mcity will provide evidence to test and improve the algorithms, as well as a demonstration to showcase how the AV can interact with weaving vehicles.

Institution(s): University of Michigan Transportation Research Institute

Award Year: 2019

Research Thrust(s): Human Factors, Modeling & Implementation

Project Form(s):

Project Information Form
Final Report