Lane-free multi-agent reinforcement learning-based control of mixed autonomy traffic

PI: Jia Li (WSU),, ORCID:

Co PIs: none

AMOUNT & MATCH: $40,000 federal from PacTrans; $40,000 federal Match

PERFORMANCE PERIOD: 8/16/2023 – 8/15/2025

STATUS: Active

CATEGORIES: Automated Vehicles, Traffic, Mobility


FINAL PROJECT REPORT: will be available once completed

PROJECT DATA: will be available once completed

DESCRIPTION: The rapid advancement of Automated Vehicle (AV) technology is reshaping the transportation landscape, ushering in new possibilities for the design and operation of future transportation systems. Much of the current literature has a strong focus on the development of AV behaviors within multilane mixed autonomy environments, typically assuming the continued use of fixed lanes, as is common practice today.

In this research, we are exploring a groundbreaking concept: lane-free mixed autonomy traffic control. This innovative approach envisions a traffic system where vehicles are not confined to fixed lanes for their longitudinal movements but can instead utilize the roadway space freely. The underlying idea is that by enabling seamless information exchange between automated vehicles and infrastructure, we can optimize road capacity utilization. While this paradigm is still in its early stages and remains experimental, recent studies have indicated its potential benefits in alleviating congestion and improving traffic efficiency through advanced simulations.

This research aims to contribute to this emerging field by designing a multi-agent reinforcement learning (RL) framework tailored for lane-free AV control within mixed autonomy environments so that transportation system efficiency and fuel consumption are optimized.

Research Project Progress Report #1 10/10/2024
Research Project Progress Report #2 4/10/2025
No Cost Extension Request 6/15/2025
Draft Report 6/15/2025
Final Project Report 7/15/2025