Physics-informed Machine Learning for Autonomous Vehicle Control

Details

• Creators:
  Li, Xianning ; Ozbay, Kaan ; Jiang, Zhong-Ping
• Corporate Creators:
  New York University. Tandon School of Engineering
• Corporate Contributors:
  Connected Communities for Smart Mobility Toward Accessible and Resilient Transportation for Equitably Reducing Congestion (C2SMARTER) Tier-1 University Transportation Center (UTC) ; United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology
• Subject/TRT Terms:
  Autonomous Vehicles Congestion Management Systems Connected Vehicles Machine Learning Trajectory Control Vehicle Mix
• Resource Type:
  Manuscript
• Right Statement:
  Public Domain
• Geographical Coverage:
  United States
• Corporate Publisher:
  Connected Communities for Smart Mobility Toward Accessible and Resilient Transportation for Equitably Reducing Congestion (C2SMARTER) Tier-1 University Transportation Center (UTC)
• Abstract:
  In this paper, a physics-informed machine learning approach is proposed for the lane changing of autonomous vehicles. Combining vehicle longitudinal and lateral control, the lane changing problem is formulated as a nonlinear motion planning and control optimization problem based on model predictive control (MPC). To address the computational challenge in solving the nonlinear MPC problem, a neural network controller is trained based on differential predictive control self-supervised learning framework. Through numerical simulation, the learned controller is compared with a conventional MPC controller. It improves the computational efficiency by 97.36% and provides reduced overshoot of the control inputs. Index Terms— Lane Changing, Physics-informed machine learning, Differentiable predictive control (DPC).
• Content Notes:
  Public Domain, except where noted in the document. Li, Xianning, Kaan Ozbay, and Zhong-Ping Jiang. Physics-informed Machine Learning for Autonomous Vehicle Control.
• Format:
  PDF
• Funding:
  69A3552348326
• Collection(s):
  Autonomous Vehicles University Transportation Centers
• Main Document Checksum:
  urn:sha-512:6867e93622975da306bccfe048d5c5d9f6b83a6f32ef7e290aea25fa4c9785f06efb4bc3baed33ac7e5de958bd2e17986b672925d2ae858906c1f325aebcb370
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/86347/dot_86347_DS1.pdf
• File Type:
  [PDF - 1.46 MB ]