Evaluating and comparing longitudinal control strategies for autonomous vehicles

Cho, Louis Sungwoo

Permalink

https://hdl.handle.net/2142/129188 Copy

Description

Title

Evaluating and comparing longitudinal control strategies for autonomous vehicles

Author(s)

Cho, Louis Sungwoo

Issue Date

2025-05-06

Director of Research (if dissertation) or Advisor (if thesis)

Talebpour, Alireza

Department of Study

Civil & Environmental Eng

Discipline

Civil Engineering

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Spacing control
• Autonomous vehicles
• Genetic algorithm

Abstract

Maintaining appropriate inter-vehicle distances is crucial in enhancing the safety and efficiency of traffic flow for autonomous vehicles (AV). Various spacing control policies have been introduced in the literature. These policies affect traffic stability and dynamics differently, creating variations in road capacity and driving patterns in mixed traffic. This comparative study analyzes five common policies utilized in the literature: Constant Spacing Policy (CSP), Constant Time Headway (CTH), Traffic Flow Stability (TFS), Constant Safety Factor (CSF), and the Intelligent Driver Model (IDM), and compares their applicability to modeling real-world AV behavior. Accordingly, each model is calibrated using the genetic algorithm for optimizing the parameters of each spacing policy. Three distinct datasets were utilized for calibration: (1) Third Generation Simulation (TGSIM) data from I-294L1 in Chicago that contains SAE Level 1 AVs, (2) TGSIM data from I-90/94 containing SAE Level 2 AVs, and (3) data collected from Level 4 AV operations in Phoenix, AZ. Calibration and simulation results show that CSP and CSF models had the most consistent performance, achieving the lowest RMSE and highest R2 values. The CSP model achieved a good level of performance under low-density highway conditions by maintaining consistent vehicle spacing with minimal perturbations. The CSF policy was determined to be the most optimal policy for high-density traffic, effectively managing safety-critical scenarios involving abrupt acceleration and braking. The CTH model showed reliable performance though it was sensitive to speed fluctuations, resulting in higher errors during dynamic traffic scenarios particularly in stop-and-go traffic. The TFS model consistently had higher errors, reflecting its limited adaptability to congested or complex traffic environments. The IDM model demonstrated strong adaptability and realistic driving behavior across diverse conditions but required precise calibration to achieve optimal performance in high-density and unpredictable scenarios.

Graduation Semester

2025-05

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/129188

Copyright and License Information

Copyright 2025 Louis Cho

Owning Collections