Traffic State-Based Adaptive Cruise Control

An intelligent ACC system that classifies traffic states and dynamically adjusts control strategies to improve traffic flow, reduce stop-and-go waves, and enhance passenger comfort.

---

Demo

Watch Road Test Video

---

Overview

Standard ACC systems react only to the immediate vehicle ahead, amplifying phantom traffic jams and causing string-unstable conditions. Our Traffic State-Based ACC addresses this by:

• Classifying traffic state using onboard sensors to identify free-flow, entering congestion, in-wave, or exiting conditions
• Adapting control strategy dynamically based on classified state
• Improving comfort by mitigating hard-braking events

Traffic State Classification

Mode	State	Description
0	No Wave	Free-flowing traffic or no lead vehicle
1	Into Wave	Vehicles slowing down, entering congestion
2	In Wave	Slow or stopped traffic
3	Out of Wave	Accelerating out of congestion

Control Equation

dv/dt = α(s - (τv + s_min)) + k·Δv

Mode	α	τ	s_min	k
No Wave	0.15	2.0	10	0.424
Into Wave	0.7	2.4	10	0.23
In Wave	0.2	2.5	10	0.35
Out of Wave	1.1	2.4	10	0.24

---

Performance Results

Optimal Penetration Rate: 25-50%

Penetration	String Stability	Safety	Comfort	Recommendation
5 - 25%	Moderate	Excellent	Excellent	Initial Rollout
25 - 50%	Good	Excellent	Good	Target Range
50 - 75%	Excellent	Good	Degrading	Needs Tuning
100%	Catastrophic	Crashes	Poor	Not Recommended

Key Finding: Mixed fleets outperform homogeneous fleets. Human drivers provide crucial damping at high penetration rates.

ROS/Docker test after road test improvements

Synthetic Simlation: String stability at 100% penetration rate under certain conditions

Synthetic Simulation: Stable performance until penetration exceeds 75%

NGSIM Simulation: Time space diagram, 50% penetration

NGSIM Simulation: Time space diagram, 100% penetration

Speed distributions, 100% penetration

---

Requirements

• MATLAB 2025b (or 2025a with .r2025a files)
• Toolboxes: Simulink, MATLAB Report Generator, Control System Toolbox, Simulink Control Design, DSP System Toolbox

---

Testing

86 test cases across two suites with automated PDF report generation.

Single Vehicle Tests (10 cases)

run('test/single_veh_test/run_all_single_test.m')

Multi Vehicle Tests (76 cases)

run('test/multi_veh_test/run_all_multi_test.m')

Large-Scale Simulations

Python-based fleet simulations with NGSIM data. See test/large_scale/README.md.

---

CI/CD Pipeline

Automated testing via GitHub Actions with Grafana monitoring.

View Live Dashboard

See .github/CI_CD_SETUP.md for setup details.

---

Team

Name	Responsibilities
Daechul Jung	Into Wave / Exit Wave Control, Road Test Analysis
Quang Pham	CI/CD Pipeline, In Wave / No Wave Control, Fleet Testing
Kate Sanborn	Classification Subsystem, NGSIM Simulations, ROS Integration

---

References

1. FHWA, "Next Generation Simulation (NGSIM) Program" - Link
2. M. Treiber, "The Intelligent-Driver Model" - Link

---

Vanderbilt University | CS 5892: Projects in Autonomous Vehicles and Traffic | 2025