Autonomous Track-Following Car (TI Cup)

This repository contains the firmware and supporting materials for an autonomous track-following car developed for the Texas Instruments (TI) Cup, a competitive embedded systems project hosted by the Rochester Institute of Technology (RIT) Computer Engineering Department.

The system demonstrates real-time embedded control, sensor processing, and closed-loop motor control on a resource-constrained microcontroller platform.

📄 Full Technical Report:
Car_Report.pdf

---

Project Overview

The goal of this project was to design and implement firmware capable of autonomously navigating an unknown track using a 128-pixel line scan camera, a servo steering system, and dual DC motors.

Instead of traditional edge-detection steering, this system uses a Center of Mass (COM)–based steering algorithm with dynamic thresholding to achieve smoother, more stable control at high speeds.

The firmware was written in C and runs on a Texas Instruments MSPM0 LaunchPad.

---

Key Features

• Center of Mass (COM) Steering

  • Computes the weighted average of camera intensity values to locate the center of the track
  • More robust and less oscillatory than edge-based steering

• Dynamic Thresholding

  • Threshold adapts based on camera lighting conditions
  • Prevents noise and background interference from affecting control decisions

• 5-Point Moving Average Filter

  • Smooths raw camera data to reduce noise
  • Optimized for low latency using integer arithmetic and bit shifts

• Variable Speed Control

  • Vehicle speed decreases on sharper turns
  • Speed is proportional to distance from track center

• Differential Motor Steering

  • Inner and outer wheels rotate at different speeds during turns
  • Improves stability and cornering performance

• Carpet Stopper / Off-Track Detection

  • Motors automatically disable when track is lost
  • Prevents runaway behavior during testing

• Optional PID Servo Control

  • Proportional control is always used
  • Integral and Derivative control were implemented and tested (commented in code)

---

Hardware Components

• MCU: Texas Instruments MSPM0 LaunchPad
• Sensor: TSL1401-DB 128-pixel line scan camera
• Motors:
  • 2× DC motors (H-bridge controlled)
  • 1× hobby servo motor (PWM steering)
• Motor Driver: External H-bridge board
• Power: Dual battery pack
• Chassis: Pre-assembled competition kit with custom PCB

---

Software Architecture

Camera Processing Pipeline

1. Acquire raw camera data via ADC
2. Apply 5-point moving average smoothing
3. Compute dynamic threshold
4. Calculate Center of Mass (COM)

Steering Control

• Servo duty cycle range: 5.0% – 10.0%
• Center position: 7.5%
• Duty cycle adjusted proportionally based on COM offset

Speed Control

• Speed decreases as COM deviates from center
• Configurable speed modes via onboard switches

Motor Control

• PWM-based DC motor control using TimerA0
• Independent forward/reverse channels for each motor
• Differential steering applied during turns

---

Repository Structure

├── main.c            --> Primary firmware source
├── Library Files/  --> Camera, UART, ADC, timer, and motor drivers
└── Car_Report.pdf --> Full IEEE technical report

---

Results & Performance

• Successfully completed all pre-race demonstrations
• Achieved class-best unofficial lap time of 26.8 seconds
• Demonstrated stable steering at high speeds
• Identified and corrected a race-day configuration error post-competition

Detailed performance metrics, diagrams, and analysis are available in the full report: 👉 Car_Report.pdf

---

Lessons Learned

• COM-based steering significantly reduces oscillations compared to edge detection
• Minimizing UART output is critical for real-time responsiveness
• Latency-aware algorithm design is essential in embedded control systems
• Differential steering provides substantial performance gains in tight turns

---

Future Improvements

• Adaptive differential steering based on servo angle
• Full PID tuning with dynamic gain adjustment
• Hardware interrupt optimization for lower latency
• Improved fault detection and recovery

---

Authors

• Jean-Charles Mve
  B.S./M.S. Computer Engineering, RIT
  Firmware, algorithms, and control logic

• Joshua Derosier
  B.S. Computer Engineering, RIT
  Hardware integration and mechanical design

---

Acknowledgments

Special thanks to the RIT Computer Engineering faculty and teaching assistants for their guidance throughout the project, including:

• Dr. Stephanie Soldavini
• Gino Porretta
• Christian Ryan
• Maya Kaul
• Oliver Vinneras
• Eliot Nagy

---

License

This project is provided for educational and portfolio purposes.
Reuse is permitted with attribution.