This Raspberry Pi-based vehicle to features closed-loop PID motor controls, real-time signal processing (FFT), and OpenCV computer vision for navigation, target tracking, and environment-aware cruise control.
- Description: In this objective, I was tasked with implementing closed-loop velocity control on a suspended, no-load vehicle to target a stable 4.0 RPS.
- Signal Validation: In order to collect enough data, I tuned the vehicle to utilize a 200 Hz Analog-to-Digital (AD) sampling rate. I validated my real-time calculations using a Fast Fourier Transform (FFT) analysis, confirming a dominant steady-state frequency spike at 7.94 Hz (3.97 RPS).
-
Tuning Parameters:
$K_p = 0.03$ ,$K_i = 11.0$ ,$K_d = 0.6$ . - My data is logged in
data/car_noload_4rps.txt.
- Description: I tracked the vehicle's velocity under full body load and floor friction. The PiCar required aggressive gains to overcome the rolling resistance and static friction.
-
Tuning Parameters:
$K_p = 11.0$ ,$K_i = 7.0$ ,$K_d = 0.1$ . - I logged my data in
data/manual_car_[speed]rps.txt.
- Description: I programmed the camera to sweep and locate a blue object over 10 feet away. I then aligned the trajectory using a visual geometric center-of-mass (COM) error loop, and I utilized ultrasonic deceleration to park the car safely without collision.
- Vision Pipeline: Images taken on the Pi camera was converted from RGB to HSV color space to build a robust binary pixel mask against ambient shadows.
- My execution logs are in
data/Seeker.txt.
| Original RGB & Center of Mass | HSV Binary Mask |
 |
 |
trashbin.mp4
- Description: I achieved direct line driving by utilizing an MPU-6050 accelerometer gyroscope tracker for real-time heading correction.
- Color Decision Matrix: I processed three concurrent HSV masks (Red, Yellow, Green). For the Red mask, I utilized a bitwise OR operation to join the split bounds (0–10 and 160–180 Hue).
- Behavioral Logic: I designed the system to continue on Green, apply a linear braking deceleration curve on Yellow, and execute a full halt at 170 cm on Red.
- My execution logs are in
data/Traffic_light.txt.
| Original Frame | Red Mask | Yellow Mask | Green Mask |
 |
 |
 |
 |
traffic.mp4
- Description: For this cumulative challenge, I drove the vehicle through a hallway grid involving an uphill ramp climb, a wall-bounded 180° three-point turn, and a downhill ramp descent.
-
Dynamics Management: I raised the parameters (
$K_i = 15.0$ ,$K_d = 0.8$ ) to counteract gravity. I also built a specialized active braking routine that reverses motor polarity if gravity accelerates the vehicle 1.5 RPS beyond my target threshold. - My telemetry data is in
data/Objective_5.txtorplots/Objective_5.png.
For explicit design equations, transfer functions, and full mathematical derivations of my feedforward ratios, please read my complete technical paper: Read my Full Technical Report