PID Controls

The goal of this project is to implement a PID controller that determines the steering angle in order to keep a car in the center of the lane track during driving.

Demostration

The full demostration is at https://youtu.be/jeNe3Oa1ZVo

Terminologies

Cross Track Error (CTE): The distance from the car to the trajectory
PID: Proportional-Integral-Derivative
P component: The steering angle is set to proportion of CTE with a proportional factor tau_p.
```
steer_angle = - tau_p * cte
```
D component: The differential component of the controller which helps to take temporal derivative of error. When the car turned enough to reduce the error, it will help not to overshoot the road.
```
diff_cte = cte - prev_cte
prev_cte = cte
steer_angle += - tau_d * diff_cte
```
I component: The sum of cte overtime to minimize the average CTE.
```
sum_cte += cte
steer_angle += tau_i * sum_cte
```

The PID controller controls the steering value.

cte = robot.y
diff_cte = cte - prev_cte
prev_cte = cte
sum_cte += cte
steer = -tau_p * cte - tau_d * diff_cte - tau_i * sum_cte

PID Fine-Tune

1. Manually fine-tune parameters for Kp, Ki, Kd

Step 1: I set parameters to zeros. Obviously, the car drives straight.
Step 2: I increased the Kp of the P component with an increment of 0.01 until the car start going on following the road and drives with constant oscillations.
Step 3: I increased the Kd of the D component with an increment of 0.1 to try to reduce oscillations.
Loop the 2nd step and the 3rd step.
Step 4: When the car drives the track with really small oscillations and without going out of the road, I increased the Ki of the I component to minimize the average CTE.

Finally, I ended up with the parameters: Kp = 0.15, Ki = 0.0001, Kd = 1.0

2. Applying Twiddle algorithm to optimize the parameters

I also implemented the Twiddle algorithm to optimize the parameters. I'll update the results of the algorithm later.

Code Style

Google's C++ style guide.

indent using spaces
set tab width to 2 spaces (keeps the matrices in source code aligned)

Source code structure

The directory structure of this repository is as follows:

${ROOT}
├──build.sh
├──clean.sh
├──CMakeLists.txt
├──README.md
├──run.sh
├──src/
    ├──json.hpp
    ├──main.cpp
    ├──PID.cpp
    ├──PID.h

Dependencies

cmake >= 3.5
make >= 4.1
gcc/g++ >= 5.4
uWebSockets

How to compile and run

Download the Term 2 Simulator here.
Install uWebSocketIO:
This repository includes two files that can be used to set up and install uWebSocketIO for either Linux or Mac systems. For windows you can use either Docker, VMware, or even Windows 10 Bash on Ubuntu
You can execute the install-ubuntu.sh to install uWebSocketIO.
Once the install for uWebSocketIO is complete, the main program can be built and ran by doing the following from the project top directory.

mkdir build
cd build
cmake ..
make
./pid

References

PID controller

maudzung/Self-Driving-Car-08-PID-Control-CPP