CarND-Controls-PID

Self-Driving Car Engineer Nanodegree Program

PID Control

In this project,PID is focus on reduce CTE which is cross-track error.The main control output is the steering angle.In the following discussion, mainly for the bicycle model, the output is considered to be the steering angle.

P-Proportional

Represents the reverse steering angle, which is used to bring the system closer to expectations, ie to reduce cte. The output is p * cte . But just using the p parameter results in the model continuing to oscillate in the x-axis. The p control considers the current error. p代表反向的转向角度，用来使系统接近预期，即减少cte。输出结果为 p * cte 。但是只是用p参数的结果是模型将会持续在x轴震荡。p控制考虑当前误差。

I- Intergral

Consider the error in the past, add the error of a certain period of time to the system in a certain proportion. But when there are too many i parameters, there will be overshoot.

考虑过去误差，将一段时间的误差以一定比例加入到系统中。但是当i参数过多时，会出现过冲的现象。

D-derivative

The difference between this cte and the last cte is added to the system through a certain percentage. Consider future errors. The larger the value, the faster the system will reflect on the output. It can improve the setting time and system stability. 将本次cte和上次cte的差值通过一定比例加入系统。考虑将来的误差。该值越大，系统对输出结果反映越迅速。可以提升整定时间和系统稳定性。

PID Hyperparameter Selcection

twiddle

my first ideas is blow(in pid.cpp line 69 - 126):

I chose a ramdom value which P I D equal [0.5,0.5,0.0005]
I achieved a twiddle in PID class and let it output the result and CTE(the detail in PID::twiddle() line 59)
I start run the car ,and before the car away from the road,I select the PID when the cte is best
I input the new pid param into the param, and repeat the 2 ,3 ,4 step until the car could always run in the road smoothly. but in the real implementation,I found that, when I use the twiddle ,the car is easily out of the road. so I changed the idea,and manually tuning the hyperparameters.

manual tuning

1.我首先调节P 参数设置 i和d参数为0. 2. 增加d参数到震荡消退。 3. 观察车驶出道路的原因，如果是因为反应过慢增加kp 和kd ki 如果是因为震荡(过冲) 减少 kp 和 ki

Dependencies

cmake >= 3.5
All OSes: click here for installation instructions
make >= 4.1(mac, linux), 3.81(Windows)
- Linux: make is installed by default on most Linux distros
- Mac: install Xcode command line tools to get make
- Windows: Click here for installation instructions
gcc/g++ >= 5.4
- Linux: gcc / g++ is installed by default on most Linux distros
- Mac: same deal as make - [install Xcode command line tools]((https://developer.apple.com/xcode/features/)
- Windows: recommend using MinGW
uWebSockets
- Run either ./install-mac.sh or ./install-ubuntu.sh.
- If you install from source, checkout to commit e94b6e1, i.e.
```
git clone https://github.com/uWebSockets/uWebSockets 
cd uWebSockets
git checkout e94b6e1
```
  Some function signatures have changed in v0.14.x. See this PR for more details.
Simulator. You can download these from the project intro page in the classroom.

There's an experimental patch for windows in this PR

Basic Build Instructions

Clone this repo.
Make a build directory: mkdir build && cd build
Compile: cmake .. && make
Run it: ./pid.

Tips for setting up your environment can be found here

shenmutong/CarND-PID-Control-Project