Thanks to visit codestin.com
Credit goes to github.com

Skip to content

severina19/pidcontroller

Repository files navigation

CarND-Controls-PID

Self-Driving Car Engineer Nanodegree Program


Effect of the P, I, D components on the performance

The Proportional Component

The proportional part of the controller will steer the car in proportion to the CTE, which is the distance between the y-position of the car and the middle line of the road. But if we only use a p-controller, the car will overshoot and get off track quickly. A high gain of this component will also cause high oscillation.

Integral Component

The integral term will not only consider the current error but also the sum of all previous offset. The I-controller can be used to eliminate bias.

Derivative Component

The d-controller considers the rate of change of the error, and will try to bring this rate to 0. So if the car is deviating from the middle of the lane, the steering angle will correct itself faster, and if the car is moving toward the center, the steering angle will be more smooth. If the gain of the D-Component is high, the steering angle of the car will stay nearly constant. A low gain will lead to a high oscillations and result in overshooting.

How I chose the parameters

To begin with, I set the K_p and K_d to 1, and K_i to 0.05. The car was oscillating and immediately went of track. To reduce the oscillation, I increased K_d and reduced K_p, and the performance was slightly better but the car still went off track soon. I reduced the gain of the I-component, and found out that setting this gain to 0 leads to a pretty good result, as there is no bias in our system.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages