Finding Lane Lines on the Road

The goal of this project is to create a pipeline that finds lane lines on the road

---

1. Pipeline description.

My pipeline consisted of 5 steps. First, I converted the images to grayscale, then I applied Gaussian smoothing and Canny procedure to detect edges.

In order to draw a single line on the left and right lanes, I created the draw_lines() function by averaging line coordinates. Also, I decided to make separate mask for left and right sides.

You can see the final Jupyter Nonebook with all steps by the here

As the result you can see detected lines on the following picture:

2. Potential shortcomings with the current pipeline

The pipeline is unreliable when a car turns left or right and on too bright road segments.

3. Possible improvements to your pipeline

A possible improvement would be to replace an average function by weighted average or to take into account X and Y of lines that are closest to the car or make a polynomial (n = 2) regression

Another potential improvement could be to make two regressions (Y = AX + B and Y = AX^2 + B*X + C) and extrapolate lines by one which is the most predictive.