When we drive, we use our eyes to decide where to go. The lines on the road that show us where the lanes are act as our constant reference for where to steer the vehicle. Naturally, one of the first things we would like to do in developing a self-driving car is to automatically detect lane lines using an algorithm.
In this project you will detect lane lines in images using Python and OpenCV. OpenCV means "Open-Source Computer Vision", which is a package that has many useful tools for analyzing images.
Finding Lane Lines on the Road
The goals / steps of this project are the following:
- Make a pipeline that finds lane lines on the road
- Reflect on your work in a written report
1. Describe your pipeline. As part of the description, explain how you modified the draw_lines() function.
My pipeline consisted of 6 steps:
- Step 1. Convert to grayscale the image
-
Step 2. Apply gaussian blur
-
Step 3. Find edges by canny alghorithm
- Step 4. Remove all edges out of the region of interest NOTE: It's better remove useless edges after the canny algorithm in order to avoid edges caused by the mask of the region of interest
- Step 5. Draw lane lines using Hough alghorithm
- Step 6. Draw lanes on the original image
In order to draw a single line on the left and right lanes, I modified the draw_lines() function by sperate hough lines in two sets, one for left lane and one for the right lane. The process also remove lines with the slope out of a specific tuned range and note the lenght and the lowest y position on the image for each line that will be used as weight for an average coefficients calculation. In the end calculate lines vertices using an weighted average coefficients.
One potential shortcoming would be what would happen when the road is too bright and the canny algorithm can't identify the edges on grayscale color space.
Another shortcoming could be cracks on the road that can be represented as edges and some cracks in right lane with left slope can change the slope of the left line. Obviously the same could happen for the righ lane
A possible improvement would be to use more than one color space to identify more edges and give more weight on lines that are present in more colorspaces.
Another potential improvement could be to split region of interest in two, one for each lane to identify better edges involved in each lane and exclude potential cracks on the road.