Road Lane Detection with Raspberry Pi

Simple road lane detection on Raspberry Pi 3 using OpenCV and Python. Frame rates obtained up to 17 FPS.

Road Lane Detection System

Self-driving cars are one of the new trends in the modern world. They use very sophisticated control systems and engineering techniques to maneuver the vehicle. Road lane detection is one of the important things in the vehicle navigation. Here I'm describing a simple and fast lane detection using Raspberry pi 3 and computer vision techniques. For fast calculation I just avoided linear regression method. This method gives the good result for low noise environment but for complex scenes advanced statistical and image processing techniques are needed.

Hardware setup

Connect the camera with your Pi

Camera configuration

Follow this link for camera setup https://www.raspberrypi.org/documentation/configuration/camera.md

Software setup

Install OpenCV for python. Follow these instructions to install OpenCV. These instructions were copied from https://raspberrypi.stackexchange.com.

Generic stuff

sudo apt-get update
sudo apt-get upgrade
sudo rpi-update
sudo reboot
sudo apt-get install build-essential git cmake pkg-config
sudo apt-get install libjpeg-dev libtiff5-dev libjasper-dev libpng12-dev
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev
sudo apt-get install libxvidcore-dev libx264-dev
sudo apt-get install libgtk2.0-dev
sudo apt-get install libatlas-base-dev gfortran
cd ~
git clone 
cd opencv
git checkout 3.1.0
cd ~
git clone 
cd opencv_contrib
git checkout 3.1.0

If you want to use OpenCV with python 2.7 :

sudo apt-get install python2.7-dev
wget 
sudo python 
pip install numpy
cd ~/opencv
mkdir build
cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE \
    -D INSTALL_C_EXAMPLES=OFF \
    -D INSTALL_PYTHON_EXAMPLES=ON \
    -D OPENCV_EXTRA_MODULES_PATH=~/opencv_contrib/modules \
    -D BUILD_EXAMPLES=ON ..
make -j4
sudo make install
sudo ldconfig

If you want to use OpenCV with Python 3:

sudo apt-get install python3-dev
wget 
sudo python3 
pip install numpy
cd ~/opencv
mkdir build
cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE \
    -D CMAKE_INSTALL_PREFIX=/usr/local \
    -D INSTALL_C_EXAMPLES=OFF \
    -D INSTALL_PYTHON_EXAMPLES=ON \
    -D OPENCV_EXTRA_MODULES_PATH=~/opencv_contrib/modules \
    -D BUILD_EXAMPLES=ON ..
make -j4
sudo make install
sudo ldconfig

It takes around 2 hours so wait it.During this time we can learn about Hough-Transform. This technique is the key thing behind most of the practical lane detection algorithm.

What is Hough Transform?

Watch this video for better understanding. Thanks to Thales Sehn Körting -the creator of this video.

For opencv-python documentation http://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_houghlines/py_houghlines.html

python code

from picamera.array import PiRGBArray
import RPi.GPIO as GPIO
from picamera import PiCamera
import time
import cv2
import numpy as np
import math
GPIO.setmode(GPIO.BOARD)
GPIO.setup(7, GPIO.OUT)
GPIO.setup(8, GPIO.OUT)
theta=0
minLineLength = 5
maxLineGap = 10
camera = PiCamera()
camera.resolution = (640, 480)
camera.framerate = 30
rawCapture = PiRGBArray(camera, size=(640, 480))
time.sleep(0.1)
for frame in camera.capture_continuous(rawCapture, format="bgr", use_video_port=True):
   image = frame.array
   gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
   blurred = cv2.GaussianBlur(gray, (5, 5), 0)
   edged = cv2.Canny(blurred, 85, 85)
   lines = cv2.HoughLinesP(edged,1,np.pi/180,10,minLineLength,maxLineGap)
   if(lines !=None):
       for x in range(0, len(lines)):
           for x1,y1,x2,y2 in lines[x]:
               cv2.line(image,(x1,y1),(x2,y2),(0,255,0),2)
               theta=theta+math.atan2((y2-y1),(x2-x1))
   #print(theta)GPIO pins were connected to arduino for servo steering control
   threshold=6
   if(theta>threshold):
       GPIO.output(7,True)
       GPIO.output(8,False)
       print("left")
   if(theta<-threshold):
       GPIO.output(8,True)
       GPIO.output(7,False)
       print("right")
   if(abs(theta)<threshold):
      GPIO.output(8,False)
      GPIO.output(7,False)
      print "straight"
   theta=0
   cv2.imshow("Frame",image)
   key = cv2.waitKey(1) & 0xFF
   rawCapture.truncate(0)
   if key == ord("q"):
       break

Some sample output results:

left road

right road

Straight road

mechanical setup

The GPIO pins are connected to Arduino mega for servo motor control.

#include <Servo.h>
Servo myservo;
void setup() {
  myservo.attach(10);//attach servo motor PWM(orange) wire to pin 10 
  pinMode(0, INPUT);//attach GPIO 7&8 pins to arduino pin 0&1
  pinMode(1,INPUT);
void loop() {
           if(digitalRead(0)==HIGH && digitalRead(1)==LOW)
                {
                      myservo.write(118);
                }
          if(digitalRead(1)==HIGH && digitalRead(0)==LOW)
                {
                      myservo.write(62);
                }
          if(digitalRead(1)==LOW && digitalRead(0)==LOW)
                {
                       myservo.write(90);
                } 
}

I am not a good coder please give suggestions and feedback in the comment section.