This is our best autopylot yet. It finds a line and follows it, obeys traffic signs and is a good driver. It can also drive you to the desired destination. It uses OpenCV for traffic sign, lane and marker detection.
- script runs and exits if you use force exit command
- multithreaded opencv image process preview in browser on port 1234
- single line curve detection and adaption
Use pip to manage dependencies.
- opencv-python
- opencv-contrib-python
- numpy (comes with opencv-python)
- (OPTIONAL: for traffic sign detection) tensorflow
- (OPTIONAL: for opencv preview in browser) flask
- (OPTIONAL: for use on actual Raspberry Pi) picamera
- (OPTIONAL: for use on actual Raspberry Pi with ROS) rospy
- (OPTIONAL: for lights control) PoKeysLib (can be found on bitbucket)
OPTIONAL libraries are not required. The code runs even without them, but some of the functions are ignored in such case. Without picamera, it uses sample.jpg for processing, without rospy, there is no control, but the commands are printed in console, without flask, there is no preview in browser.
You need to run both main.py (on Python 3) and ros_server.py (on Python 2). They communicate using unix socket. On test environment, if there is no unix socket, data is not sent.
In order for PoKeysLib to operate with our project, uncomment usefastusb option PoKeysLib.h
