Projects related to self driving cars

During my second year of college (2011), I made a prototype to steer a car using hand gestures using computer vision. Fast forward six years, and I got selected for an exhilarating course on self-driving cars. In the final project, the code I developed with my team was tested on a real car. Ever since the AlexNet was introduced in 2012 and progress in GPU hardware and data availability, there is incredible progress in computer vision.
In this course, I worked on several projects related to computer vision, sensor fusion, path planning and control systems. Some of the selected projects are :

Lane detection

Using traditional CV techniques, I was able to identify and track lanes.

Traffic sign detection

Using Object detection techniques like YOLO I trained a NN to detect traffic signals and read the signs.

Behavioural cloning

AI learns from our training data. Before this project, I trained several ML models on different tasks. In this project, I trained the NN to learn the way I drive and imitate my style in different situations. The way it works is I train a CNN-based model to work on images from the dash cam and predict the driving angle. Watch the above video to learn more.

Road identification

In this project, I used a particular task of CV called image segmentation to identify pixels of the road.

Final project

For the final project, a team of three and I developed code with different algorithms and deployed it in ROS and eventually on a car. Our code was tested in a real car, and the car autonomously manoeuvred through obstacles and obeyed traffic signals.
The link to my code and project files : link