panther97 / self-driving-car-using-behavioral-cloning

Self-driving-car-using-behavioral-cloning

The goal of this project is to train a convolutional neural networks based behavioral cloning model to drive a car autonomously. This will be done in a simulator provided by Udacity. Various computer vision principles were used in the development. The simulator is used for both training and testing. To collect the data, the car was run on a virtual track in training mode. This data was fed for training to our model. The network picks up on our driving decisions corresponding to the state of the road, such as when to turn. The model is then saved and is run in the autonomous/ testing mode of the simulator in order to test its performance (on a track that it has never seen before).

For further understanding of project, read the project report attached.
CV_Project_Report.pdf
For resources such as Demo of car running in virtual environment,model semantics,etc. follow the following link. https://drive.google.com/file/d/1cqTQ37J3oo8hfgV2vsYfKQLsgv0v4OID/view?usp=sharing

Tech Stack Used:
1. Python
2. Keras
3. Flask
4. Socketio
5. Udacity Simulator

How to run:
1. Run requirements.txt
2. Run Drive.py
3. Open the simulator
4. Click on "Autonomous Mode" in the simulator
5. Watch the car self-drive :)