An implementation of NVIDIA's Convolution Neural Net with quantized weights using AlphaRT to train autonomous vehicle
The folder modified_carla_scripts contains carla py files modified for training and for running on controls(steer, speed) returned by the model
The contents of the documentation is as follows. • Motivation and Brief Introduction. • Behavioural Cloning on the Udacity Simulator. • Introduction to the Carla Simulator and the code architecture. • Added Scripts to extract information from the Carla Client. • Comparison of performance between RGB, Depth and Semantic Segmented Images. • Current Scenario and future prospects. • Brief ending with the final Python Client Structure