CNN-TensorFlow-Keras-and-Image-processing---Behavioral-clonning-of-self-drive-car

In this project a convolutional neural network (CNN) is trained to learn the behavior of a car using data from a simulator that allows real-time information gathering from the car’s chassis, position and its speed. As a first step the vehicle is driven in a manual mode of simulation for collecting data. Then the neural network uses information from the front-facing, left and right cameras, the car’s position on the lane and its speed to learn the internal representations of the necessary processing steps, such as the detection of road lanes, required speed, and track position. Then the simulation is done in autonomous mode, applying the trained weights of neural network which predicts the steering angle for the car in autonomous mode.