Multivariate Prediction of Correct Lane for Autonomous Electric Vehicle Using Deep Learning Models

The target of a self-driving car researches is to build a better autonomous driver. We control the car to be able to drive itself without falling off the track using appropriate accelerating and braking. Using udacity as an open source simulator to depict a real-time environment. With the help of a model trained by deep neural networks, we mimic the driving behavior of a human on the simulator (Behavioral Cloning) [1]. We use Keras and a high-level API that uses TensorFlow as the backend for dataflow programming. Keras provides sequential models for building a linear stack of network layers. CNN models (Neural network layers were optimized in series combinations of Time-Distributed Convolution Layers, Maxpooling, Flatten, Dense, Dropout and so on) are experimented for reaching the best model of control. Finally, we concluded that model 2 performed the best at the end of 60 epochs with the least loss = 0.007.