Driverless Car : Software Modelling and Design Using Python and Tensorflow

EasyChair Preprint no. 1446

Abstract

For autonomous vehicles, several real – time systems must work tightly together. These real – time systems include environmental mapping and understanding, localization, route planning and movement control. For these real – time systems to have a platform to work on, the self-driving car itself needs to be equipped with the appropriate software infrastructure. For this, the software programming architecture using ROS is presented. One of the key functions i.e. Object Detection is in self-driving cars. Also implementation of different object detection methods for detecting objects in images like Deep Learning and Deep Reinforcement Learning are presented. The Deep Reinforcement Learning uses a CNN and DQN where an agent extracts features and localize the object. The results of different techniques are comparable with high prediction accuracies.

Keyphrases: Deep Reinforcement Learning, driverless car, object detection, TensorFlow

@Booklet{EasyChair:1446,
  author = {Poondru Prithvinath Reddy},
  title = {Driverless Car : Software Modelling and Design Using Python and Tensorflow},
  howpublished = {EasyChair Preprint no. 1446},

  year = {EasyChair, 2019}}