Overview / Usage

Product Scope

Road safety has been an issue for as long as cars have been in existence. Over 1.3 million people die of road accidents every year across the globe, most of which are preventable. Ever-rising road traffic has led to an exponential increase in commute time. This has a direct impact not only on people's productivity but also on the environment.

Recent developments in machine learning and artificial intelligence along with the ever-increasing performance of modern day computers have enabled the use of these technologies in developing self-driving cars. These cars have several advantages, as described below:

Better road safety: Machines are not prone to human-error and distractions, leading to swift and appropriate responses in real-time road conditions.
Reduced commute time: With cars communicating with each other and using modern GPS systems, commute times can be greatly reduced as self-driving cars reduce the "phantom effect" in modern-day traffic.
Increased productivity: Reduced commute times mean more time can be spent on what matters more.
Reduced expenditure: Reduction in accidents will directly lead to reduced expenditure on damages.
Environment-friendly: Efficient driving styles of the self-driving car will lead to lower emissions.
Solution to parking problem: Most of the modern cities face parking problems and which can be resolved by this solution.
Better traffic discipline: Better law enforcement can be achieved and traffic can be managed by capping speed in various regions.
Potential for a new design: Because a vehicle may eventually function as a self-guided train car, the potential for new car designs is huge. With no need for complicated driving tools, self-driving cars could include new ways to relax or to stay entertained.

Novelty of the Project

Our project was squarely aimed at developing a simple RC self-driving car. This involved developing code for a cloud-based machine learning solution, an Android application for data collection and an embedded system to actuate the car's motors. This project, thus, represents a complete self-driving car solution, instead of forming a part of a complete system.

The entire concept of self-driving cars is alien to the general Indian population. Not much research is being done in India right now on this. This car represents a small, albeit significant, step towards the development of a full-sized, fully-functional self-driving car. Ours is a scalable solution, i.e. with minor modifications, it can be scaled up to a full network of self-driving cars. Communication between such a network is a future research topic that can be looked into.

Methodology / Approach

Visit the URL: https://github.com/GeekyShiva/Self-Driving-Car/blob/master/Project%20Docs/SRS.md

Technologies Used

The tools used in this project and described in this document are:

TensorFlow library for machine learning
Vision API by Google Cloud Platform and TensorFlow.

The hardware used in this project and described in this document are:

RC Car serving as the actual self-driving car
Arduino Uno used for controlling the car speed and turning radius
Ultrasonic sensors to detect objects in front of the car
Raspberry Pi used for capturing and transmission of camera data and communicating with Google Cloud Platform
Android App used for capturing objects in front of the car

Repository

https://github.com/GeekyShiva/Self-Driving-Car