Autonomous Car - Final Project

Final Project - Ariel University Department of Computer Science 2022.

Instructed by Professor Boaz Ben Moshe.

Introduction :

The world is progressing towards autonomous vehicles with autonomous control, one of the main systems required for the field is a reliable system that can scan its environment and give an accurate indication of when an accident danger occurs and prevent it by braking / acceleration.

For this to happen - the system needs to be sensitive and accurate and know how to scan the environment in real time and act accordingly - which is why we developed the tool we chose.

Project goal : the project is designed to develop a solution for navigation in autonomous vehicles (in a simulator or a real car), emphasizing the reliability of the autonomous software and in addition to conducting a review of the existing technologies on the market and comparing them.

Part 1 - Overview :

The following research is written for promotional purposes and the development of broader research in the field of autonomous vehicles. In the review we will compare ARCore - a platform For building augmented reality experiences by Google map and RTAB map - SLAM approach based on RGB-D, stereo and Lidar Graph Close detector. Goal - getting to know the platforms and the technology behind them. Conducting various experiments - in the light and in the dark, inside and outside. Forming conclusions - advantages and disadvantages of each of the platforms. Using the appropriate and correct platform for part 2 of the project.

For the full review - https://github.com/EladVaknin/Final_Project_autonomous_car/tree/main/Overview

Part 2 - Augmented Reality APP :

After performing the above review, we chose to focus on Google's ARcore technology and use it to develop a mobile application that will be a kind of car simulator - the application knows how to identify surfaces and distinguish between them, the application knows to detect potential transitions (we created a data set and taught the application to identify . In further development the application will know how to create a trajectory and an object that will execute the trajectory. With the help of the route that was opened and the features we developed, the application will serve as a simulator for autonomous driving.

The application was developed in the Unity work environment with the AR Foundation and ARCore XR Plug-in libraries.

Exampels:

Surface identification and modeling:

Detection of potential passages (eg door):

How to install ?! :

1. Make sure your Android device is listed https://developers.google.com/ar/devices?hl=en
2. Make sure your device supports OpenGLES3 and Android 8.0 and above.
3. Download the APK file from Builds folder - https://github.com/EladVaknin/Final_Project_autonomous_car/tree/main/Builds.
4. Allow access to the camera when using the application.

Architecture schema:

Participants :

• Afik Peretz
• Elad Vaknin
• Avigael Abitbol