Session

Mechatronics, System Engineering and Robotics

Description

As we know the artificial intelligence, and specifically artificial neural networks, have improved rapidly in the last decade which leads to the application of these systems in commercial fields like in buying online products, medical applications, financial applications, etc. But we know also that ANN usually are complex systems that need a lot of computing power in order to function properly which limits their application in number of fields, including here embedded systems because of their limited hardware and software properties. In this paper our goal is to implement a fully functional ANN in ATmega328p microcontroller, which will be programmed and trained in microcontroller. Our study case is solar tracker, where we aim to create a functional tracker which works based on ANN and with very limited hardware resources. With this implementation we aim to prove that ANN can function in practical systems without need of high computing power but just with simple low cost embedded system.

Keywords:

Artificial Intelligence, artificial neural networks, embedded systems, atmega328p.

Session Chair

Peter Kopacek

Session Co-Chair

Fatmir Azemi

Proceedings Editor

Edmond Hajrizi

ISBN

978-9951-437-96-7

First Page

4

Last Page

11

Location

Lipjan, Kosovo

Start Date

31-10-2020 1:00 PM

End Date

31-10-2020 2:30 PM

DOI

10.33107/ubt-ic.2020.429

Download

Included in

Engineering Commons

Share

COinS
 
Oct 31st, 1:00 PM Oct 31st, 2:30 PM

Implementation of Artificial Neural Network in Embedded Systems

Lipjan, Kosovo

As we know the artificial intelligence, and specifically artificial neural networks, have improved rapidly in the last decade which leads to the application of these systems in commercial fields like in buying online products, medical applications, financial applications, etc. But we know also that ANN usually are complex systems that need a lot of computing power in order to function properly which limits their application in number of fields, including here embedded systems because of their limited hardware and software properties. In this paper our goal is to implement a fully functional ANN in ATmega328p microcontroller, which will be programmed and trained in microcontroller. Our study case is solar tracker, where we aim to create a functional tracker which works based on ANN and with very limited hardware resources. With this implementation we aim to prove that ANN can function in practical systems without need of high computing power but just with simple low cost embedded system.