CAR SIMULATOR TEST ENVIRONMENT FOR ARTIFICIAL NEURON SYSTEM DEVELOPMENT
Abstract
The Department of Electron Devices (Semiconductor Technology Laboratory) is conducting ongoing research projects on implementing Physical Artificial Neurons (ANs ) using Vanadium Oxide (VO2). The networked ANs allow us to get closer to the brain-like operation devices. Understanding the HW-based ANs’ operation requires a stable test environment.[1]
This research focuses on designing a flexible Artificial Neural Networks (ANN) test environment. One interesting engineering problem is a self-driven car study. A key point of HW-based ANN development is using a self-driven car simulator to examine different types of ANNs’ behaviour and operation in a simulated, well-defined environment.[2]
To address the need for energy-efficient ANs, this work employs an STM32 microcontroller unit (MCU) based on the Arm Cortex-M processor. A Python-based car simulator is developed using the PyGame library, featuring manual and self-driving modes. The Neural Network is created for the self-driving car simulator by STM32, and two-way communication protocols are analysed between the MCU-based ANN and PC simulator.
The self-driving car simulator successfully measures distances using radar sensors and translates steering commands into vehicle directions. The communication between the deployed neural network model on the STM32 MCU and the PC car simulator is established, facilitating data exchange.[3]
This research addresses the challenge of testing evolving hardware in real-time environments and emphasises simplifying algorithmic systems and data gathering. The study also highlights the importance of converting data between digital and analogue formats for seamless communication. Successful completion of this work contributes to the testing of physical ANN hardware and efficient communication protocols for real-time deployment.
Keywords:
Physical Artificial Neuron, Self-Driven Car Simulator, Communication Protocols, STM32, MCU, Neural Network, Real-Time Environments.
References:
[1] Q. Zhang, H. Yu, M. Barbiero, B. Wang, and M. Gu, “Artificial neural networks enabled by nanophotonics.
[2] D. Kuan, G. Phipps, and C. Hsueh, “Autonomous Robotic Vehicle Road Following,” IEEE Trans Pattern Anal Mach Intell
[3] STM32 Official Website. [Online]. Available: https://www.st.com/en/microcontrollers-microprocessors/stm32-32-bit-arm-cortex-mcus.html
szerző
-
Megalla Antony Ifj.
Villamosmérnöki szak, mesterképzés
mesterképzés (MA/MSc)
konzulens
-
Dr. Neumann Péter
adjunktus, Elektronikus Eszközök Tanszék
