AUTONOMOUS VEHICLE CONTROL SYSTEM USING CONVOLUTIONAL NEURAL NETWORK
Author(s):
Ravi k Sah, Omkar Poshatwar, Abhijit Howal, Prashant Sharma
Keywords:
Convolutional Neural Networks, Deep learning, Autonomous Vehicle, Image Recognition, Obstacle Detection, Depth Estimation, Deep Learning, Machine Vision, Autonomous/Self-Driving Vehicles
Abstract
Convolutional neural networks (CNNs) are a type of layered deep neural network comprised of artificial neurons. These neurons are initially taught a set of rules and conditions, through training, which dictate whether they will fire when given varying inputs. CNNs learn as they are used and make future decisions based on both the taught and learned information. A common application of CNNs is object and feature recognition in images. The CNN identifies features in an image by analyzing data pixels through layers of neurons. This is particularly useful in the field of autonomous vehicles where CNNs can be used to process driving footage and identify possible obstacles. CNNs will often classify sections of the preset image grid that potentially contain an obstacle. Errors that occur are fed back into the network for reclassification and further learning. After the analysis is complete and a final conclusion has been reached, the CNN outputs a signal for the vehicle to perform an action: keep driving, stop, turn, etc.
Article Details
Unique Paper ID: 146596
Publication Volume & Issue: Volume 5, Issue 1
Page(s): 154 - 158
Article Preview & Download
Share This Article
Join our RMS
Conference Alert
NCSEM 2024
National Conference on Sustainable Engineering and Management - 2024