Detección y clasificación de obstáculos mediante visión estéreo en la toma inteligente de decisiones para robots autónomos en espacios reducidos.
Year: 2020
Type: dissertation
Abstract: As the use of autonomous navigation robots within human environments increases, the need \nto improve the security measures they integrate so that their mobility does not represent a risk \nfor people and animals, becomes more evident. This work presents a software tool for the \ndetection and classification of nearby obstacles using artificial vision algorithms, aimed at \nmaking decisions that protect the physical integrity of children and pets that are in the around \nautonomous robots. Depth maps generated by a stereo camera are used to identify the critical \npath and obstacles, along with a convolutional neural network for their classification. A final \ndecision-making stage collects the information obtained to suggest the most appropriate \nactions according to each case, exploring the possibility of manipulating non-critical objects if \nnecessary.
Author: José Daniel Báez-Maldonado
Institution: Universidad Internacional de La Rioja
Language: Spanish
Topic: Advanced Vision and Imaging
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3125529238
