Authors:
Ricardo Santos
;
Mateus Silva
;
Rodrigo Lucas Santos
;
Emerson Klippel
and
Ricardo A. R. Oliveira
Affiliation:
Departmento de Computação - DECOM, Universidade Federal de Ouro Preto - UFOP, Ouro Preto, Brazil
Keyword(s):
Autonomous Mobile Robot, Inspection Robot, Robot, Edge AI, Artificial Intelligence, Deep Learning, CNN, YOLOv7, Feedback Control, Object Detection, Jetson Xavier NX.
Abstract:
Recent technological advances have made possible what we call industry 4.0 in which the industrial environment is increasingly filled with advanced technologies such as artificial intelligence and robotics. Defective products increase the cost of production and in such a dynamic environment manual methods of equipment inspection have low efficiency. In this work we present a robot that can be applied in this scenario performing tasks that require automatic displacement to specific points of the industrial plant. In this robot we use the concept of Edge AI using artificial intelligence in a edge computing device. To perform its locomotion the robot uses computer vision with the brand new YOLOv7 CNN and feedback control. As hardware this robot uses a Jetson Xavier NX, Raspberry Pi 4, a camera and a LIDAR. We also performed a complete performance analysis of the object detection method measuring FPS, consumption of CPU, GPU and RAM.