A Review of Art and Real World Applications of Intelligent Perception Systems

Abstract

Sensory data and AI/ML techniques are crucial to several robotics applications, which is why perception in robots is a hot topic. Some of these applications include: object recognition, scene understanding, environment representation, activity identification, semantic location classification, object modeling, and pedestrian/human detection. Robotic perception, as used in this article, is the collection of machine learning (ML) techniques and methods that allow robots to process sensory data and form conclusions and perform actions accordingly. It is clear that recent development in the field of ML, mostly deep learning methodologies, have led to improvements in robotic perception systems, which in turn make it possible to realize applications and activities that were previously unimaginable. These recent advancements in complex robotic tasks, human-robot interaction, decision-making, and intelligent thought are in part due to the fast development and widespread usage of ML algorithms. This article provides a survey of real-world and state of the art applications of intelligent perception systems in robots.