Artificial Vision Algorithms for Socially Assistive Robot Applications: A Review of the Literature

Abstract

Today, computer vision algorithms are very important for different fields and applications, such as closed-circuit television security, health status monitoring, and recognizing a specific person or object and robotics. Regarding this topic, the present paper deals with a recent review of the literature on computer vision algorithms (recognition and tracking of faces, bodies, and objects) oriented towards socially assistive robot applications. The performance, frames per second (FPS) processing speed, and hardware implemented to run the algorithms are highlighted by comparing the available solutions. Moreover, this paper provides general information for researchers interested in knowing which vision algorithms are available, enabling them to select the one that is most suitable to include in their robotic system applications.

SoftSAR: The New Softer Side of Socially Assistive Robots—Soft Robotics with Social Human–Robot Interaction Skills

“…The main sensory systems that have been used in SAR design are 2D and 3D vision [ 70 ], microphones [ 71 ], and tactile sensors [ 71 , 72 ]. They enable SARs to engage in effective HRI.…”

“…Recently, the incorporation of ML and DL methods has been used for object or person classification in SARs [ 106 ], affect classification [ 107 ], and robot assistive behavior decision-making [ 108 ]. To-date, SARs have mainly used CPUs to meet the computational needs of their respective tasks when a SAR does not need to execute high computational algorithms and for cost reduction [ 70 ]. However, when incorporating algorithms, such as real-time object detection YOLO (You Only Look Once) and FCNN (Fully Convolutional Neural Network), to be executed together, it is necessary to use computer resources with greater-than-regular hardware (RAM and slow processors) and video graphic cards with GPUs [ 70 ].…”

“…To-date, SARs have mainly used CPUs to meet the computational needs of their respective tasks when a SAR does not need to execute high computational algorithms and for cost reduction [ 70 ]. However, when incorporating algorithms, such as real-time object detection YOLO (You Only Look Once) and FCNN (Fully Convolutional Neural Network), to be executed together, it is necessary to use computer resources with greater-than-regular hardware (RAM and slow processors) and video graphic cards with GPUs [ 70 ]. Table 3 presents a number of existing SAR platforms with their corresponding onboard CPU and input/output computation requirements.…”

Socially-Assistive Robots Using Empathy untuk Mengurangi Nyeri dan Distress selama Pemasangan IV Line Perifer pada Anak

“…SAR bisa berperan dalam terapi afektif, pelatihan kognitif, fasilitator sosial, persahabatan dan terapi fisiologis (Abdi et al, 2017) Penelitian dari Hannah Louise Bradwell (2021) merekomendasikan beberapa perubahan desain untuk penggunaan Socially-Assistive Robots di ruang lingkup kesehatan dan sosial, diantaranya peningkatan mobilitas, pengenalan suara, desain dan penampilan (Bradwell et al, 2021). Montaño-Serrano, V.M et al (2021) dalam penelitiannya menyatakan algoritma visi yang sangat penting untuk SAR adalah kinerja dan kecepatan (Manuel Montaño-Serrano et al, 2021).…”

The Use of Terrestrial and Maritime Autonomous Vehicles in Nonintrusive Object Inspection

“…During a passive inspection, security should be ensured without direct interaction with the object of the search, even without informing them about the searching process [ 7 , 8 , 9 ]. With this aim, different object and behavior recognition techniques are employed, typically based on processing the images received from surveillance video cameras installed in public places [ 10 , 11 , 12 , 13 , 14 ].…”