The Indonesian And Malaysian Research Team Examined The Use Of Eye Signs On Servant Robots

JAKARTA - The service robot may be familiar to your ears, because nowadays, service robots are common in many industries, especially to assist humans in carrying out various repeated tasks.

So, to give orders to robots, Human Robot Interactions (HRI) are important. HRI Non Verbal, plays an important role in social interactions, highlighting the need to accurately detect the subject's attention by evaluating programmed gestures.

In a recent study entitled "Non-Verbal Human-Robot Interaction Using Neural Network for the Application of Service Robot" published in the journal IIUM Engineering Journal in January 2023, a team of experts from Indonesia and Malaysia introduced a conceptual attention model algorithm called Attractive Recognition Model (ARM) to recognize a person's attention by increasing the accuracy of detection and subjective experience during HRI Non Verbal.

This research was conducted by several experts, including Prof. Dr. Andi Adriansyah from Mercu Buana University, Universiti University of Tun Hussein Onn Malaysia, namely Zubair Adil Soomro, Abu Ubaidah Shamsudin and Ruzari Abdul Rahim, as well as from Move Robotic SDN BHD, namely Mohd Zoli.

According to Andi in his research, the ARM algorithm they used used used three combined detection models, namely: face tracking, sliding tracking, andhauling.

Face tracking models are trained using the Long Short-Term Memory (LSTM) neural network, which is based on in-depth learning. While eye-slic tracking and eye-brightening use a mathematical model. The eye-brightening model uses random facial marker points to calculate the Eye Aspect Ratio (EAR), which is much more reliable than the previous method, explained the Chancellor of Mercu Buana University.

Andi also mentioned that the facial tracking and slicing experiments carried out were able to detect directions up to a distance of 2 meters. Meanwhile, the tested eyebright model provides an accuracy of 83.33 percent at a distance of up to 2 meters. More than that, the accuracy of the overall attention of ARM reached 85.7 percent.

Based on these results, Andi said this experiment shows that service robots can understand programmed signaling and hence perform certain tasks, such as approaching interested people.

"With the ability to read more accurate human nonverbal signals, service robots will be more effective in helping humans in carrying out repeated tasks in many industries," he concluded.