Human-robot interaction: predicting research agenda by long short-term memory.

Area of Science:

• Robotics and Artificial Intelligence
• Human-Robot Interaction (HRI)
• Industrial Technology (Industry 4.0/5.0)

Background:

• The scientific literature lacks defined research agendas for human-robot interaction (HRI).
• This gap hinders advancements in collaborative robotics and Industry 4.0/5.0 technologies.
• A structured approach is needed to guide future HRI research.

Purpose of the Study:

• To propose and validate a multilayered model for identifying and predicting HRI research topics.
• To establish a precise research agenda for the scientific community in HRI.
• To guide new developments in collaborative robotics and HRI technologies.

Main Methods:

• Data mining and visualization (VantagePoint, Gephi) on 2020-2021 HRI articles to identify themes (Layer 1).
• Natural Language Processing (NLP) and word embeddings to extract key terms (Layer 2).
• Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM) for future topic prediction (Layer 3).

Main Results:

• Layer 1 revealed intensive HRI applications across sectors, emphasizing trust.
• Layer 2 highlighted the importance of vision, sensors, communication, collaboration, and anthropomorphism.
• Layer 3 predicted future topics including design, performance, methods, and controllers for enhanced robot interaction.

Conclusions:

• The proposed multilayered model effectively defines a robust and relevant HRI research agenda.
• The methodology successfully identifies future trends and needs in HRI research.
• This work provides a valuable tool for the scientific community, filling a critical gap in the literature.