Development and Testing of a Multimodal Acquisition Platform for Human-Robot Interaction Affective Studies

Nicole Lazzeri, Daniele Mazzei, Danilo De Rossi

Abstract


Human-Robot Interaction (HRI) studies have recently received increasing attention in various fields, from academic communities to engineering firms and the media. Many researchers have been focusing on the development of tools to evaluate the performance of robotic systems and studying how to extend the range of robot interaction modalities and contexts. Because people are emotionally engaged when interacting with computers and robots, researchers have been focusing attention on the study of affective human-robot interaction. This new field of study requires the integration of various approaches typical of different research backgrounds, such as psychology and engineering, to gain more insight into the human-robot affective interaction.

In this paper, we report the development of a multimodal acquisition platform called HIPOP (Human Interaction Pervasive Observation Platform). HIPOP is a modular data-gathering platform based on various hardware and software units that can be easily used to create a custom acquisition setup for HRI studies. The platform uses modules for physiological signals, eye gaze, video and audio acquisition to perform an integrated affective and behavioral analysis. It is also possible to include new hardware devices into the platform. The open-source hardware and software revolution has made many high-quality commercial and open-source products freely available for HRI and HCI research. These devices are currently most often used for data acquisition and robot control, and they can be easily included in HIPOP.

Technical tests demonstrated the ability of HIPOP to reliably acquire a large set of data in terms of failure management and data synchronization. The platform was able to automatically recover from errors and faults without affecting the entire system, and the misalignment observed in the acquired data was not significant and did not affect the multimodal analysis. HIPOP was also tested in the context of the FACET (FACE Therapy) project, in which a humanoid robot called FACE (Facial Automaton for Conveying Emotions) was used to convey affective stimuli to children with autism. In the FACET project, psychologists without technical skills were able to use HIPOP to collect the data needed for their experiments without dealing with hardware issues, data integration challenges, or synchronization problems. The FACET case study highlighted the real core feature of the HIPOP platform (i.e., multimodal data integration and fusion). This analytical approach allowed psychologists to study both behavioral and psychophysiological reactions to obtain a more complete view of the subjects’ state during interaction with the robot.

These results indicate that HIPOP could become an innovative tool for HRI affective studies aimed at inferring a more detailed view of a subject’s feelings and behavior during interaction with affective and empathic robots.


Keywords


human-robot interaction, affective computing, multimodal approach, physiological signals

Full Text:

PDF


DOI: https://doi.org/10.5898/JHRI.3.2.Lazzeri

Refbacks

  • There are currently no refbacks.




Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.