Home Page > Media > Paper > Culturally Sensitive Human-Robot Interaction: A Case Study of the Pepper Robot

Culturally Sensitive Human-Robot Interaction: A Case Study of the Pepper Robot

In the current wave of technology,people will engage in more interactions with social robots in their daily lives. However, for these interactions to be effective, people must trust the robots. A robot that can understand human expectations, recognize emotions, and act independently accordingly is a culturally competent social robot.

Just like in other parts of the world, Africa also needs robotics technology, using social robots and human-robot interaction to support various industries and sectors such as healthcare, education, and services. Therefore, understanding and incorporating cultural sensitivity into human-robot interaction is crucial for building trust and promoting the adoption of social robots in different cultural environments. In this process, it is necessary to construct human-robot interaction behavior patterns that reflect linguistic, non-linguistic, and spatial cues based on the socio-cultural context of Africa.


Research Process


Africa is a diverse continent with approximately 1,500 to 2,000 officially recognized languages spoken by 3,000 ethnic groups. The study conducted an initial survey among individuals from 8 African countries (Benin, Burundi, Ethiopia, Ghana, Kenya, Nigeria, Rwanda, and Gambia), with participants aged between 20 and 25 years. The initial focus of the survey was on Rwanda and South Africa.

文化敏感.jpg

Sample of African Cultural Knowledge


Software Interaction Primitives for Human-Robot Interaction


To ensure that social robots can engage in predictable and effective interactions that align with African socio-cultural norms, researchers have embedded them within reconfigurable and reusable interaction primitives. Based on outlined scenarios, researchers identified several software interaction primitives and their corresponding parameters. Each software interaction primitive is a function with a set of different parameters, such as bowing (angle, speed). Software interaction primitives are categorized into four types: iconic gestures, deictic gestures, gaze, and posture.

敏感2.jpg

Specific Software Interaction Primitives


System Architecture


敏感3.jpg


Interaction Flowchart


The Interaction Flow consists of two main subsystems and two external knowledge bases. The two subsystems are the Robot Perception and Interpretation Subsystem and the Robot Behavior Subsystem. The Robot Perception and Interpretation Subsystem includes two parts: (1) Sensing and Analysis, and (2) Event Detection and Classification.


The Sensing and Analysis subsystem is responsible for detecting and locating humans, faces, eyes, gaze direction, hands, sounds, and the robot itself (enabling the robot to navigate within its environment). The Event Detection and Classification subsystem is responsible for detecting and classifying events in the robot's environment, including speech, gestures, proximity, engagement, and retreat events of the interaction partner, as well as mutual gaze.


he Robot Behavior Subsystem consists of four parts: Reactive Behaviors, Attention, Interaction Manager, and Expression and Actuation.


The two external knowledge bases are the African Cultural Knowledge Base, which describes the social interactions between the Pepper robot and humans. The Interaction Manager component uses this African Cultural Knowledge Base to translate it into culturally sensitive robot behaviors that are necessary.


Laboratory Demonstration


The system architecture and components required for culturally sensitive human-robot interaction have been completed and are being used in the laboratory for the cultural sensitivity testing of the Pepper robot. Although the system is still under development, three culturally sensitive behaviors have already been generated and tested on the Pepper robot: welcoming, greeting, and demonstrating gestures.

敏感4.jpg


(a) Illustrates a greeting gesture. Pepper robot opens its hands with palms facing up and slightly tilts its head to show respect.


 In (b), Pepper extends its right hand while moving the left arm back, and slightly bends its torso and head to show respect.


 In (c) and (d), Pepper presents a demonstration behavior. First, Pepper makes eye contact to invite the interaction partner to participate in what Pepper is demonstrating (c); second, Pepper and the user focus together (d) to indicate interest in what Pepper is demonstrating.

研究


Research Conclusions


The research results indicate that the majority of participants perceived the Pepper robot as empathetic, sociable, capable, and reliable, which aligns with the researchers' expectations of positive evaluations of culturally sensitive robot behaviors.


In our culturally diverse present, socio-cultural factors play a significant role in achieving the trust and acceptance necessary for realizing the potential of artificial intelligence and robotics technology. Therefore, culturally sensitive human-robot interaction is crucial for the application of social robots, whether in Africa or across the globe.


reference:https://ieeexplore.ieee.org/document/10293271


Follow SoftBank Robotics
Explore the Boundless Possibilities of Robotics



相关推荐

Discover the Future of Robotics: Subscribe Now to the SoftBank Team for the Latest News.

Curious for More Details?

Get Started with a Consultation.

Home Page / Media / Paper / Culturally Sensitive Human-Robot Interaction: A Case Study of the Pepper Robot