Comparison of Quantitative Proxemics to measure trust in HRI

Abstract

With the increase in the affordability of robots and development in the robotics field, humans have to work in collaboration with robots in many domestic and industrial tasks. With these advancements in the robotics field, it is required to address the challenge of creating a trustworthy environment. However, while working with robots, it is necessary to maintain some distance from the robot to ensure the safety and comfort of a human. To address this problem, we proposed a virtual reality (VR) based human-robot interaction (HRI) task where we can practice interaction between humans and humanoid robots to analyze the trust of humans in robots in terms of social interaction parameters such as the distance of the robot from the human while moving towards the human. In our research, we show a novel idea to measure the trust of human in a robot through EEG signals and then compare it with the real-world HRI task. The comfortable distance while interacting with the robot is determined through Brain Computer Interface (BCI). Along with the use of survey-based assessment of the subject, a standard, and more efficient BCI system is also used to record the users' brain activity in different HRI zones for the study of human emotional state in these zones. Real-time data is required to analyze the effect of social parameters such as the speed of the robot and the comfortable distance on human mental state and this is done by collecting the electroencephalography (EEG) signals of the participants while they are performing the HRI tasks in both VR and real world. The questionnaire is used to compare the results of the BCI for each subject. Experimental results showed that the level of closeness between the robot and the human can affect the way that the human perceives and interacts with the robot. According to the BCI results most of the participants felt comfortable when the robot enters their personal zone with 28% relaxation and 26% stress level as compared to when the robot entered their intimate zone and the stress level increased to 32% and relaxation decreased to 25%. Participants trust virtual robots more than real robots as they are more comfortable in VR interaction as compared to interacting with real robots. BCI results proved that training in the VR framework improves the real-world HRI experience. This study provides valuable insights into the human's cognitive and emotional response to different HRI zones and highlights the importance of considering social parameters, such as proximity, in the design and development of robots for humans.