Developing Technologies for Regulating Robot Persuasion in Human-Robot Interaction

点击翻译按钮获取中文摘要

Robots are emerging as viable tools in domains including personal care, education, and search and rescue, where they increasingly work alongside people as autonomous or semi-autonomous tools and assistants. In addition to the difficult robotics engineering challenges, robots that enter people's spaces must also have effective interface technologies that support human-robot interaction and enable the general public - not only robotics experts - to work with robots, give commands, and to get data from them. Robots collocated with people use human-like social techniques such as gestures or gaze, leveraging people's existing inter-personal skills to simplify interaction. These social robots will often be in positions where they offer recommendations to people, and research has shown how expert-reference robots can be more persuasive than anticipated, in part due to people's trust in robotic memory, knowledge, and computational ability. This can be dangerous, for example a healthcare provider may doubt their assessment when contradicted by a specialized healthcare robot, but can also be desirable, for example a persuasive rehabilitation robot could help with recovery exercises. Research has established that robots can be persuasive, but little is known about how to practically create a robot to interact with people to either avoid or encourage persuasion. In the short term, I propose to develop the first interaction framework and technologies that demonstrate how robots can modify their persuasiveness by changing how they interact with people, specifically targeting how a robot can use its body language. In the long term, my program will help build a solid understanding and array of tools for programming robots to interact effectively and appropriately with people. My program will methodologically investigate how robots can use their physical body to impact persuasion; specifically, how robots can use their gestures, posture, gaze, and use of interpersonal space, to increase or decrease persuasiveness. I will implement and formally evaluate original interface technologies and algorithms for enabling robots to leverage these techniques for interaction. For this I will establish the first methodologies and metrics for measuring and comparing persuasiveness of robotic designs, and conduct a series of experiments to develop interface design frameworks for engineering robot persuasion. This work will provide foundations for understanding how nuanced robotic behaviors can impact persuasion, including persuasion-oriented interface guidelines, and some of the first interaction techniques and algorithms for architecting robotic persuasion. Ultimately, this will be important for enabling human-robot interaction programmers to be in control of their robots: to reduce the persuasiveness of robots in potentially dangerous situations, or make robots more convincing when appropriate.