網頁

星期日, 2月 26, 2017

Socially Adaptive Robots is on its way


By Lucas DiLeo
Contributing writer


Socially Adaptive Robots is a new field that addresses how robots can better interface with and engage with humans. Increased personalized and human-like interaction will be important as the industry moves from performing repetitive tasks or responding to requests, to developing robots who can assist with rehabilitation, coaching, behavioral modification or patient care.

Dr. Maja Matarić from the University of Southern California, presented her work at the Harvard Institute for Applied Computing. Dr. Matarić is professor and Chan Soon-Shiong chair in Computer Science Department, Neuroscience Program, and the Department of Pediatrics at the University of Southern California, founding director of the USC Robotics and Autonomous Systems Center (RASC), co-director of the USC Robotics Research Lab and Vice Dean for Research in the USC Viterbi School of Engineering. 

Dr. Matarić's work explores how people interface with robots and how to make robots more engaging, communicative and responsive. This will determine the effectiveness of robots serving as trainers and health monitors and educators. 

She presented the results of projects that utilized robots for interfacing and motivating patients for stroke and rehabilitation, Alzheimer's, and Autism, as well as educating young children. 

A project for stroke rehabilitation, for instance, had the robot playfully interacting with a patient who was trying to game the sensors tracking her exercise actions. Some therapies require patients to be motivated to participate. Ordering or instructing will not work for many.

Future Robots will need to learn how to interact with people during a conversation. Such as judging how much feedback to give and the type of feedback - say based on goals or speed of learning. And style of training: should the robot be programmed to teach thru imitating, or understanding the issues that matter, or teach thru demonstrating.

According to Dr. Matarić, having a body is complex - we are hard-wired for many aspects of how we perceive people, such as body language. However, it is difficult for a robot to change what it looks like, so her lab experiments with behavior instead. One example: testing how a robot faces a person and interacts with them- looking directly at them, establishing social distance. 

Moreover, social dynamics - conversations and interactions - change over time as the mood and level of interest of the human interface changes over time. And facial signals can be tricky. As people are often immersed in problem solving so their expressions may indicate their attention to the problem, not their emotional response.

Another lab project entailed small group encounters, where the robot served as a moderator. Within groups there is often the challenge of a strong leader crowding out the participation of others. Where the robot moderator was instructed to look at and ask questions of the person saying the least - there was greater participation and group adhesion.

A final project presented using robots to explore ways to work with children visiting doctor's offices and testing different pain coping strategies. They looked at empathy, distraction and content - explaining what was going on.

In the future, Dr. Matarić's lab will be undertaking a robotic care project in an elder home.


Key takeaway from Dr. Matarić's talk: To be effective companions, trainers or health coaches, robots will need to better judge the emotions of the humans they work with, and intuit their needs and intentions, and better relate to and engage with their human counterparts in terms of the style, content and tone of their conversation.