Human-Robot Interactions to Bidirectional Align Human-Robot Value for Manipulation Robots

PhD Qualifying Examination


Title: "Human-Robot Interactions to Bidirectional Align Human-Robot Value for 
Manipulation Robots"

by

Miss Hanfang LYU


Abstract:

Bidirectional communication and understanding are essential for teammates to 
work together effectively. The recent surge in popularity of large artificial 
general intelligence (AGI) models has brought the value alignment research 
between humans and AGI systems to a more critical position. Specifically, in 
intelligent manipulation robot systems requiring human assistance to accomplish 
complex tasks, the challenge is to align information, knowledge, intention, or 
social norms of varying importance between humans and robots. On the one hand, 
robots need to understand humans' preferences and take appropriate actions. On 
the other hand, humans need to make sense of the robot's actions and teach or 
provide feedback to the robot.

Recently, considerable alignment research in the human-robot interaction (HRI) 
community aims to bridge the gap by enabling agents to infer human values and 
increasing the system's explainability. This survey reviews the literature on 
interaction techniques for human-robot value alignment in current robot 
learning and HRI research and categorizes them according to the alignment 
directions. The directions of alignment include robot learning human values, 
human understanding robot values, and bidirectional communication between 
humans and robots. The survey also highlights the research challenges and 
opportunities in human-robot interaction for robot value alignment.


Date:                   Monday, 29 April 2024

Time:                   4:30pm - 6:30pm

Venue:                  Room 4472
                        Lifts 25/26

Committee Members:      Dr. Xiaojuan Ma (Supervisor)
                        Prof. Fugee Tsung (Co-Supervisor)
                        Prof. Raymond Wong (Chairperson)
                        Prof. Andrew Horner
                        Prof. Chiew-Lan Tai