AI-based Decision Making for Strategically Empathetic Robot

Description

Multiple PhD positions are available on developing strategically empathetic robot. In this project, we are interested to approach the problem from a more general viewpoint, rather than bogged down to the particular types of ethics or normal principles. Specifically, we are interested to endow robots with the ability to accomplish its task while accounting for the concerns of others —that is, to strategically emphatize with the human it collaborates or interacts with. We will frame this problem as decision making in a partially observable world, where the partial observability is the concerns of other humans. Specifically, we will design and develop novel approaches to account for the mentioned concerns as extensions of Partially Observable Markov Decision Processes (POMDPs).
 
This work is also part of the ANU Humanising Machine Intelligence project.
 
 

Requirements

  • A first class honours or equivalent in Computer Science, Software Engineering, Mathematics, Mechatronics, or related fields.
  • Fluent in programming with high level programming language
  • Performing well in Artificial Intelligence or Machine Learning course is a plus.
  • Experience working with robots is beneficial but not required
 

Keywords

Planning under uncertainty, Stochastic Games, Reinforcement Learning, POMDPs

Updated:  10 August 2021/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing