NSF National Robotics Initiative Project Funded (Communicating Physical Interactions)

A new project, entitled “Communicating Physical Interactions” has been funded through the National Science Foundation’s (NSF) National Robotics Initiative (NRI).  An overview of the project is given below:

PI:    
  • Prof. Michael Gleicher
  • Department of Computer Science
  • University of Wisconsin – Madison
Co-PI:
  • Prof. Bilge Mutlu
  • Department of Computer Science
  • University of Wisconsin – Madison
Co-PI:
  • Prof. Michael Zinn
  • Department of Mechanical Engineering
  • University of Wisconsin – Madison

Overview:

In order for robots to become ubiquitous collaborators, they must interact physically with objects in unstructured human environments. They must grasp, push, squeeze, snap, balance, stabilize and hand-off objects. However, not only must they perform these actions adeptly, they must be able to communicate about them effectively with their human collaborators. People must be able to specify to a robot what to do and how to do it; they must be able to interpret what a robot intends to do (and how); and they must be able to monitor the robot as it acts so they can collaborate correctly. Effective communication about physical interactions will be essential if a robot manipulator is going to be driven in real time to deliver care to an older adult, responsively help a technician in a repair task, or be trained by a non-expert to perform a repetitive assembly task. Unfortunately, such communication about physical interactions are difficult as they involve invisible and unfamiliar quantities (e.g., forces and compliances), require communicating plans and contingencies, and often require communicating about what did not (or should not) happen as well as what did (or should).


This project will improve our ability to communicate with robots about physical interactions. We will develop a better understanding of how people think about and communicate about physical interactions through a series of human studies. We will use these insights to develop methods that address specific issues in human robot interaction. We will embody these methods in prototype systems that demonstrate the potential for human robot collaboration in various scenarios and allow us to evaluate the performance  of collaboration methods in context. Our success will enable a new generation of robot applications, and advance the vision of ubiquitous, collaborative robots.