|
|
Robotics Institute
Seminar, April 29
Time
and Place | Seminar Abstract | Speaker
Biography | Speaker Appointments
Shared Control of Assembly Tasks by Robots
and Humans
|
|
Sanjiv Singh Associate Research Professor |
|
Time and Place |
Mauldin
Auditorium (NSH 1305)
Refreshments 3:15 pm
Talk 3:30 pm
An
important class of multi-robot coordination consists of tasks that require
tightly-coupled coordination of more than one machine. For this class of tasks, coordination of
multiple independent degrees of freedom is a matter of necessity rather than
one of efficiency. For example, within
our lifetime, teams of robots will assemble structures in space. Such assembly is expected to be complex
enough that humans and robots will need to share control to be both reliable
and efficient.
To
test our ideas, we are developing a coordinated team of robots to assemble
structures, a task that can't be performed by any single robot. Even simple operations in this domain require
complex interaction between multiple robots and the number of contingencies
that must be addressed by a fully autonomous system is prohibitively
large. This scenario forces
incorporation of a human operator. Ideally, we would like a seamless interface
between the robots and the operator such that the operator can interact with
the system by helping it be more efficient or get out of a stuck condition or
by explicitly controlling a robot to perform a task that it is not capable of
performing itself.
We
have developed an architecture that allows autonomy to “slide”
between the robot team and a human operator. The operator can step in and take
over control of part or all of the assembly either when the system discovers
that it is failing and requests help or if the operator notices an opportunity
to make performance more effective. I
will present recent results in efficiency and reliability from an extended
series of experiments in which a team of three robots and an operator work
together to assemble structures across the spectrum of autonomy from fully teleoperated to fully autonomous. These results quantify
the benefit of the shared control of assembly by robot and human teams. I will conclude with pointing to current
research questions in this area—how we can analyze plans for assembly by
multiple agents, how the robots autonomously learn about the operator and how
humans participating in such teams should direct the robots.
|
Speaker Biography |
Arriving at Carnegie Mellon in 1985, Sanjiv Singh has been an engineer, student, scientist and
more recently a professor at the Robotics Institute. His current work focuses
on two research themes. The first
develops perception (collision avoidance and egomotion
estimation) for robots operating in natural environments. The other develops coordination for multiple
agents in dynamic environments. He is
the Editor-in-Chief, Journal of Field Robotics.
For appointments, please contact Sanjiv Singh (ssingh@ri.cmu.edu).
The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.