The Adaptive Selfreconfiguring Robotic Factory

1
The Adaptive Self-reconfiguring Robotic Factory

• Daniela Rus, PI, MIT
• Eric Klavins, Co-PI, U. Washington
• Hod Lipson, Co-PI, Cornell University
• Mark Yim, Co-PI, U. Pennsylvania

2
Adaptive Reconfiguring Stochastic FactoryPIs D.
Rus (MIT), E. Klavins (Washington), H. Lipson
(Cornell), M. Yim (Penn)
This rendition illustrates a wheeled and a
climbing robot made of shared materials. They are
in the process of reconfiguring two structures in
a stochastic environment subject to uncertainty
in the availability of components. A testbed will
be created to demonstrate this concept.
Aims
Vision

• Self-assembling structures using parallel
  stochastic and deterministic methods for
  scalability and robustness.
• Distributed adaptive algorithms and analytical
  tools for synthesizing self-organized activities
  and methods for proving performance guarantees
  and stability
• Demonstration using new and existing platforms
  (modular robots and swarm robots) for
  construction and for situational awareness

• Statistical physics modeling and design
  principles for large scale distributed real world
  systems that self-organize.
• Distributed process that autonomously manages
  energy, information, and structure to turn
  resources into useful products according to
  changing needs.
• Reliable computation for the physical world with
  heterogeneous systems robots, inert construction
  materials w/embedded comms and computation.

3
Expected Transformative Benefits

• Novel approach to robust construction from
  elementary components with resource variability
• Distributed algorithms, control theory and
  statistical physics for modeling system behavior
• Analysis and synthesis by analyzing information
  flow
• Experimental demonstrations for robotic
  construction and swarming

4
Hod Lipson, Cornell
Mark Yim, U. Pennsylvania
Eric Klavins, U. Washington
Daniel Rus, MIT
control of robotic self-assembly
modular robotics
reconfigurable structures
self-replication
distributed algorithms
statistical dynamics
mechanism design
5
Project Management
Institution-specific and collaborations for
theoretical advances Shared testbed, yearly goals
for joint experiments Graduate student
rotations Monthly progress teleconfereces,
meetings every 6 months