PARC Modular Robotics: Proteo

PARC - Chain - Lattice - Pubs -

Lattice Systems: Proteo • Telecube • Digital Clay

A Self-assembling Lattice Reconfiguration Robot

Java Simulator

RD Motion Planning

Once you have lots of modules that can move, attach and detach, how do you get them to globally do what you want them to do? The initial problem is getting the modules to form global shapes.

Issues in programming:

• Search space for path planning of n moving modules is exponential in n
• Some robots will not work properly
• Computational resources are distributed. Each module has limited computation and communication ability

Sequence of 441 modules moving from flat square into a teacup shape.

There are many possible motion planning approaches. One of the questions is how many move, and where?

One RD moving at a time to goal

All moving each to his assigned goal

All move to one goal at a time

All move towards closest unfilled goals

Each of these methods have been tried with various results.
Here are some snap shots of a planner that uses the 4th method planning the motion of 441 modules from a square plate shape to a tea cup shape. When a module reaches a goal position, it is shown as solid white, otherwise it is shown as wireframe magenta.

Proteo Java Simulator

A simulation environment is presented here to experiment with planning. (paper on planning)