Warning: this page is currently under construction!
 
Overview
COCoA is a framework to facilitate multirobot coordination. COCoA is specifically designed for teamwork in time critical domains requiring tightly coupled coordination. Robots respond to uncertainty in the environment by interleaving planning and execution. COCoA enables the robots to generate a team plan by modeling the problem as a MILP which can be solved using an anytime algorithm that combines standard MILP solution techniques with domain specific heuristics.
The declarative nature of the MILP framework makes it easy for COCoA to accomodate a wide range of additional system constraints on the robots, goals, and resources of the system. For example, COCoA can handle constraints such as: unless a room is checked for chemicals before looking for victims in that room, robot 1 cannot participate in the search since it cannot be decontaminated.
Benchmarks
Please contact (mberna AT andrew DOT cmu DOT edu) if you would like
access to the current set of benchmarks or if you have your own
benchmarks to donate to the repository.
COCoA and an abstract simulator, COCoASim have been implemented in
Java. I have not yet made any official releases of the source code
but if you are interested, please email me (mberna AT andrew DOT cmu DOT edu).
CPLEX is currently required in order for COCoA to be fully functional.
I will investigate open source ILP solvers in the near future.
This work was made possible through the generosity of the ILOG corporation and the NSF Award IIS-0205526.
Code
Relevant Publications
Acknowledgments