Support: NSF grants IIS-0093581 and CCR-0330342

• Each source that appears in the enviroment can only be detected locally.
• The agents that detect a source have a mechanism to evaluate how many robots are needed to deal with the event in a promptly manner.  They can agree on the total number of robots by using the aggregation algorithm.
• Agents detecting the same source can communicate among them.
• Under this algorithm, coverage can be sacrificed for the sake of  a quickly completion for the sensing task

We define a utility function that describes the percentage of agents that are doing sensing (with respect to the number of agents needed to complete that task).

Additionally, we define an invitation algorithm as a function of such utility function.

We show that, for the invitation algorithm, the utility function behaves as a Lyapunov function.  This ensures the stability of the algorithm.

We then show that the network evolves towards a formation on which the sensing task is addressed as best as possible. 

The situation becomes interesting when multiple sources appear and there are not enough agents to ensure the promptly completion of all the sensing tasks.

The invited agents thus decide which task to attend, based on the parameters on which they are being invited with. 

As expected, different parameters lead to different performances of the system.  Some of them being optimal for the family of parameters which define them.

An invited agent, by locally evaluating a min-max rule among the invitation it receives, can chose one such that it is locally optimal such that when all the other invited agents behave in the same way, it becomes a globally optimal solution.

Note: The simulations on this project have been based on the Mathematica code provided by the authors of  Coverage Control for Mobile Sensing Networks. IEEE Transactions on Robotics and Automation, 20(2):243--255, 2004.

Back                        Home