eaxattic / phd_old Goto Github PK

This is README file for PhD work by Ekrem Aksoy
(c) 2010
Decentralized Data Centric Middleware for Heterogeneous Environments (Robots/Agents)
Heterogeneous Robotic Middleware in short :) or HRM

HRM is a middleware for simplifying multi robot/agent application development. Also, the environment is modeled as totally decentralized and data centric.
Therefore there are very large variety of application addressed. Moreover, with its Policy based communication reliability mechanisms, it is assured that if a Task T
can be accomplished by any robot in the Active Environmet (i.e. LOS of Primary Coordinator), it is done, of course Policies are met.

There are three roles in environment: 

1) Primary Coordinator
2) Shadow Coordinator
3) Task Member

PC, SC and all TMs form a "Task Force". The Task Force is assigned a Task Board. The TMs communicate thru this board during Task Progress.

In HRM Environment these are the assumptions:

1) Any Agent could be PC for a Task, and PC is the agent that the Task is defined from. (e.g. an agent who acquires the Task)
2) Any Agent could be SC in order to keep Task Repositories alive.
3) There is no central, dedicated or assigned role for either PC or SC or TM
4) Comm is totally decentralized, there is no single point of failure.
5) LOS Comm can be lost in anytime. BLOS comm is accomplished by proxies under this assumption
6) An Agent could be in any or all of the roles at the sametime. e.g. an agent could be PC in Task Force#1 where SC in Task Force#2 and TM in Task Force#3
7) An Agent could be assigned only in one Task Force as a TM.
8) Task Forces identified by PCs. So there are several Task Forces identified by a PC since a Task could be defined from a PC and each Task could require different capabilities.