This will serve as an area to publish the work I do on the MARS (Mobile Autonomous Robot Software)-2020 project. The MARS-2020 project is a DARPA-sponsored robotics research project concerned with robot teams of Unmanned Air Vehicles (UAVs) and Unmanned Ground Vehicles (UGVs). These teams will rely on communication-sensitive: mission specification, planning, and behaviors, to cooperate and complete their task.

The communication-sensitive behaviors and plans will have to accomodate for all the different types of communication scenarios. One of my tasks will be to incorporate communication models of the mobile adhoc networks (MANETS) into MissonLab for experimentation and testing. The models will decide whether one robot can communicate with another robot (and the quality of that communcation) either directly or indirectly via a multi-hop message. The models will take into account things like the distance between the team members, surrounding foliage and buildings, and delay and congestion due to network topology and performance. From the proposal, "These communication models will be used to test and stress the network of UAVs and UGVs as it moves through the environment, forcing the underlying behavioral controllers, and hence individual agent performance, to conform to the global needs of the mobile network."

4.22.2003 kjo A more realistic communication model is now in MissionLab, it models buildings and foliage. An interesting visualization can be found here. 02.23.2003 kjo The official MRL web page for MARS-2020. 11.11.2002 kjo Some new results showing how the team is effected by the communication preserving behavior. 11.04.2002 kjo We have some preliminary results concering the communication preserving behavior. 10.21.2002 kjo Here is a screen shot of the newest communication model. It takes walls into consideration when deciding whether two robots can talk. I will post a screen shot of Patrick's communication preserving behaviors in action soon. 09.22.2002 kjo Begining to integrate basic comm. models into MissionLab. This is the simplest (level 0) pathloss model, the plane-Earth model for an obstruction free path to determine whether two robots can speak either directly or transitively via a mulit-hop message. Here is a screen shot, the lines between robots mean direct connectivity. Some early graphs showing connectivity between five robots moving through 2-D space. There is a graph for each robot pair, showing how many hops a message takes to talk to each other over time. 09.06.2002 kjo Swiki created. 08.30.2002 kjo Web page created!

• The Mobile Robot Laboratory in the College of Computing at the Georgia Institute of Technology
  
  
  • Dr. Ronald C. Arkin
  • Dr. Tucker R. Balch
  • Dr. Robert R. Burridge
  • Alan Wagner
  • Patrick Ulam
  • Keith J. O'Hara
  • Matt Powers

• The GRASP Laboratory at the University of Pennsylvania

• The Robotics Research Lab at the University of Southern California

• BBN Technologies

• So far, nothing for this particular project.

• The Mobile Robot Lab's publications.

• MissionLab -- MissionLab takes high-level military-style plans and executes them with teams of real or simulated robotic vehicles. MissionLab supports execution of multiple robots both in simulation and actual robotics platforms.

• The official MRL web page for MARS-2020.
• Georgia Tech's previous MARS project.
• The Mobile Robot Laboratory Swiki -- You can find relevant up-to-date documents, web links, and other resources here.