Name
Abstract | ||
|---|---|---|
This paper describes a new multi-agent testbed developed to support research in multi-agent systems that in- corporate both autonomous operation and human input. Teams compete in a modified version of the game of capture-the-flag in a maze-like environment of arbitrary complexity. Each team consists of a number of mobile robots, a human commander, and a UAV that provides essential position information. The makeup of each team, the motion constraints, and the nature of the game present a significant challenge to any multi-agent coordination scheme. This paper describes the rules of the game and the hardware and software infrastructure we have developed to play it. The paper further describes some of our work in devising an effective technique for planning paths through the maze for the ground-based robots. I. INTRODUCTION Among researchers in robotics there is increasing interest in using teams of robots to provide solutions in many application domains. To expedite the development of the control and coor- dination strategies required for complex multi-agent systems, researchers have adopted a variety of testbed problems, of which competitive games for teams of mobile robots are well known examples. Competitions using real robots advance the overall state of the art because successful teams are likely to be based on effective coordination techniques that can be applied to other multi-robot applications with real-time constraints. To support our research in adjustable autonomy, coordi- nating autonomous robots with human directives, we hoped to identify an existing competition that met a short list of essential requirements. First and foremost, input from a hu- man manager had to be allowed. Some potentially hazardous applications will always have a human in the command loop to address reliability, safety, and legal concerns. Second, we wanted an application that required the collaborative efforts of heterogeneous teams of robots, with the resulting complexity in managing multi-agent interactions and coordination. Third, the operating environment needed to be both dynamically varying and configurable using static obstacles to model vari- ous urban or indoor settings. Despite our interest in and experience with robot soccer (1), we realized it did not meet our criteria for multiple reasons: This work was funded by DARPA grant NBCH1020013 the rules of the most popular leagues require fully autonomous operation, the playing field does not contain arbitrary obsta- cles, and the robots on each team usually have nearly identical capabilities (2). Existing search and rescue competitions come closer in that they support human operators and provide obstacle-rich environments, but their operational environments have little if any dynamic variation, and solutions do not require different kinds of robots on each team (3). In the absence of a standard competition that met our criteria, we elected to define our own multi-robot game and to create the infrastructure required to play it using our lab facilities. We created a version of capture-the-flag in which two teams play against each other in an easily reconfigurable maze environment that can represent the streets of a city or the inside of a building. The goal of each team is to retrieve the opponent's flag and bring it back to base while preventing the opposing team from doing the same. Each team has a human manager, mobile robots on the ground, and a (simulated) UAV flying over the playing area that collects vital position information for the team. The capabilities of the robots are such that a team can win only if its robots explicitly commu- nicate and collaborate with each other. Moreover, robots can communicate directly with each other, with the UAV, or with the human operator only if both are within a specified distance of each other. Indirect communication through another robot is allowed provided that each link meets the maximum distance constraint. After our project was underway, we were intrigued to discover other research efforts based on capture the flag. Researchers at the California Institute of Technology and Cornell University have created a competition called RoboFlag that is loosely based on a combination of capture the flag and paintball (4), (5). Teams have 6-10 robots and a human operator, and they attempt to bring their opponent's flag back to their Home Zone. Robots in enemy territory can be tagged by shooting them with small balls, a number of which are randomly distributed on the field at the beginning of each game. Although the rules and regulations of RoboFlag con- tinue to evolve, all software required to run the game has been developed and is freely distributed (6). Like our competition, RoboFlag is a very dynamic game and includes humans in the |
Year | DOI | Venue |
|---|---|---|
2004 | 10.1109/ROBOT.2004.1307982 | Robotics and Automation, 2004. Proceedings. ICRA '04. 2004 IEEE International Conference |
Keywords | Field | DocType |
man-machine systems,mobile robots,multi-agent systems,multi-robot systems,path planning,user interfaces,autonomous operation,ground based robots,hardware infrastructure,human assisted capture-the-flag game,human commander,mobile robots,motion constraints,multiagent coordination scheme,multiagent systems,multiagent testbed,path planning,software infrastructure,unmanned air vehicle,urban environment | Motion planning,Urban environment,Control engineering,Multi-agent system,Software,Engineering,Robot,User interface,Mobile robot | Conference |
Volume | ISSN | ISBN |
2 | 1050-4729 | 0-7803-8232-3 |
Citations | PageRank | References |
1 | 0.37 | 6 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Matthew A. Blake | 1 | 1 | 0.37 |
| Gerrit A. Sorensen | 2 | 1 | 0.37 |
| James K. Archibald | 3 | 632 | 161.01 |
| R. W. Beard | 4 | 2522 | 311.64 |