The primary mission of our team is to uncover the basic principles that will best govern a group of robots trying to do useful work in difficult and hazardous environments. The foundation of our work begins with the idea that robot existence must be modeled probabilistically. Robots, like humans, are subject to physical laws and can be damaged or destroyed by both random and intentional events. In the extreme environments posed by space exploration, military operations, firefighting, and nuclear cleanup, the likelihood that robots will be injured is amplified. In many situations, the danger posed is so great that a single robot expected to perform adequately in these scenarios must be designed to mitigate every conceivable circumstance. Clearly, this task is either very difficult or impossible for most operations.
A promising approach is to use tightly coordinated groups, or colonies, of smaller, simpler robots to perform tasks in these dangerous locales. The fundamental advantage of this approach is redundancy. If managed properly, the loss of a robot, although painful, will not be catastrophic and task execution capabilities will degrade gracefully across multiple robot failures. While very promising, the implementation of these ideas into a working system is fraught with the difficulties inherent in any highly redundant system. Specifically, our team is focusing on the following major problem areas:
Behavioral Strategies for Colonization: What is the "critical mass" of agents necessary to form an effective colony? Once formed, what kind of internal, environmental, and task-level cues can be monitored to maintain the optimal colony through pruning and growth behaviors? Are there times when it is more efficient to disband the colony and work at the level of the individual robot?
Group Learning Algorithms: Can the sharing of experiences amplify the effectiveness and minimize the convergence time of the learning processes within a robot colony? What is the proper mix of individual and group learning strategies? Can effective learned behaviors be distributed as a commodity within the colony?
Protected and Secure Resources: What is the best way to distribute data and processes within the colony to insure that loss of robots will minimize loss of learned knowledge and data? Once losses are realized, what is the best way to redistribute existing information to make it more secure? What are the most efficient and secure methods of routing critical information and data within the colony and to the outside world?
Task-Level Reprogramming: Can we obtain an efficient system for doing real work in hazardous locales using behavioral seeds from individual robots to grow a task-specific organization? If so, what is the optimal matrix of basic behaviors and learning algorithm required to foster this system?
Scalability: Robot colonies that rise as a dynamic response to general task-level requests will need to have great variability in their computational, power consumption, communications, and physical capabilities. Thus, our architecture must scale to problems at both ends of the spectrum. Can we make a colonization architecture that supports ten robots in one task and ten thousand for another?
Although the focus of our work is fundamental, we believe the ultimate measure of success of any robotic system should be evaluated in terms of doing useful work out in the world. For this reason, we have chosen to apply our work to the task of Distributed Mapping of Urban Environments. The unique feature of our distributed mapping system, and the eventual metric of our success, will be its ability to doggedly pursue this task when faced with multiple robot failures. Our initial demonstration, tentatively scheduled for the Fall of 2001, will be to deploy ten small robots into a "mock-up" of an urban facility. These robots will form a colony whose sole purpose is the generation of a map of this area. After an initial period during which basic distributed mapping operation is demonstrated, our sponsors will be asked to "disable" robots of their choice and observe the reaction of the colony to this loss. This process will continue until critical mass is lost and the colony is unable to function in terms of its primary mission. Thus, observers will be given an "on-line" demonstration of how our system adapts to multiple and catastrophic failures.
Top of Page