Berman, Spring Melody
Now showing 1 - 2 of 2
PublicationAbstractions, Analysis Techniques, and Synthesis of Scalable Control Strategies for Robot Swarms(2010-05-16) Berman, Spring MelodyTasks that require parallelism, redundancy, and adaptation to dynamic, possibly hazardous environments can potentially be performed very efficiently and robustly by a swarm robotic system. Such a system would consist of hundreds or thousands of anonymous, resource-constrained robots that operate autonomously, with little to no direct human supervision. The massive parallelism of a swarm would allow it to perform effectively in the event of robot failures, and the simplicity of individual robots facilitates a low unit cost. Key challenges in the development of swarm robotic systems include the accurate prediction of swarm behavior and the design of robot controllers that can be proven to produce a desired macroscopic outcome. The controllers should be scalable, meaning that they ensure system operation regardless of the swarm size. This thesis presents a comprehensive approach to modeling a swarm robotic system, analyzing its performance, and synthesizing scalable control policies that cause the populations of different swarm elements to evolve in a specified way that obeys time and efficiency constraints. The control policies are decentralized, computed a priori, implementable on robots with limited sensing and communication capabilities, and have theoretical guarantees on performance. To facilitate this framework of abstraction and top-down controller synthesis, the swarm is designed to emulate a system of chemically reacting molecules. The majority of this work considers well-mixed systems when there are interaction-dependent task transitions, with some modeling and analysis extensions to spatially inhomogeneous systems. The methodology is applied to the design of a swarm task allocation approach that does not rely on inter-robot communication, a reconfigurable manufacturing system, and a cooperative transport strategy for groups of robots. The third application incorporates observations from a novel experimental study of the mechanics of cooperative retrieval in Aphaenogaster cockerelli ants. The correctness of the abstractions and the correspondence of the evolution of the controlled system to the target behavior are validated with computer simulations. The investigated applications form the building blocks for a versatile swarm system with integrated capabilities that have performance guarantees. PublicationOptimized Stochastic Policies for Task Allocation in Swarms of Robots(2009-08-01) Berman, Spring; Halasz, Adam; Kumar, Vijay; Hsieh, M. AniWe present a scalable approach to dynamically allocating a swarm of homogeneous robots to multiple tasks, which are to be performed in parallel, following a desired distribution. We employ a decentralized strategy that requires no communication among robots. It is based on the development of a continuous abstraction of the swarm obtained by modeling population fractions and defining the task allocation problem as the selection of rates of robot ingress and egress to and from each task. These rates are used to determine probabilities that define stochastic control policies for individual robots, which, in turn, produce the desired collective behavior. We address the problem of computing rates to achieve fast redistribution of the swarm subject to constraint(s) on switching between tasks at equilibrium. We present several formulations of this optimization problem that vary in the precedence constraints between tasks and in their dependence on the initial robot distribution. We use each formulation to optimize the rates for a scenario with four tasks and compare the resulting control policies using a simulation in which 250 robots redistribute themselves among four buildings to survey the perimeters.