| Abstract: |
Cooperative Distributed Problem Solving (CDPS) studies how a network of problemsolvers (Intelligent Agents) may work together to solve problems that are beyond theirindividual capabilities. Cooperation between these nodes is necessary because nosingle node has sufficient information to solve a problem. These nodes mustcoordinate their parallel problem solving to build compatible solution, and to avoidredundant work.CDPS has opered a new door in the study of intelligence because it continues to revealthe connections between coordination and other aspects of intelligence. Coordinationresearch brings forth issues in areas such as planning, scheduling, languages, andreasoning. In other words, coordination forces researchers to address many of thebasic problems :,n Artificial Intelligence.Coordination is defined to be the act of managing interdependencies between activities.Coordination behaviors can be divided roughly into specification behaviors (creatingshared goals), planning behaviors (selecting potential sets of tasks to accomplishgoals), and scheduling behaviors (assigning tasks to groups or individuals, creatingschedules, allocating resources, etc.). Coordination behaviors, in turn, rest on moreprimitive agent behaviors such as acqumng knowledge, following rules,communicating information, inferencing , and negotiation.These behaviors are intensively located in Distributed Sensor Networks (DSN). ADSN consists of intelligent nodes scattered across a specific area. DSN problems arecommon in many environments, where a large amount of detailed work needs to bedone, and the results are collected at a single location and integrated, so on up ahierarchy. TWCl examples of such problems are presented. These are DistributedInformation Gathering and Distributed Vehicle Monitoring. The problem is toschedule problem solving activities that are simulated as task hierarchies (usingT AEMS framework).
|
|
|