‘ˆξ“c‘εŠw—HŠwp‰@@ŠξŠ²—HŠw•”EŒ€‹†‰Θ@ξ•ρ—HŠw‰ΘEκU@›Œ΄Œ€‹†ŽΊ@Sugawara Lab. Waseda Univ.

The Intelligent Software Laboratory

Professor Toshiharu Sugawara

The Intelligent Software Laboratory (ISL) explores the theory, development, and analysis of autonomous agents and multi-agent systems (MAS) and their applications in various domains, including Internet monitoring management, Web applications, interface agents, and multi-agent simulations.

Research on multi-agent systems is one of the active research fields in artificial intelligence (AI). Autonomous agents are intelligent programs that are capable of perceiving local environments and taking appropriate action to move toward their goals based on independent decision-making. An MAS consists of loosely-coupled autonomous, rational agents capable of coordinated/collaborative actions to achieve individual or joint goals in cooperation with other agents. Research activities include a broad range of efforts to develop and apply this type of coordinated actions.

Our research topics include:

  • Coordination and Cooperation in Multi-Agent Systems
  • Multi-Agent Planning
  • Learning for Coordinated Activities
  • Negotiation in Massively Multi-agent Systems
  • Effects of organizational and inter-agent network structures
  • Internet Monitoring and Management by Multiple agents
  • Swarm Intelligence
  • Multi-Agent Simulations
  • Coexistence of agents (including people and computer programs)
  • Web Agent Applications

In particular, we have addressed the issues related with large-scale MAS: Although recent advances in Internet services, sensor networks, pervasive computing, and grid computing exhibit the need for MAS, they further require more sophisticated MAS technologies for large-scale and busy environments. For example, e-commerce transactions, which frequently appear in the current Internet era, consist of coordinated tasks including interactions among a variety of agents in charge of customer authentication and management, stock management, shipping control, and payment processing. These kinds of tasks simultaneously and frequently occur throughout in the world. In sensor-network applications, agents reside in many sensor and computational devices and, in grid computing, many computational entities should concurrently process the assigned subproblems into which a large computation problem is decomposed. In these applications, many of the tasks should be allocated appropriately to maximize the entire abilities of MAS, in order to realize efficientcy and high-quality services.

We also think about the coexistence with computer programs that are deployed everywhere in the Internet, smartphone, and often home appliances. They affect each other at many levels of social activities. Taking optimal actions are usually good for individuals (people and porograms), but may not be good from social viewpoints due to excessive/wasteful competition. We believe that socially harmonious coexistence between people and programs will be stronly required in near future.

If you are interested in studing our lab as a master or doctor cource student, please visit this page.

to TOP

 

Japanese
hover