Sensible Agents Testbed

The following objectives guide the implementation of the Sensible Agent architecture, Run-Time Environment and Testbed: (1) parallel development of SA module functionality, (2) rapid integration and prototyping of SA module and system functionality, (3) repeatable experimentation, (4) insightful visual­ization and monitoring of module-level, agent-level and system-level operations, and (5) accessibility by 3^rd parties. These objectives drove the design decisions discussed in this paper.

All components of the Sensible Agents system communicate through OMG^ä CORBA®.  CORBA (Common Object Request Broker Architecture) is an inter-lingual open standard for inter-process communication.  This allows diverse programs, including agents, simulations, user interfaces, and the experimental scripting system to communicate regardless of their computing platform or implementation language.  Another middleware foundation is OMG’s Interface Definition Language (IDL^ä), which specifies across programming languages the services provided by each software module.

The four Sensible Agent modules and the Sensible Agent System Interface(SASI) implement IDL-specified interfaces.  The SASI is a layer of abstraction between the agents and the runtime environment and testbed, providing a common interface to a Sendible Agent despite an agent’s possible internal differences in code.   Sensible Agents modules are written in Java, C++, and LISP, and currently run on Linux™ and Windows™ operating systems.

This text taken from Agents 2001 Infrastructure paper with minor modifications.