並列モジュール記述モデルに基づく自律ロボットの情報システム開発環境

抄録

In this paper, we propose an environment for developing autonomous robot brains in which the developer can describe a brain based on an abstract model in a generic programming language, and verify it immediately on a multi-processor system. We define BeNet, an abstract module network model, for describing a robot brain completely and claim that it is an appropriate model for designing brains and simulating them on multi-processor systems. Since a BeNet is a static network of asynchronous modules that interact each other at regular intervals, it represents a dynamical system completely and concisely and can be simulated efficiently by various multi-processor systems. You can describe a brain frexibly as a BeNet out of desirable modules. We describe our model and show how to implement our environments for developing BeNets for robot brains to show that they are simulated efficiently by multi-processor systems. Our environments make it possible to describe brains in C or C++ in a simplified manner, verify the behaviors of them in the real world and modify them in a short term to develop desirable systems.