Improving the development process for teleo-reactive programming through advanced composition

Abstract

Teleo-Reactive programming as applied to autonomic systems is both a viable and exciting prospect as it inherently recovers from errors and unexpected events whilst proceeding towards its goal. It does this without the need to know in advance, the specific events which might occur, thus greatly reducing the human maintenance element. Whilst the benefits of this technique are great once the program has been composed, the challenge of validation attached to these programs also increases. We aim to ease the composition process needed when constructing T-R programs by building new abilities in to our existing Teleo-Reactive framework. Firstly, these new abilities will make it possible to detect validation issues at design time, thus reducing the likelihood of problems and increasing the ease of detecting them. Secondly, it will be possible for programs to be automatically composed from existing elements, thus further reducing the cost of validation issues. These ideas can even be extended to allow for dynamic composition and runtime changes to goals and actions.