Proposal Abstract

We plan to investigate the use of advanced modeling techniques to improve the development of embedded systems. In particular, we propose to develop domain-specific models that describe the layout of a “hostile” environment. This could symbolically represent an area infected with land-mines, or a disaster site where it is too dangerous for humans to search for survivors. From these models, model interpreters will be developed that generate the embedded code that will be downloaded into a LEGO MindStorms robot. The code that is generated will assist the robot in maneuvering around the dangerous obstacles. Thus, high-level models of the terrain are all that are needed to drive the navigation of the robot. Changes to the models can represent different configurations of the hostile environment and produce different code for the robot. When a new environment is encountered, all that is needed are modifications to the models. This saves time in development because low-level code does not need to be written by hand; it is generated from the models (i.e., changes to the models result in representative changes to the generated embedded code to control the robo). The domain knowledge of the specialized techniques for engineering such an embedded system is built into the code generators, rather than residing in the minds of embedded domain engineers. We plan to also investigate the coordination among multiple robots. This could represent the situation where multiple industrial robots are being used to scan a search area (of course, our smaller LEGO robots are but a facsimile of the real problem, but the underlying principles are similar).