Enhancements of PECOS embedded real-time component model for autonomous mobile robot application

Abstract

Recently, Component-Based Software Engineering (CBSE) has becoming a popular approach for developing embedded software. In CBSE, a component model is required to specify the standards and conventions imposed on developers of components. Industrial component models such as CORBA, COM and JavaBeans are generally not suitable for embedded real-time (ERT) system, Consequently, a number of component models suitable for CBSE of ERT software, such an PBO, Koala, PECOS and ReFlex are introduced. Assessments of the PECOS component model were conducted to evaluate the suitability of PECOS component model for adoption in CBSE of autonomous mobile-robot (AMR) software. The assessments emphasize on three requirements facilitates predictable real-time performance, support for resource constraint systems, and support platform-independent implementation. Three enhancements were proposed for the PECOS component model. These enhancements were implemented on a real two-wheeled mobile robot, and results show that, the PECOS component model together with the proposed modifications can generate application suitable for resource constrained AMR systems, the new mapping of component behavior to tasks process can be used to guarantee the run-time predictability and performance, and the new implementation framework proposed enable platform independent development of AMR software.