Modern cyber-physical systems interact closely with continuous physical processes like kinematic movement. Software component frameworks do not provide an explicit way to represent or reason about these processes. Meanwhile, hybrid program models have been successful in proving critical properties of discrete-continuous systems. These programs deal with diverse aspects of a cyber-physical system such as controller decisions, component communication protocols, and mechanical dynamics, requiring several programs to address the variation. However, currently these aspects are often intertwined in mostly monolithic hybrid programs, which are difficult to understand, change, and organize. These issues can be addressed by component-based engineering, making hybrid modeling more practical. This
paper lays the foundation for using architectural models to
provide component-based benefits to developing hybrid pro-
grams. We build formal architectural abstractions of hy-
brid programs and formulas, enabling analysis of hybrid programs at the component level, reusing parts of hybrid programs, and automatic transformation from views into hybrid programs and formulas. Our approach is evaluated in the context of a robotic collision avoidance case study.