;Software integration is an important and difficult issue in software engineering. In this study, we were inspired by cell theory and implemented the CellS software framework to address software integration issues. The difficulty in software integration is that the individual modules are executable, but they are not guaranteed to be executable after integration. In CellS, the smallest unit of software is the cell, and when all the cells pass its unit test, the integration of the software is completed. On the other hand, traditional software cannot leverage completely the computing power of the CPU, and the root cause is that the parallelized program accounts for a very low percentage of the total program. The CellS software architecture improves software parallelism, and engineers do not have to write parallelism program because the framework itself is a parallel architecture. This study uses the CellS implementation the dialogue engine and is applied to navigation systems and robotic arm control. The dialogue-based design gives the navigation system and robotic arm control the advantage of usability; the CellS-based design gives the navigation system and robotic arm control the advantage of maintainability. Performance experiments show that our application can increase speedup to 1.94 times, and with the increase in the number of cores of the CPU, the application is significantly speedup. The CellS software framework designed in this study has the ability to continuously evolve, and this framework is especially suitable for software applications that need to autonomously process the information generated by the sensor and respond quickly.
