The DnSPro project set out to create a foundation for smart Industry 4.0 production lines by replacing the traditional hierarchical automation pyramid with small, embedded, modular, and decentralised subsystems. The chosen application was a bottle‑filling plant, a representative process that demands tight control of volume, pressure, temperature and fluid dynamics while remaining highly adaptable to changes in bottle geometry and filling medium. The project’s goal was to design, evaluate and prototype a cyber‑physical production system (CPPS) that integrates sensor‑actor communication, adaptive control, and secure data handling within a cost‑effective hardware platform.

A key technical outcome was the development of a modular demonstrator concept that combines an Infineon XMC4800 microcontroller with a Xilinx Zynq‑7010 system‑on‑chip (SoC). This multi‑board architecture separates peripheral handling from programmable logic, allowing the full hardware/ software co‑design to be implemented and later reconfigured to match the computational demands of the control algorithms. The demonstrator was built on a Zedboard evaluation platform, enabling rapid integration of existing algorithms as hardware accelerators written in high‑level synthesis languages. During the concept phase the team evaluated several field‑bus systems and newer communication stacks, performing cycle‑time calculations to ensure that the chosen protocols could meet the real‑time requirements of the filling process. The evaluation was supported by a free license from the company Emta, which provided a CANopen test environment.

Three distinct control concepts for the proportional valve were formulated, each differing in expected control quality and computational complexity. This scalability allows the control algorithm to be tuned to the actual benefit derived from the system, ensuring that the hardware can be adapted after evaluation. The control concepts target a fast and accurate filling operation, with the objective of reaching a set‑point or following a trajectory without overshoot, even in the presence of foaming fluids and varying bottle shapes. Sensor data fusion was addressed through a dedicated subsystem that aggregates pressure, temperature, and visual data from a smart camera, forwarding only processed results to the higher‑level controller. Trajectory planning was explored using both classical methods and reinforcement learning; the learning component is trained offline, while online prediction is performed in real time. Security features were also incorporated, following the guidelines of the project’s security work package.

Performance metrics reported include cycle times for process‑flow tasks ranging from 45 ms to 400 ms, and the communication stack evaluation demonstrated that the selected protocols could sustain the required data rates for both horizontal and vertical communication layers. The modular hardware design allows for future scaling, and the evaluation of the communication stacks ensures that the system can be integrated into larger production networks without compromising real‑time performance.

The project was carried out by the Chair for Embedded Systems of Information Technology at Ruhr‑University Bochum, in close collaboration with industrial partners and the company Emta. The team performed requirement analysis, a failure‑mode and effects analysis, and drafted the specification for the DnSPro system. The project ran until 31 October 2018 and was funded by the German Federal Ministry of Education and Research (BMBF) under grant number 16ES0392. The collaborative effort produced several conference papers, including presentations at NASA/ESA AHS 2017, DASIP 2017, and CASE 2018, and contributed to the doctoral thesis of Florian Kästner, who is expected to defend his dissertation in the first quarter of 2020. The knowledge gained in novel algorithms, communication mechanisms, low‑power hardware, and operating systems positions the team to pursue further research and apply for future funding from BMBF, the EU, and the German Research Foundation.