The second Push Innovation Voucher awarded in the canton of Fribourg marks a significant step toward smarter, more responsive industrial automation. This time, it supports a project led by Asyril, a company specializing in high-performance vibratory feeding systems, in collaboration with the HEIA-FR and its research institutes iCoSys and SeSi.
The project, named FiLeML (Fill Level detection using Machine Learning), builds directly on the results of a previous collaborative initiative, ModIA. That initial project developed a hybrid methodology combining physical simulation and machine learning to create digital twins for predictive maintenance. FiLeML now takes the concept a step further, with a concrete goal: enabling Asyril’s Asyfill hopper system to autonomously detect the fill level of its parts tray—in real time, with high accuracy.

From promising method to deployable solution
In ModIA, promising algorithms and models were built, but more work was needed to turn them into deployable features. The 25,000 CHF Push Innovation Voucher, funded by Fribourg’s Nouvelle Politique Régionale (NPR), is now helping bridge that gap. The project kicked off in early 2025 and will run through September.
FiLeML will integrate a MEMS sensor directly into the Asyfill hopper and leverage hybrid machine learning models to process the vibratory signals. By combining simulation data and real-world signals, the system aims to estimate the fill level for various part types and hopper sizes with acceptable uncertainty. The operator can be notified with sufficient lead time—early enough to refill the tray without causing production downtime, but not so early that it disrupts workflow.
Beyond technical performance, usability is a core focus. The fill-level detection must run in under 500 ms, and the system should be easy to calibrate in under 3 minutes, even by non-experts. This user-centric approach reflects Asyril’s strategy: combining high-tech innovation with plug-and-play simplicity.

A win for customers—and for Swiss innovation
For Asyril’s customers, the benefits are immediate: reduced production downtime, better process monitoring, and early warnings when parts are running low. For Asyril, the added functionality enhances its value proposition in a competitive market—offering more intelligence, without adding complexity.
The FiLeML project helps companies take a promising innovation, refine it with expert support, and bring it one step closer to real-world use. By backing these final stages of development, the NPR helps Swiss SMEs stay at the forefront of innovation—not just in ideas, but in products ready to perform.