Embedded learning and optimization for the next generation of smart industrial control systems

Project Information

ELO-X

Grant agreement ID: 953348

DOI

10.3030/953348

Project closed

EC signature date 13 August 2020

Start date 1 January 2021

End date 30 June 2025

Funded under

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Total cost

€ 3 884 353,92

EU contribution

€ 3 884 353,92

Coordinated by

ALBERT-LUDWIGS-UNIVERSITAET FREIBURG
Germany

Periodic Reporting for period 2 - ELO-X (Embedded learning and optimization for the next generation of smart industrial control systems)

Reporting period: 2023-01-01 to 2025-06-30

Summary of the context and overall objectives of the project

Digital technologies are transforming all sectors of our economy and will increasingly do so in the years to come. Thanks to the increasing capabilities of digital technologies, the next generation of smart industrial control systems (SICS) are expected to learn from streams of data and to take optimal decisions in real-time on the process at hand, leading to increased performance, safety, energy efficiency, and ultimately value creation.

However, to develop this new control systems bottlenecks have to be overcome, the two most critical being: (1) the fact that the computation availability for industrial control systems is locally embedded in each system or subsystem, possibly with limited communication capabilities and distributed topologies; (2) the fact that industrial applications require reliable algorithms, with interpretable and verifiable behavior. Both these bottlenecks are related to safety aspects, which are crucial in applications where a single computation error can cause high economic and environmental cost or even damage to people.

Numerical optimization is at the very core of both learning and decision-making, since both the extraction of information from data and the choice of the most suitable action are naturally cast as optimization problems and solved numerically. Therefore, the overall objectives of the project were to develop embedded learning- and optimization-based control methodologies for SICS, while training highly qualified and competent researchers.

Over the course of the project, ELO-X has developed novel methods for model predictive control, learning-based control with robustness guarantees, and optimization algorithms for embedded and mixed-integer systems. These were implemented and demonstrated in real-world applications such as autonomous vehicles, hydraulic systems, robotic manipulators, and temperature control units. The project produced over 100 scientific publications, several high-TRL software tools, and six open-source packages. As of June 2025, five PhD theses have been successfully defended, with others nearing completion, and several ESRs have already transitioned into research positions across Europe.

Visit https://elo-x.eu/ for more information on the ELO-X network.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The project began in January 2021 with a focus on the setup of the consortium and in particular with recruitment of the early stage researchers (ESRs), which was finalized in spring 2022. All 15 ESR candidates were selected before autumn 2021, and an online kick-off event was organized on November 17, 2021 to present the whole Consortium and the ESRs to each other. In the first months, the work of all ESRs was devoted to literature research and studying to reach the state of the art of their respective domains.

Each ESR began working on specific research topics aimed at novel solutions, with early results submitted to journals and international conferences. ESRs engaged in laboratory work, discussion groups, and secondments, including with industrial partners. The consortium also introduced a group of eight associated PhD fellows, fully integrated into the network.

Over the course of the project, the network organized four seasonal schools, four workshops, and two Advisory Board meetings, which brought together ESRs, supervisors, and external experts to support training, collaboration, and dissemination.

Scientific progress included the development of new algorithms in optimization-based control, robust learning, symbolic feedback stabilization, mixed-integer optimization, and physics-informed modeling. Applications ranged from autonomous vehicles and hydraulic systems to robotic manipulators and temperature control units. Several methods were demonstrated on real systems and released as open-source tools.

The project produced over 100 scientific publications, six open-source software packages, and multiple high-TRL demonstrators. Results were disseminated through international conferences, a public website, professional videos, and social media. Exploitation activities included industry-oriented secondments, software transfer, and integration into experimental platforms.

As of June 2025, five ESRs have defended their theses, and the remainder are on track for defense by mid-2026. The project has successfully met its objectives in research, training, and real-world impact.

Visit https://elo-x.eu/ for more information about the ELO-X fellows' results, news of their progress, and information about the events organized by the network.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

ELO-X has made significant contributions beyond the state of the art in learning-based control and embedded optimization. ESRs developed new algorithms for nonlinear model predictive control, mixed-integer optimization, symbolic feedback stabilization, and robust control using Gaussian processes and neural network models. Several of these methods were tested on real systems, including autonomous vehicles, hydraulic actuators, and temperature control units, demonstrating both technical feasibility and industrial relevance.

The project produced over 100 scientific publications and six open-source tools, accelerating research and enabling uptake in both academia and industry.

These proof-of-concept implementations are expected to have a significant impact on industrial practice, by setting the stage for the deployment of the next generation of smart industrial control systems. These systems are expected to improve performance, safety and efficiency of industrial processes, prolong the life of industrial machinery, and to save energy and natural resources.

elo-x-header.jpg

elo-x-consortium-leuven-2mb.jpg