Mehdi Heydari Shahna, a researcher at the Division of Robotics, Perception and Learning at KTH Royal Institute of Technology in Stockholm, Sweden, has developed a control framework aimed at advancing the electrification and autonomy of heavy-duty mobile machines (HDMMs). These machines, traditionally powered by diesel-hydraulic systems, are facing a dual transition: towards cleaner electric systems and greater autonomy, driven by climate goals and technological advancements.
The research addresses the challenges of transitioning HDMMs from diesel-hydraulic to electric systems, which include significant technical and economic hurdles. While advanced artificial intelligence (AI) could facilitate higher levels of autonomy, its adoption in HDMMs has been limited due to stringent safety requirements. Heydari Shahna’s work focuses on developing a control framework that simplifies control design for electrified HDMMs and integrates AI while ensuring safety and stability.
The framework consists of three main lines of investigation. First, it proposes a generic robust control strategy for multi-body HDMMs that ensures strong stability across different types of actuation and energy sources. Second, it offers control solutions that maintain strict performance under uncertainty and faults, balancing robustness and responsiveness. Third, it develops methods to interpret and trust black-box learning strategies, enabling their stable integration and verification against international safety standards.
The research is validated through three case studies involving different actuators and conditions, including heavy-duty mobile robots and robotic manipulators. The findings, published in five peer-reviewed publications and one unpublished manuscript, contribute to the fields of nonlinear control and robotics, supporting both the electrification and autonomy transitions in the heavy-duty machinery sector.
For the energy industry, this research offers practical applications in the design and control of electrified heavy-duty machinery. The modular approach simplifies control design, making it easier to integrate new energy sources and technologies. The guaranteed performance and stability of the control policies ensure safe and reliable operation, which is crucial for the adoption of autonomous systems in the energy sector. Additionally, the methods for interpreting and trusting AI strategies can facilitate the integration of advanced AI technologies in energy systems, enhancing their efficiency and autonomy.
This article is based on research available at arXiv.