This is the pre-proceedings for the RLC 2026. You may expect minor changes.
Reinforcement Learning Journal, vol. 7, 2026, pp. TBD.
Will be presented at the Reinforcement Learning Conference (RLC), Montréal, Quebec, Canada, August 15–17, 2026.
Keywords: Applied RL, Constrained RL, Shielding, Microgrid optimization, Energy dispatch
Reinforcement learning (RL) is a powerful framework for optimizing decision-making in complex systems under uncertainty, a key challenge in real-world settings such as the energy transition. A representative example is remote microgrids that supply power to communities disconnected from the main grid. Enabling the energy transition in such systems requires coordinated control of wind turbines, fuel generators, and batteries to meet demand while minimizing fuel consumption and battery degradation under uncertain load and wind conditions. These systems must comply with extensive regulations and operational constraints, requiring interpretable guarantees when deploying RL agents. In this paper, we introduce Shielded Controller Units (SCUs), a systematic and interpretable approach that leverages prior knowledge of system dynamics to ensure constraint satisfaction. Our shield synthesis methodology decomposes the environment into a hierarchy where each SCU explicitly manages a subset of constraints. We demonstrate the effectiveness of SCUs on a remote microgrid optimization task with strict operational requirements. The resulting RL agent achieves a 24\% reduction in fuel consumption without increasing battery degradation, outperforming current industry heuristics and standard constrained RL baselines while satisfying all constraints. We hope SCUs facilitate the safe deployment of RL agents. All code and supporting data will be open-sourced upon acceptance.
Hadi Nekoei, Alexandre Blondin Massé, Rachid Hassani, Sarath Chandar, and Vincent Mai. "Shielded Controller Units for RL with Operational Constraints Applied to Remote Microgrids." Reinforcement Learning Journal, vol. 7, 2026, pp. TBD.
BibTeX:@article{nekoei2026shielded,
title={Shielded Controller Units for RL with Operational Constraints Applied to Remote Microgrids},
author={Hadi Nekoei and Alexandre Blondin Massé and Rachid Hassani and Sarath Chandar and Vincent Mai},
journal={Reinforcement Learning Journal},
volume={7},
pages={},
year={2026}
}