In the realm of energy and technology, a significant piece of research has emerged from the work of Diaeddin Rimawi, a researcher affiliated with the University of California, Berkeley. This study delves into the intricate world of cyber-physical systems (CPS), particularly focusing on Online Collaborative AI Systems (OL-CAIS), which blend computational and physical components to collaborate with humans in achieving common goals.
The research addresses a critical challenge in OL-CAIS: balancing resilience and energy efficiency. These systems are vulnerable to disruptive events that can degrade their performance. Decision-makers face a trade-off between restoring performance quickly and minimizing energy impact. Rimawi’s work aims to strike a balance between these two properties by developing models and policies that optimize the greenness-resilience trade-off.
Rimawi models the behavior of OL-CAIS through three operational states: steady, disruptive, and final. To support recovery during disruptions, the researcher introduces the GResilience framework. This framework offers recovery strategies through multi-objective optimization, game-theoretic decision-making, and reinforcement learning. Additionally, a measurement framework is designed to quantify resilience and greenness, providing a clear metric for evaluating the system’s performance.
Empirical evaluations using real and simulated experiments with a collaborative robot learning object classification from human demonstrations show promising results. The resilience model effectively captures performance transitions during disruptions. GResilience policies improve green recovery by shortening recovery time, stabilizing performance, and reducing human dependency. Reinforcement learning-based policies achieve the strongest results, although with a marginal increase in CO2 emissions. The research also highlights the phenomenon of catastrophic forgetting after repeated disruptions, which the proposed policies help mitigate. Notably, containerized execution is found to cut CO2 emissions by half, underscoring the importance of efficient execution environments.
This research, published in Rimawi’s doctoral dissertation, provides valuable models, metrics, and policies that ensure the green recovery of OL-CAIS. For the energy sector, the insights gained from this study can be applied to improve the resilience and energy efficiency of various cyber-physical systems, from smart grids to renewable energy management systems. By optimizing the greenness-resilience trade-off, energy companies can enhance their operational efficiency and reduce their environmental impact, contributing to a more sustainable energy future.
This article is based on research available at arXiv.