The CityLearn Challenge: Four Years of Advancing Common Task Frameworks for Energy Management in Smart Buildings
Year: 2025
Type: conference-paper
Abstract: We show the development and outcomes of the CityLearn Challenge series (2020-2023), a competition focused on advanced control algorithms for building energy management.We examine how the challenges evolved in complexity and realism over four consecutive years, from testing control policy transferability across climate zones to incorporating real-world datasets, battery management, and grid resilience scenarios.The series has consistently attracted increasing numbers of participants and demonstrated the effectiveness of hybrid approaches combining forecasting, optimization, and learning techniques.The standardized benchmarking environments of the CityLearn Challenge series offer valuable insights for the organization of similar machine learning competitions and highlights the importance of common task frameworks in advancing energy-efficient building technologies.
Authors: Zoltan Nagy, Kingsley Nweye
Institution: The University of Texas at Austin
Language: English
Subfield: Building and Construction
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4411345750