Control systems underpin a vast array of modern technologies, from autonomous vehicles to aircraft guidance systems. In safety-critical domains, understanding why a system acts as it does is as important as achieving high performance. However, many control policies, especially those derived from modern machine learning techniques, remain opaque, prioritizing performance over interpretability. Compounding this issue is the lack of standardized methods to measure interpretability in control contexts.

The Interpretable Control Competition, now in its third edition, aims to advance research at the intersection of control theory, machine learning, and interpretability. The goal is to foster the development of control policies that are both effective and understandable, encouraging new approaches to designing, explaining, and evaluating intelligent controllers.

Building on previous editions, which explored both continuous and discrete control settings (from robotic locomotion to game-based environments), this year’s competition will take a step toward more realistic, safety-relevant scenarios. In collaboration with Airbus, a global leader in aerospace innovation and manufacturing, we will focus on a challenge related to aeronautical decision-making, focusing on relevant concrete scenarios that exemplify the complexity and reliability demands of real-world systems.

Participants will be welcome to enter the competition using their preferred methods to develop and interpret control policies for addressing the proposed task. We particularly encourage the incorporation of evolutionary computation techniques to enhance either policy generation or interpretability.

Submissions will be evaluated based on both performance and interpretability. Performance will be assessed through simulations of each submitted policy, while interpretability will be evaluated by a panel of judges, including domain experts from the industry.