LENS Lab

Welcome to the Laboratory for Learning Evaluation and Naturalization of Systems (LENS Lab). We focus on operationalizing machine learning (ML) models in autonomous systems, robots, and beyond. Our mission is to develop algorithmic tools that not only evaluate these models but also help naturalize them—making them more compatible with the complex, dynamic requirements of the real world. We believe that evaluating models solely by accuracy on a held-out dataset is insufficient for true operational readiness. That’s why we explore deeper questions: When and how do models fail? How can we explain black-box neural networks? How can we assess decision uncertainty? Are models fair and unbiased? Will a model adapt effectively to new or changing environments? Our work supports engineers in building systems that are robust, adaptable, and trustworthy in practice, while also aiding policymakers in setting legal and ethical standards.