LIDIA: Localizing in the Dark with Illumination-Awareness Toward Perception-Aware Planning
Iason Georgios Velentzas, Kento Tomita
AI summary
Problem
Visual localization performance degrades under varying or low illumination, yet existing perception-aware planning methods ignore photometric effects by relying solely on geometric visibility.
Approach
LIDIA efficiently models localization quality by jointly evaluating geometric visibility and direct illumination from estimated light sources, enabling accurate prediction of informative camera poses without real-time rendering.
Key results
- Unifies geometric and photometric reasoning for efficient localization quality prediction
- Accurately predicts information gain of candidate poses under varying illumination
- Generates planning trajectories that achieve higher localization accuracy than geometric baselines
- Validates robust performance in synthetic urban and spacecraft scenarios with dynamic lighting
Why it matters
Enables robust visual localization and autonomous navigation in GPS-denied, illumination-constrained environments like space exploration and subterranean mapping.
Abstract
Accurate Localization is a fundamental challenge in robotic autonomy, with applications ranging from au- tonomous driving to space proximity operations. Visual Local- ization is a viable choice in GPS-denied environments, such as subterranean, indoor, urban, or space environments; however, its performance degrades under often encountered conditions, such as low light or varying illumination. This paper introduces LIDIA — an illumination-aware model of localization quality for Perception-Aware Planning. LIDIA involves the efficient integration of light source direction into the planning frame- work, enabling the prediction of visually informative regions in the map under varying lighting. Unlike prior geometric approaches, LIDIA jointly exploits geometric and photometric information without requiring computationally expensive real- time rendering, thereby preserving online applicability. Our re- sults demonstrate that LIDIA consistently outperforms existing geometric methods such as FIF in predicting the information gain of candidate camera poses and in planning trajectories that achieve higher localization accuracy. To the best of our knowledge, this is the first approach to unify geometric and photometric reasoning in an efficient, active localization system, paving the way for robust autonomy in illumination-constrained environments.