Building on previous surveys that often focus on specific components of robot localization, this work offers an integrated review spanning algorithmic estimators, mapping paradigms, and sensor modalities. The survey proposes a functional taxonomy that links probabilistic inference, map representations, and sensor-derived spatial information within a unified comparative framework. By examining classical approaches alongside emerging trends such as semantic mapping and learning-augmented estimators, the paper outlines performance trade-offs, open challenges, and potential directions for future research. Structured tables provide a comparative overview of each class of methods across accuracy, robustness, scalability, and computational cost.
Alipourvarmazabadi et al. (Mon,) studied this question.