This chapter will discuss advantages and disadvantages of different approaches for indoor localization and contrast them to the proposed method. While there is a wide range of methods for indoor localization—from laser range scanners over depth cameras to RFID tag based localization—only methods that use the same technical setup (a monocular camera) are discussed. Two types of localization are distinguished: local techniques and global techniques \cite{fox1999monte}. Local techniques need an initial reference point and estimate a robot’s coordinates based on the change in position over time. Once they lost track, the robot’s position can typically not recovered. The approaches also suffer from “drift” since errors are accumulating over time. Global techniques are more powerful and do not need an initial reference point. They can recover when temporarily losing track and address the kidnapped robot problem, in which a robot is carried to an arbitrary location \cite{engelson1992error}.

In general, comparing the accuracy and run-time of different localization methods is difficult: target systems and test environments are often too different to draw comparisons. The annual Microsoft Indoor Localization Competition¹¹https://www.microsoft.com/en-us/research/event/microsoft-indoor-localization-competition-ipsn-2016/ aims at setting a standardized testbed for comparing near real-time indoor location technologies. However, since the competition does not require lightweight platforms and allows for using external infrastructure such as WiFi routers, no vision-only approach was presented at the competition yet.