Sensors for localization: an overview

Project description

Robot localization presents challenges in indoor navigation. To localize, a map and at least one sensor is required, but it remains unclear which set of sensors provide the highest localization accuracy, while also making it impractical to test each sensor due to the high cost.

Context

Localization is one of the big problems for robot navigation due to dynamic environments, the build up of uncertainties, and drift. To achieve accurate localization, robots rely on maps and sensors. Multiple sensors are usually used to solve this problem by using sensor fusion. However, determining which sensors contribute most to localization accuracy remains unclear, and it would be cost inefficient to blindly test each sensor. This study aims to get more knowledge in the field of localization and sensors, to create an overview of usable sensors for localizaton, and the possible metrics to compare all types of different sensors.

Results

An overview of sensors is presented, compared based on localization
accuracy, drift, range, computational load, and cost. However, all the experiments have been conducted in a different context and environment, which causes some uncertainty about how truthful the values are. Therefore, future research is suggested, where many experiments are conducted on many sensors within the same context and environment.