Mapping and Localization with RFID Technology
Outline
Introduction
Related Work
Probabilistic Sensor Model
Mapping
Localization
Experimental Results
Conclusion
Opinion

Localization

In Monte Carlo localization, belief of robot is a set of random samples
Each sample consists of state vector which is pose of robot and weight factor ‘w’
Weight factor provides importance of particle
Beliefs updated on two alternating steps:
In prediction step, new sample for each sample based on weight and model probability of robot’s dynamics since previous update
In correction step, new observation is integrated into sample set by bootstrap resampling
For RFID sensors, samples placed on in potential detection range of sensor