Robot Cleaner Localization Using RFID and Vision System

Localization is a fundamental element required in various applications across fields such as vehicle navigation, smart factories, automation systems, and product shipping services. This paper discusses the fusion of two popular systems- RFID and vision systems to maximize their individual advantages and minimize their respective drawbacks. Using the RFID system, the Received Signal Strength Indicator (RSSI) and tag ID are collected using the fingerprinting technique by utilizing two different types of antennas the far-field antenna (transmits and receives in range of meters) and the near-field antenna (transmits and receives in range of centimetres) and deploying passive tags. After comparing the two in terms of location error accuracy, it was found that the near-field antenna achieved significantly better results, with an accuracy level 55 % better than that of the far-field antenna. In the vision system, a fixed camera captures images for image processing algorithms to detect and localize the cleaner using the appearance-based approach. The accuracy level of this system alone was found to be 81.8%. However, upon fusing both systems, the accuracy levels improved to 90.9 % as each system contributed to the other to reduce each system's drawbacks eliminating the time incompatibility present in the vision system due to computational cost and the sensitivity of the RFID system from the materials used in the environment also, to get the best of RFID system two types of antennas where compared to choose the most accurate, the fusion resulted in an improved and a decreased average error localization system.