Ultra Wideband Assisted Localization of Semi-Autonomous Floor Scrubber

Abstract

The paper describes the design and features of the novel semi-autonomous floor scrubber add-on module, used for cleaning large indoor spaces. Module is designed in such a manner that it can be easily attached and detached from scrubber machine and that additional sensors can be introduced if needed. The paper focuses on the localization capabilities of the machine in several sensor setups with emphasis on the use of ultra wideband (UWB) real-time localization system (RTLS). It also proposes fusion of sensor data from several sources including novel use of wheel encoder’s data in UWB setup. Analysis is performed in terms of localization accuracy and reliability as well as associated advantages and disadvantages. Obtained results demonstrated that inclusion of UWB subsystem, despite its price and accuracy (20 cm in ideal, line of sight, conditions), based on behavior switching yields more reliable and accurate results in open spaces (up to 25 times in position and 2 times in orientation) and that its accuracy can be further improved with inclusion of wheel encoder data.

Keywords

• UWB
• robot localization
• navigation
• EKF
• service robotics
• sensor fusion