Particle Smoother-Based Landmark Mapping for the SLAM Method of an Indoor Mobile Robot with a Non-Gaussian Detection Model

Table 2

Related works: an RFID-based localization system.

Works	Sensors	Dense of tags	Manually built tag map	Robot localization	Tag localization	SLAM	Precision (mm)
Yang et al. [24]	UHF-RFID	9 tags/m^2	✓	Hybrid particle filter	✗	✗	186
Hähnel et al. [25]	UHF-RFID+LRF	Attached to objects	✗	Monte Carlo localization (with known tag position)	Particle filter (with known robot route)	✗	Much better than only using a laser
Joho et al. [26]	UHF-RFID	350 tags	✗	Monte Carlo localization (with known tag position)	Particle filter (with known robot route)	✗	355
Kleiner et al. [15]	UHF-RFID+odometry	10tags/90,000 m^2	✗	Graph optimization	Graph optimization	✓	5,640
Kodaka et al. [27]	LF-RFID	Lattice, 16 tags/m^2	✓	Particle filter	✗	✗	<100
Yang and Wu [28]	HF-RFID	100 tags/m^2	✓	Particle filter	✗	✗	53
Mi and Takahashi [29]	HF-RFID	100 and 16 tags/m^2	✗	Monte Carlo localization	✗	✗	25 and 22
Wang and Takahashi [16]	HF-RFID	100 and 16 tags/m^2	✗	Particle filter	Particle filter	✓	14.5 and 21 (with detection range of 30)
Our approach	HF-RFID	4 tags/m^2	✗	Particle filter	Particle smoother	✓	>25 and <55 (with detection range from 50 to 250)