PDR/UWB Based Positioning of a Shopping Cart

In this article, we consider indoor positioning of a shopping cart in a store with a hybrid approach combining ultra wideband (UWB) and a people dead reckoning (PDR) system. While UWB can provide a very accurate positioning estimate in ideal circumstances, its accuracy reduces in non-line-of-sight s...

Full description

Saved in:
Bibliographic Details
Published in:IEEE sensors journal 2021-05, Vol.21 (9), p.10864-10878
Main Authors: Vandermeeren, Stef, Steendam, Heidi
Format: Article
Language:English
Subjects:
Acceleration
Acceleration measurement
accelerometer
Accelerometers
Accuracy
Dead reckoning
Filtration
Inertial platforms
Inertial sensing devices
Kalman filters
Position measurement
Sensor fusion
Sensor systems
Shopping
shopping cart
Shopping carts
Step counter
Systems architecture
Ultrawideband
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this article, we consider indoor positioning of a shopping cart in a store with a hybrid approach combining ultra wideband (UWB) and a people dead reckoning (PDR) system. While UWB can provide a very accurate positioning estimate in ideal circumstances, its accuracy reduces in non-line-of-sight situations and the update rate decreases when the number of users increases. To solve these issues, each shopping cart is not only equipped with a UWB tag, but also with an inertial measurement unit (IMU) sensor to determine the step length and heading of the user moving the shopping cart, in order to track the shopping cart in between two UWB measurements. As the IMU is not attached to the body, the measured acceleration will be different than in other works considering PDR systems. In this article, we therefore first extract a model for the acceleration, and use the resulting model in the PDR system, where we look for the best acceleration component to track the cart. To combine the PDR and UWB information, we consider two approaches, i.e. Kalman and particle filtering, and compare both approaches. Moreover, we investigate the effect of the presence of map information of the store on the trajectory information. Our experiments show that the average positioning error using UWB only equals 62.6 cm, while the Kalman and particle filter result in an accuracy of respectively 34.1 cm and 41.3 cm, and when using map information in combination with particle filtering, the accuracy improves to 28.0 cm.
ISSN:1530-437X
1558-1748
DOI:10.1109/JSEN.2021.3060110