Abstract
The location of people, robots, and Internet-of-Things (IoT) devices has become increasingly important. Among the available location technologies, solutions based on ultrawideband (UWB) radio are having much success due to their accuracy, which is ideally at a centimeter level. However, this accuracy is degraded in most common indoor environments due to the presence of obstacles which block or reflect the radio signals used for ranging. One way to circumvent this difficulty is through robust estimation algorithms based on measurement redundancy, permitting to minimize the effect of significantly erroneous ranges (outliers). This need for redundancy often conflicts with hardware restraints put up by the location system’s designers. In this work, we present a procedure to increase the redundancy of UWB systems and demonstrate it with the help of a commercial system made by Decawave. This system is particularly easy to deploy, by configuring a network of beacons (anchors) and devices (tags) to be located; however, its architecture presents a major disadvantage as each tag to be located can only measure ranges to a maximum of four anchors. This limitation is embedded in the Positioning and Networking Stack (PANS) protocol designed by Decawave, and therefore is not easy to bypass without a total redesign of the firmware. In this paper, we analyze the strategies that we have been able to identify in order to provide this equipment with multiple range measurements, and thus enable each tag to be positioned with more than four measured ranges. We will see the advantages and disadvantages of each of these strategies, and finally we will adopt a solution that we implemented to be able to measure up to eight ranges for each mobile device (tag). This solution implies the duplication of the tags at the mobile user, and the creation of a double interleaved network of anchors. The range among tags and the eight beacons is obtained through an API via a wireless BLE protocol at a 10 Hz rate. A robustified Extended Kalman filter (EKF) is designed to estimate, by trilateration, the position of the pair of mobile tags, using eight ranges. Two different scenarios are used to make localization experimentation: a laboratory and an apartment. Our position estimation, which exploits redundant information and performs outlier removal, is compared with the commercial solution limited to four ranges, demonstrating the need and advantages of our multi-range approach.
Highlights
The location of devices or people is becoming increasingly important
While global navigation satellite systems (GNSS) have become quite successful in outdoor environments, there is not yet an equivalent system operative indoors. Devices designed for this purpose receive several names in the literature: Local Positioning Systems (LPS), Indoor Positioning Systems (IPS), Real Time Location Systems (RTLS), etc
We have seen that Positioning and Networking Stack (PANS) protocol is not prepared to work with more than four ranges for each tag to be located, which is a serious hindrance for operation of the system in real indoor scenarios
Summary
The location of devices or people is becoming increasingly important. While global navigation satellite systems (GNSS) have become quite successful in outdoor environments (with unrestricted view to the satellite constellations), there is not yet an equivalent system operative indoors. Devices designed for this purpose receive several names in the literature: Local Positioning Systems (LPS), Indoor Positioning Systems (IPS), Real Time Location Systems (RTLS), etc. A few UWB characterization papers analyze the performance of UWB radio ranging and positioning. In [6], a performance comparison among ten impulse-radio UWB localization systems is referenced. Robust positioning or navigation in spite of these circumstances is a challenging research topic [7]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have