Indoor positioning systems (also called as indoor localization, indoor navigation or indoor location-aware systems) consist of technologies (devices and algorithms) providing an estimate of the physical location(s) of some device(s) which cannot be otherwise localized by typical outdoor positioning systems (e.g., GPS, cellular network).
This research area also includes tracking systems (i.e. continuous estimation of trajectories of moving devices), radar systems (where the target is usually non-cooperative), robot navigation systems, etc. Wikipedia contains a page about Real-time Location Systems (RTLS) which provides an overview about some aspects of this wide research area.
State of Art
There is a large body of technical literature dealing with localization systems. A few journals covering this area are IEEE Trans. on Communications, IEEE Trans. on Signal Processing, IEEE Trans. on Information Theory; searching these journals for “pedestrian tracking”, “indoor localization” or “indoor navigation” yields many results.
Indoor localization may be performed using a wide range of different technologies: a) radio technologies (e.g., UWB, RFID, Bluetooth, etc); b) inertial technologies (e.g., to perform dead-reckoning); c) ultra sound technologies; d) optical technologies (e.g., computer-vision systems).
In CommLAB, we have been focusing on radio and inertial systems. The former may be based on Time-Of-Arrival (TOA), Angle-Of-Arrival (AOA) or Received Signal Strength (RSS) measurements, to perform point-to-point ranging between the device to localize, the agent, and fixed devices with known positions, the anchors; multi-lateration algorithms are then used to find the position of the agent, processing all TOA, AOA, RSS measurements. Inertial systems instead, are infrastructure-less positioning systems, based on continuous processing of each infinitesimal movement of the agent, as sensed by an Inertial Measurement Unit (IMU).
Our Research Project
CommLAB has both narrowband radio equipment (operating at 169MHz and 2.4GHz unlicensed bands) and a few IMUs. The ongoing research aims at exploiting radios to perform robust, absolute localization (with room-level accuracy), and exploiting IMUs to perform robust, relative localization (with accuracy of about 1 meter).
A non-technical document summarizing our project follows:
A simple, commented video showing our application running:
Note that the video shows the system employing the inertial technology only, currently.
From the technical point of view, in CommLAB the following topics have been studied:
- Statistical modelling and mitigation of NLOS bias in UWB systems; unfortunately UWB radios remain pretty much expensive, so that later research activity has been focused on low-cost RSS radios.
- Map-aware localization: knowledge of the map of the environment where the localization system is deployed is crucial and, if correctly exploited, may greatly improve localization performances. We have studied the performance limits of map-aware systems, to find answers to questions like “what is the performance gain given by map-awareness?”, “how much additional complexity adds map-awareness?”, etc.
- Statistical modelling of NLOS bias in RSS-based systems: we obtained a simple, yet effective model to compensate for NLOS bias; furthermore, we have been working on quasi-optimal, reduced-complexity algorithms for indoor localization, exploiting such model.
- Non-linear, reduced-complexity, map-aware sequential estimation: map-aware inertial navigation poses many challenges, because of non-linear nature of the dynamic and measurement models involved; conventional Bayesian approaches (e.g., particle filtering) are unfeasible because of the high-dimensionality of the problems involved and new approaches are required. We are actively working on an innovative filtering solution.
Open Research Topics
Many open issues deserve additional research:
- Human body shadowing on RSS and optimal placement of antenna for pedestrian tracking RSS-based;
- Wearable antenna design and choice of optimal frequency for indoor RSS localization;
- Statistical modelling of pedestrian walk for inertial system improvement;
- Fingerprinting of indoor magnetic anomalies for navigation;
- Implementation of non-linear tracking filters on embedded devices (e.g., particle-based methods on FPGA);
All these topics are well-suited as Master degree theses; if you are interested please write or get in touch with the lab head.
Other Research Groups on this Topic
- LOPSI: group working on pedestrian dead-reckoning, ultrasonic and RF-based localization
- Cooperative Systems Group: group working on pedestrian dead-reckoning
- WITACA: group working on pedestrian dead-reckoning and RSS-based localization
- DCS Navigation Group: group working on augmentation of radio navigation systems using miniature sensors
For more information on this project, get in touch with us!