Indoor Positioning Systems (IPS) traditionally rely on odometry and building infrastructures like WiFi, often supplemented by building floor plans for increased accuracy. However, the limitation of floor plans in terms of availability and timeliness of updates challenges their wide applicability. In contrast, the proliferation of smartphones and WiFi-enabled robots has made crowdsourced radio maps – databases pairing locations with their corresponding Received Signal Strengths (RSS) – increasingly accessible. These radio maps not only provide WiFi fingerprint-location pairs but encode movement regularities akin to the constraints imposed by floor plans. This work investigates the possibility of leveraging these radio maps as a substitute for floor plans in multimodal IPS. We introduce a new framework to address the challenges of radio map inaccuracies and sparse coverage. Our proposed system integrates an uncertainty-aware neural network model for WiFi localization and a bespoken Bayesian fusion technique for optimal fusion. Extensive evaluations on multiple real-world sites indicate a significant performance enhancement, with results showing ∼ 25% improvement over the best baseline.

Wi-Fi radiomaps offer a convenient solution to position users in GNSS denied areas, however the process of obtaining them through manual surveys is expensive and time consuming, while the use of crowdsourced uncalibrated data is noisier and lacks coverage. In this work we propose to calibrate the crowdsourced data using an optimization framework based on GraphSLAM using GPU processing, in order to automatically cluster the data into floors and perform 2D graph optimization and alignment, combined with opportunistic map matching when floor plans are available. Using our approach we are able to generate radio maps of multi floor buildings, processing data from thousands of user and millions of fingerprints. In our experiments we are able to produce a radiomap with a floor accuracy above 98% and median positioning error of 7.63 m in a 14 floor shopping mall (Joy City) in Beijing, China using single shot WKNN for localization.

Traditional Indoor Positioning Systems (IPS) use odometry, WiFi, and often building floor plans for accuracy. However, floor plan limitations have shifted attention to crowd-sourced radio maps, popularized by smartphones and WiFi-integrated robots. These maps pair locations with Received Signal Strengths (RSS) and reflect movement patterns similar to floor plans. Our research explores using radio maps as an alternative to floor plans in IPS. We've developed a new framework that combines an uncertainty-aware neural network for WiFi positioning with a Bayesian fusion method. Testing in real-world scenarios showed about a 25% performance increase compared to the leading baseline.

K-Neares Neighbors (KNN) and its variant weighted KNN (WKNN) have been explored for years in both academy and industry to provide stable and reliable performance in WiFi-based indoor positioning systems. Such algorithms estimate the location of a given point based on the locality information from the selected nearest WiFi neighbors according to some distance metrics calculated from the combination of WiFi received signal strength (RSS). However, such a process does not consider the relational information among the given point, WiFi neighbors, and the WiFi access points (WAPs). Therefore, this study proposes a novel Deep Neighborhood Learning (DNL). The proposed DNL approach converts the WiFi neighborhood to heterogeneous graphs, and utilizes deep graph learning to extract better representation of the WiFi neighborhood to improve the positioning accuracy. Experiments on 3 real industrial datasets collected from 3 mega shopping malls on 26 floors have shown that the proposed approach can reduce the mean absolute positioning error by 10% to 50% in most of the cases. Specially, the proposed approach sharply reduces the root mean squared positioning error and 95% percentile positioning error, being more robust to the outliers than conventional KNN and WKNN.

In recent years WiFi became the primary source of information to locate a person or device indoor. Collecting RSSI values as reference measurements with known positions, known as WiFi fingerprinting, is commonly used in various positioning methods and algorithms that appear in literature. However, measuring the spatial distance between given set of WiFi fingerprints is heavily affected by the selection of the signal distance function used to model signal space as geospatial distance. In this study, the authors proposed utilization of machine learning to improve the estimation of geospatial distance between fingerprints. This research examined data collected from 13 different open datasets to provide a broad representation aiming for general model that can be used in any indoor environment. The proposed novel approach extracted data features by examining a set of commonly used signal distance metrics via feature selection process that includes feature analysis and genetic algorithm. To demonstrate that the output of this research is venue independent, all models were tested on datasets previously excluded during the training and validation phase. Finally, various machine learning algorithms were compared using wide variety of evaluation metrics including ability to scale out the test bed to real world unsolicited datasets.

This paper presents a new indoor location concept named Light-matching which uses the perceived gradient in the illumination from unmodified indoor lights to achieve accurate physical location. The proposed method does not require any illumination calibration, just the pre-storage of the position and size of all lights in a building, irrespective of their current on/off state. The Light-matching method also requires the estimation of the relative displacement and orientation change of a person which is done by inertial Pedestrian Dead-Reckoning (PDR). Even from an initially unknown location and orientation, whenever the person passes below a switched-on light spot, the location likelihood is iteratively updated until it potentially converges to an unimodal probability density function. The time to converge to an unimodal position hypothesis depends on the number of lights detected and the asymmetries/irregularities of light distributions. The light-matching technique can be used alone or in cooperation with other signals of opportunity (WiFi, Magnetometers or Map-matching) to obtain a continuous high accuracy indoor localization system. This paper presents the basic description of the light-matching concept, the implementation details using a particle filter, and the evaluation of the method by simulation. The performance of the integrated solution can achieve a localization error lower than 1 m in most of the cases.

In personal indoor guidance and navigation applications, it is desirable to determine both the user's position and spatial orientation, since this enables to provide him with directions in a more natural way. The common method for estimation of orientation uses the magnetic compass included in numerous portable devices; this approach becomes unreliable when the device's own orientation relative to the user is unknown, or if the environment contains metallic structures which disturb the magnetic field. In this work we propose an alternative approach for estimation of personal position and orientation which requires no additional sensors and is based solely in the attenuation of radiofrequency signal strength (RSS) introduced by the user's own body. The physical system consists in a set of emitting RF nodes located at known positions in the environment, and two RF receivers placed in the front and back parts of the user's body (alternatively, a single receiver with two antennas). We show how the position of the user can be inferred from the average received signal strength, while his orientation is estimated from the difference of signal strengths from the two receivers, by adapting two commonly used techniques in indoor localization: least-squares minimization and Bayesian filters. We have validated experimentally both techniques with an RFID-based localization system arranged in a typical building, showing that orientation can be determined with a mean error of 0.3 radians, sufficient for most applications.

Indoor positioning is usually based on individual technologies that provide estimates of the trajectory of the person, or measures the ranges or angles between the user and known positions. Each technique has its advantages and problems, and a common way to overcome the drawbacks of single-technology solutions is to fuse the information from several system, but due to their non linear measurements, there is no optimal linear solution. We propose the use of a particle filter to fuse foot mounted inertial measurements with any additional Radio Frequency (RF) measurement. The information fusion is achieved propagating the position of the particles using the relative step displacements obtained from foot mounted Pedestrian Dead Reckoning (PDR), and updating the weights of the particles according to the RF measurements. In our experiments the inertial unit was located in the foot of the user, and the RF system consisted in a Radio Frequency Identification (RFID) receiver in the waist, and an Ultra Wide band (UWB) tag in the chest, but the scheme can be used in any sensor configuration. As the UWB measurements have a significant amount of outliers due to non line of sight conditions generated by the position of the tag in the body, received reflections, etc., we propose a new outlier rejection algorithm based on the compatibility of groups of measurements. The fusion was tested evaluating the inclusion of each of the RF systems and varying the number RFID tags used. The proposed method is able to locate a person with less than 2 m of error (for 90 % of the obtained estimations) in the studied trajectory. This particle filter scheme offers robust indoor positioning with 100 % availability and smooth trajectory estimation thanks to the PDR and limited error due to the RF measurements.

Pedestrian Dead-Reckoning (PDR) and Radio Frequency (RF) ranging/positioning are complementary techniques for position estimation but they usually locate different points in the body (RF in the head/hand and PDR in the foot). We propose to fuse the information from both navigation points using a constraint filter with an upper bound in the distance between the estimated positions of both sensors.

The main problem of Pedestrian Dead-Reckoning (PDR) using only a body-attached IMU is the accumulation of heading errors. The heading provided by magnetometers in indoor buildings is in general not reliable. Recently, a new method was proposed called Heuristic Drift Elimination (HDE) that minimizes the heading error when navigating in buildings. It assumes that the majority of buildings have their corridors parallel to each other, or they intersect at right angles, and consequently most of the time the person walks along a straight path with a heading constrained to one of four possible directions. In this paper we study the performance of HDE-based methods in complex buildings, i.e. with pathways also oriented at 45°, long curved corridors, and wide areas where non-oriented motion is possible. We explain how the performance of the original HDE method can be deteriorated in complex buildings. We also propose an improved HDE method called iHDE, that is implemented over a PDR framework that uses foot-mounted inertial navigation with an Extended Kalman Filter (EKF). The EKF is fed with the iHDE-estimated orientation error, as well as the confidence over that correction. We experimentally evaluated the performance of the proposed iHDE-based PDR method, comparing it with the original HDE implementation. Results show that both methods perform very well in ideal orthogonal narrow-corridor buildings, and iHDE outperforms HDE for non-ideal trajectories (e.g. curved paths).

A common problem in the evaluation of Pedestrian Dead Reckoning (PDR) algorithms is the determination of a good ground truth. Some authors propose the use of external motion capture systems, however, their setup, complexity, synchronization and limited coverage are important limitations. We propose the generation of a simulated IMU signal for pedestrians, that is obtained from a given 3D trajectory (position and attitude). The trajectory can be artificially generated or based on a real human walk pattern. This information can be used as a ground truth for the identification of systematic errors, or to obtain a statistical analysis of the effect of any noise added to the simulated signal. Any specific IMU can be simulated by adding its characteristic error pattern, and modifying them, the most influential IMU characteristics can be determined, and if possible minimized. We tested a PDR method based on an Inertial Navigation System (INS) using an Extended Kalman Filter (EKF) with a noiseless IMU signal. Since failures were detected in the stance phase, we proposed and tested some improvements. The influence of adding specific error patterns to the IMU signal were determined measuring their effect on the evolution of the standard deviation of the position error over time. The most influential source of error for an INS mechanization is the bias in the gyroscope, however the EKF-based PDR algorithm showed to diminish in a significant way many of the positioning errors. The IMU-simulation method is proposed as a way to compare several algorithms and to test new PDR improvements during algorithm design.