Pedestrian Navigation System Research Articles

Shortest Path Algorithms for Pedestrian Navigation Systems

Efficient shortest path algorithms are of key importance for routing and navigation systems. However, these applications are designed focusing on the requirements of motor vehicles, and therefore, finding paths in pedestrian sections of urban areas is not sufficiently supported. In addition, finding the shortest path is often not adequate for urban sidewalk routes, as users of these applications may also be interested in alternative routes that, although slightly longer, possess other desirable features and properties. According to the literature, the search for alternative routes is carried out mainly using the k-shortest paths (KSP) algorithm which represents an ordered list of all available alternatives. Even though various KSP algorithms have been proposed, to the best of our knowledge, there is no research addressing all issues inherent in a pedestrian navigation system. The purpose of this paper is to present a heuristic algorithm for graph datasets that implements a penalty-based method which, by increasing certain edge weights, effectively searches for the most accessible alternative paths in multi-route cases. To demonstrate how the algorithm works, we present experimental results on finding the most accessible paths in pedestrian sections of the historical center of Thessaloniki city.

Recent Advances in Pedestrian Navigation Activity Recognition: A Review

Pedestrian navigation activity recognition (PNAR) has a significant impact on positioning and tracking performance. Smartphone-based PNAR utilizes measurements from sensors embedded in smartphone to identify pedestrian motion mode and smartphone usage mode. Compared with other PNAR technologies, smartphone-based PNAR has the advantage of autonomy and practicality. Though various PNAR recognition methods based on smartphone have been proposed, up-to-date review papers that summarize relevant technologies, methods, and solutions of PNAR are relatively less so far in the literature. This work aims to present an elaborated, timely, and valuable survey for different smartphone-based PNAR techniques, including sensor calibration, pedestrian modeling, pedestrian motion and smartphone usage modes definition, workflow, state-of-the-art research methods, public datasets, performance evaluation metrics, and relative competitions. Finally, applications, current challenges and future research trends are discussed. The work propels a better understanding of existing navigation activity recognition methods. It is helpful for researchers to further design PNAR methods, and develop more accurate and robust pedestrian positioning and navigation systems.

New Zero-Velocity Detector for Low-Cost Inertial Pedestrian Navigation

A new Zero Velocity Interval (ZVI) detector is proposed and investigated in this article to augment the computation of a pedestrian’s position. In a low-cost pedestrian navigation system, the position of a pedestrian can be computed by using inertial navigation algorithms. The low-cost pedestrian navigation system employs an inertial measurement unit (IMU) comprised of accelerometer and gyroscope sensors to record acceleration and attitude rate. These measurements, when integrated mathematically, yield velocity and position. Similarly, attitude rate changes to attitude. The algorithm will then be able to figure out the changes in pedestrian position with the proper attitude. However, due to its low-cost nature, these sensors are built in such a way that their measurements are easily corrupted by noise, and once integrated mathematically, the measurement error grows exponentially. Zero Velocity Update (ZUPT) algorithm is frequently used to limit these errors. It works by detecting the ZVI that occur when the foot is on the ground. Assuming that the foot on the ground should have zero velocity, any remaining velocity measurements detected during this period are considered an error and are fed back to the navigation algorithm for correction. To detect the ZVI, a few commonly used detectors, namely Angular Rate Energy (ARE), Acceleration Moving Variance (MV), Acceleration Magnitude (MAG), and Generalized Likehood Ratio Test (GLRT) were revisited. These detectors were tested with real-world datasets of walking pedestrians. Then, a new detector, the Angular Rate Moving Variance (ARMV) detector, is proposed, and its performance is compared to that of the existing detectors.

A Pedestrian Indoor Navigation System Using Deep-Learning-Aided Cellular Signals and ZUPT-Aided Foot-Mounted IMUs

A signal-sensor-based indoor pedestrian navigation system is developed. The proposed system: 1) utilizes a foot-mounted inertial navigation system (INS), in which the accumulated errors are mitigated via a zero velocity update (ZUPT) approach; 2) exploits opportunistically cellular long-term evolution (LTE) signals in a deep neural network (DNN)-based synthetic aperture navigation (SAN) framework, in which the pedestrian’s motion is utilized to suppress multipath-induced errors. The proposed DNN-SAN-LTE-ZUPT-INS (DUALS) indoor pedestrian navigation system utilizes the complementary the desirable characteristics of both subsystems, coupled via two architectures: (a) loosely-coupled and (b) tightly-coupled. This paper designs and assesses both architectures experimentally in an indoor environment. The experimental study demonstrates a pedestrian traversing a trajectory of 600 m in 14 minutes, including a stationary period, straight segments, up/down the stairs, and riding in an elevator, while receiving signals from 4 LTE base stations (also known as evolved node Bs (eNodeBs)). The proposed tightly-coupled DUALS system exhibited a three-dimensional (3-D) position root mean-squared error (RMSE) of 1.34 m, outperforming the loosely-coupled DUALS, ZUPT-aided INS, and LTE-DNN-SAN, which achieved a position RMSE of 1.38, 1.49, and 1.97 m, respectively.

A Robust GNSS/PDR Integration Scheme with GRU-Based Zero-Velocity Detection for Mass-Pedestrians

Aiming at the problem of high-precision positioning of mass-pedestrians with low-cost sensors, a robust single-antenna Global Navigation Satellite System (GNSS)/Pedestrian Dead Reckoning (PDR) integration scheme is proposed with Gate Recurrent Unit (GRU)-based zero-velocity detector. Based on the foot-mounted pedestrian navigation system, the error state extended Kalman filter (EKF) framework is used to fuse GNSS position, zero-velocity state, barometer elevation, and other information. The main algorithms include improved carrier phase smoothing pseudo-range GNSS single-point positioning, GRU-based zero-velocity detection, and adaptive fusion algorithm of GNSS and PDR. Finally, the scheme was tested. The root mean square error (RMSE) of the horizontal error in the open and complex environments is lower than 1 m and 1.5 m respectively. In the indoor elevation experiment where the elevation difference of upstairs and downstairs exceeds 25 m, the elevation error is lower than 1 m. This result can provide technical reference for the accurate and continuous acquisition of public pedestrian location information.

Pedestrian Inertial Navigation Based on Full-Phase Constraints of Lower Limb Kinematics

Due to the low-cost magnetic/inertial measurement unit (MIMU) is easily affected by noise, the positioning result is inaccurate. In this study, to improve the precision of pedestrian positioning, a lower limb constraint model is constructed based on the gait characteristics and biomechanical relationship. Dual MIMUs are strapped to the foot and shank of the same leg to form a pedestrian inertial navigation systems (PINS). Firstly, taking the MIMU mounted on the shank as the reference point, a new velocity constraint relation is established. Combined with kinematics and Coriolis theorem, the velocity of the reference point is calculated by two MIMUs, respectively. Then, the two calculations are optimized via a Kalman filter. Secondly, a geometric mode is built owing to the position relationship between the ankle and two MIMUs. Using cosine theorem, a correction method with complementary constraints is established. In addition, different from the traditional zero-velocity update (ZUPT), the velocity and geometric constraints are applied throughout the gait cycle. Consequently, the proposed constraint methods can effectively suppress the position drift. To verify the effectiveness of the proposed algorithm, several experiments with different trajectories and velocities are designed. The results show that the trajectories obtained by the proposed methods are closer to the actual situations.

F²IMU-R: Pedestrian Navigation by Low-Cost Foot-Mounted Dual IMUs and Interfoot Ranging

Foot-mounted inertial sensors become popular in many indoor or GPS-denied applications, including but not limited to medical monitoring, gait analysis, soldier, and first responder positioning. However, the foot-mounted inertial navigation relies largely on the aid of zero velocity update (ZUPT) and has encountered inherent problems, such as heading drift. This article implements a pedestrian navigation system based on dual foot-mounted low-cost inertial measurement units (IMUs) and interfoot ultrasonic ranging. The observability analysis of the system is performed to investigate the roles of the ZUPT measurement and the foot-to-foot ranging measurement in improving the state estimability. A Kalman-based estimation algorithm is mechanized in the Earth frame, rather than in the common navigation frame, which is found to be effective in depressing the linearization error in Kalman filtering. An ellipsoid constraint in the Earth frame is also proposed to further restrict the height drift. Simulation and real field experiments show that the proposed method has better robustness and positioning accuracy (about 0.1%–0.2% traveled distance) than the traditional pedestrian navigation schemes do.

A Reconstruction Filter for Saturated Accelerometer Signals Due to Insufficient FSR in Foot-Mounted Inertial Navigation System

In pedestrian inertial navigation, one possible placement of Inertial Measurement Units (IMUs) is on a footwear. This placement allows to limit the accumulation of navigation errors due to the bias drift of inertial sensors and is generally a preferable placement of sensors to achieve the highest precision of pedestrian inertial navigation. However, inertial sensors mounted on footwear experience significantly higher accelerations and angular velocities (10s of g and 1000s of deg/sec) during regular pedestrian activities than during more conventional navigation tasks, which could exceed Full Scale Range (FSR) of many commercial-off-the-shelf IMUs, therefore degrading accuracy of pedestrian navigation systems. This paper proposes a reconstruction filter to mitigate localization error in pedestrian navigation due to insufficient FSR of inertial sensors. The proposed reconstruction filter approximates immeasurable accelerometer’s signals with a triangular function and estimates the size of the triangles using a Gaussian Process (GP) regression. To evaluate performance of pedestrian navigation systems enhanced by the proposed reconstruction filter, we conducted two series of indoor pedestrian navigation experiments with a VectorNav VN-200 IMU and an Analog Device ADIS16497-3 IMU, which have accelerometer’s FSR of ±16 g and ±40 g, respectively. In the first series of experiments, forces experienced by the foot-mounted IMUs did not exceed the FSRs of the sensors, while in the second series, the forces surpassed the FSR of the VN-200 IMU and saturated the accelerometer’s readings. The saturated accelerometer’s readings reduced the accuracy of estimated positions using the VN-200 by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${1.34}\times $ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${3.37}\times $ </tex-math></inline-formula> along horizontal and vertical directions. When applying our proposed reconstruction filter to the saturated accelerometer’s measurements, the navigation accuracy along horizontal and vertical directions was increased by 5% and 50%, respectively, as compared to using unreconstructed signals.

A Novel Calibration Method for Gyro-Accelerometer Asynchronous Time in Foot-Mounted Pedestrian Navigation System.

Pedestrian Navigation System (PNS) is one of the research focuses of indoor positioning in GNSS-denied environments based on the MEMS Inertial Measurement Unit (MIMU). However, in the foot-mounted pedestrian navigation system with MIMU or mobile phone as the main carrier, it is difficult to make the sampling time of gyros and accelerometers completely synchronous. The gyro-accelerometer asynchronous time affects the positioning of PNS. To solve this problem, a new error model of gyro-accelerometer asynchronous time is built. The effect of gyro-accelerometer asynchronous time on pedestrian navigation is analyzed. A filtering model is designed to calibrate the gyro-accelerometer asynchronous time, and a zero-velocity detection method based on the rate of attitude change is proposed. The indoor experiment shows that the gyro-accelerometer asynchronous time is estimated effectively, and the positioning accuracy of PNS is improved by the proposed method after compensating for the errors caused by gyro-accelerometer asynchronous time.

Personalized landmark adaptive visualization method for pedestrian navigation maps: Considering user familiarity

AbstractLandmark‐based pedestrian navigation can assist pedestrians in navigating successfully. Previous studies have investigated the factors affecting the cognitive efficiency of landmark visualization in terms of both the visual salience of landmarks and the personal characteristics of users. However, empirical studies and applications that consider the influence of spatial familiarity on landmark representation are limited. In this article, we propose a personalized landmark adaptive visualization method for pedestrian navigation maps considering user familiarity. We first explore the influence of spatial familiarity on landmark salience and symbols using cognitive experiments. The results showed that unfamiliar people preferred strong visual salience landmarks and image‐based symbols, while familiar people preferred strong semantic salience landmarks and text‐based symbols. Based on these results, a mathematical model of landmark salience for selecting personalized landmarks is proposed, and association rules between landmark salience and symbols are mined. Finally, the framework of a landmark visualization method is proposed based on the rules. To verify the effectiveness of the proposed method, a prototype system is developed, and a comparative experiment is conducted with a Baidu map. Experimental results showed that the proposed method has direct practical implications for the development of pedestrian navigation systems, depending on different target users.

The Improved Constraint Methods for Foot-Mounted Pedestrian Three-Dimensional Inertial Navigation

The foot-mounted pedestrian navigation system (PNS) that uses microelectromechanical systems (MEMS) inertial measurement units (IMUs) to track the person’s position. However errors accumulate over time during inertial navigation solutions, which affects the positioning precision. In this paper, a multicondition zero velocity detector is used to detect the stance phase of gait. Then the errors are corrected in the stance phase and the swing phase, respectively, through the Kalman filter. When pedestrians are going up and down the stairs, the divergence of height will reduce the accuracy of three-dimensional positioning. In this paper, an accelerometer and a barometer are used to obtain altitude variation, and after that the stair condition detection (SCD) algorithm is proposed to correct the height of Kalman filter output and detect the walking state of pedestrians. Through theoretical research and field experiments, these algorithms are studied carefully. The results of the experiment show that the algorithm proposed in this paper can effectively eliminate errors and achieve more accurate positioning.

A survey of step length estimation models based on inertial sensors for indoor navigation systems

SummaryIn many applications (i.e., navigation of vehicles in traffic, museum guide, guidance to firefighters in poor visibility, and elderly care assistance systems), location‐based services (LBS) rely on the global positioning system (GPS) for accurate positioning. LBS can be challenging to employ in an indoor environment because of the lack of a line‐of‐sight (LOS) component and other signal propagation issues. Over the last decade, researchers worldwide have been motivated to solve the problem of continuous indoor positioning in situations when GPS is ineffective. We thoroughly reviewed research conducted by many scholars and compiled this survey article, which provides a comprehensive overview of indoor pedestrian navigation systems, prevalent step length estimation models, and the current state of the art. There are two key strategies in which substantial work on indoor localization is carried out. One is a dead reckoning in which the present location of the pedestrian is calculated based on details about previous location, and the other is an inertial navigation system in which the forward acceleration from an inertial sensor strapped down to the user's body is integrated twice to establish the current location. The step length is a crucial component of the inertial sensor‐based navigation in both these strategies. For different users, step length is variable and also changes at different walking speeds for each user. As a result, the focus of this survey is on step length estimation models. The survey describes and compares various approaches to estimating step length from various perspectives, including the research method used, the length of the test path, various walking speeds, the location of the sensor device on the user's body, and the accuracy achieved in estimating step length. Because inertial sensors are embedded in smartphones, it is vital to identify the smartphone mode (i.e., phoning, texting, or pocket) during each step taken while using smartphone‐based indoor navigation. The report concludes with a quantitative evaluation of the performance of smartphone mode recognition modules. At the end, we list out challenges still to be addressed for scholars working on indoor navigation.

Landmark selection preferences of young students under orientation task within street environment

ABSTRACT Landmarks are important spatial reference in orientation because they can help pedestrians understand the relative spatial relationship between environment and themselves, yet not all landmarks are suitable for orientation. Pedestrians usually select the proper ones based on their own cognition, which is a time-consuming process. If we can predict popular landmarks (preferred by most people) from a first-person perspective, orientation will be much easier. To better understand landmark selection behaviour, the research in this paper designs an orientation experiment within virtual street environment to investigate young students aged 21–31 about preferences for landmarks. Based on the selection results, this paper further constructs a random forest model to choose popular landmarks. Results indicate that one or two landmarks are enough to meet young students’ needs for orientation, and the time spent on orientation significantly reduces by using chosen popular landmarks in most scenes. In addition, classification model achieves a desirable performance in popular landmark selection: F1 score and AUC of the predictive results reached 0.820 and 0.855, respectively, where landmark route deviation, visual salience and semantic salience serve as essential factors in influencing young students’ selection. These findings can provide a feasible reference for optimising pedestrian navigation system.

A Zero-Position-Difference ZUPT Method for Foot-Shank-Mounted Pedestrian Inertial Navigation Systems

Pedestrian navigation system is an important application of inertial sensor technology because it does not depend on any external facilities. However, the positioning accuracy is limited due to the cumulative error of the inertial sensor. Zero Velocity Update (ZUPT) is an effective method to compensate the cumulative error. To avoid the jitter and random noise defects of single inertial sensor, a zero-position-difference ZUPT combined with foot and shank kinematics information is proposed in this work. Firstly, the zero position difference constraint matrix with ankle joint as the connection point is established through analyzing the local motion characteristics of foot and shank. In order to get accurate attitude and position by fusing the measurement information of foot and shank, an improved extended Kalman particle filter (EKPF) based on the zero position difference is proposed. Moreover, according to the gait characteristics of foot and shank, a weight assignment strategy is designed to detect the zero velocity intervals of walking. The feasibility and effectiveness of the algorithm are verified by experiments. The results show that the error of the improved zero velocity detector is about 22% less than that of the foot-mounted zero velocity detector. The improved EKPF method reduces the positioning error by more than 35% and 45% compared with the traditional EKPF method based on shank and foot, respectively.

Choose your own route – supporting pedestrian navigation without restricting the user to a predefined route

ABSTRACT State-of-the-art mobile pedestrian navigation systems often employ GPS or other positioning methods for continuous tracking of users, and thus provide them with in-situ turn-by-turn route guidance along a desired route. However, studies have shown that user experience and acquisition of spatial knowledge decrease due to the “blind” following of such turn-by-turn navigation systems. This paper proposes a novel interface concept for mobile pedestrian navigation systems that provide navigation guidance without restricting the users to a predefined route. Specifically, the proposed novel user interface was based on the concept of the Potential Route Area (PRA), which defines a dynamic area consisting of all potential routes not longer than a certain detour the user is willing to accept. Within that area, the user can freely choose his/her own route and alter it anytime, and can still arrive at the destination within the desired detour tolerance. As a proof of concept, the proposed PRA-based system was then tested against a conventional turn-by-turn navigation system, represented by Google Maps, in a real-world navigation experiment, which revealed that the acquisition of spatial knowledge and user experience were substantially improved when using the PRA-based system. This can be explained by the fact that the PRA-based system provides its users with more freedom in choosing their own route.

Deep neural network‐based adaptive zero‐velocity detection for pedestrian navigation system

The zero-velocity update (ZUPT) method is an effective way to reduce accumulated velocity errors of pedestrian navigation systems (PNSs). For a typical scheme, a stance phase detection module based on a fixed threshold is used to trigger the ZUPT algorithm. However, the detector is not robust enough for dynamic gait speeds. The false detection will degrade the navigation performance. In this letter, to improve the stance phase detector, the adaptive zero-velocity detection problem is cast under dynamic gait speeds as a sequential threshold of a traditional detector inferring problem and a zero-velocity detection framework proposed by combining a deep neural network with a traditional binary gait phase detector. Sufficient experimental results show that the proposed method outperforms other discussed learning-based methods taking into account the trade-off among model performance, structure, and size. Compared with the traditional method with a fixed threshold, the real-world high-dynamic positioning experiments show that this proposed method reduces the root mean squared error (RMSE) of absolute distance error by 48.7%, RMSE of start-end error by 12.5%, and average RMSE of position error by 19.2%.

Adaptive Cardinal Heading Aided for Low Cost Foot-Mounted Inertial Pedestrian Navigation

The use of a low-cost MEMS-based Inertial Measurement Unit (IMU) provides a cost-effective approach for navigation purposes. Foot-mounted IMU is a popular option for indoor inertial pedestrian navigation, as a small and light MEMS-based inertial sensor can be tied to a pedestrian's foot or shoe. Without relying on GNSS or other external sensors to enhance navigation, the foot-mounted pedestrian navigation system can autonomously navigate, relying solely on the IMU. This is typically performed with the standard strapdown navigation algorithm in a Kalman filter, where Zero Velocity Updates (ZVU) are used together to restrict the error growth of the low-cost inertial sensors. ZVU is applied every time the user takes a step since there exists a zero velocity condition during stance phase. While velocity and correlated attitude errors can be estimated correctly using ZVUs, heading error is not because it is unobservable. In this paper, weextend our previous work to correct the heading error by aiding it using Multiple Polygon Areas (MPA) with adaptive weighting factor. We termed the approach as Adaptive Cardinal Heading Aided Inertial Navigation (A-CHAIN). We formulated an adaptive weighting factor applied to measurement noise to enhance measurement confidence. We then incorporated MPA heading into the algorithm, whereas multiple buildings with the same orientation are grouped together and assigned a specific heading information as a priori. Results shown that against the original CHAIN, the proposed Adaptive-CHAIN improved the position accuracy by more than five-fold.

Smartphone-Based Activity Recognition in a Pedestrian Navigation Context.

In smartphone-based pedestrian navigation systems, detailed knowledge about user activity and device placement is a key information. Landmarks such as staircases or elevators can help the system in determining the user position when located inside buildings, and navigation instructions can be adapted to the current context in order to provide more meaningful assistance. Typically, most human activity recognition (HAR) approaches distinguish between general activities such as walking, standing or sitting. In this work, we investigate more specific activities that are tailored towards the use-case of pedestrian navigation, including different kinds of stationary and locomotion behavior. We first collect a dataset of 28 combinations of device placements and activities, in total consisting of over 6 h of data from three sensors. We then use LSTM-based machine learning (ML) methods to successfully train hierarchical classifiers that can distinguish between these placements and activities. Test results show that the accuracy of device placement classification (97.2%) is on par with a state-of-the-art benchmark in this dataset while being less resource-intensive on mobile devices. Activity recognition performance highly depends on the classification task and ranges from 62.6% to 98.7%, once again performing close to the benchmark. Finally, we demonstrate in a case study how to apply the hierarchical classifiers to experimental and naturalistic datasets in order to analyze activity patterns during the course of a typical navigation session and to investigate the correlation between user activity and device placement, thereby gaining insights into real-world navigation behavior.

A Pedestrian Navigation Method Based on Construction of Adapted Virtual Inertial Measurement Unit Assisted by Gait Type Classification

Aiming at the problem that the pedestrian navigation system cannot effectively perform positioning and navigation when foot-mounted IMU exceeds its measurement ranges during vigorous activities or collisions with obstacles, a novel pedestrian navigation method is proposed based on construction of adapted virtual inertial measurement unit (VIMU) assisted by gait type classification. The method utilizes attention-based convolutional neural network (CNN) to classify gait types of the pedestrian equipped with the navigation system. Then, with the inertial data from pedestrian's thigh and foot collected synchronously via actual IMUs as training and testing samples, correspondent Resnet-Gated Recurrent Unit (Resnet-GRU) deep hybrid neural network models are built according to diverse gait types. Through these models, virtual foot-mounted IMU is constructed for positioning in case of physical foot-mounted IMU overrange. The experimental results show that, the proposed VIMU construction method can effectively improve the performance and reliability of zero velocity update-based pedestrian navigation system when the pedestrian moves or hits some obstacles with foot vigorously, thus enhancing the adaptability of the navigation system in complex and unknown terrains. The positioning error under comprehensive gait types is about 1.45% of the total walking distance, which meets the accuracy requirement in military and civilian applications.

A Real-Time GNSS/PDR Navigation System for Mobile Devices

In this article, a smart pedestrian navigation system is developed to be implemented in a common smartphone. The main phases that characterize a pedestrian navigation system that is based on dead reckoning are introduced. A suitable Phase-Locked Loop is designed and the algorithm to estimate the direction of the user’s motion between one step and the next is developed. Finally, a suitable multi-rate Kalman filter (KF) is considered to merge the information from the pedestrian dead reckoning (PDR) navigation with the data provided by the global navigation satellite systems (GNSS). The proposed GNSS/PDR navigation system is implemented in Simulink as a finite-state machine and allows to define a trade-off between energy-saving and performance improvement in terms of position accuracy. The presented pedestrian navigation system is independent of the body-worn location of the smartphone and implements a compensation strategy of the systematic errors that are committed on the step-length estimation and the determination of the motion direction. Moreover, several tests are performed by walking in urban and suburban environments: the results show that a suitable trade-off between energy-saving and position accuracy can be reached by switching the GNSS receiver on and off.