Pedestrian Stride-Length Estimation Based on LSTM and Denoising Autoencoders

Data Fusion of Dual Foot-Mounted INS Based on Human Step Length Model

Pedestrian navigation methods based on inertial sensors are commonly used to solve navigation and positioning problems when satellite signals are unavailable. To address the issue of heading angle errors accumulating over time in pedestrian navigation systems that rely solely on the Zero Velocity Update (ZUPT) algorithm, it is feasible to use the pedestrian's motion constraints to constrain the errors. Firstly, a human step length model is built using human kinematic data collected by the motion capture system. Secondly, we propose the bipedal constraint algorithm based on the established human step length model. Real field experiments demonstrate that, by introducing the bipedal constraint algorithm, the mean biped radial errors of the experiments are reduced by 68.16% and 50.61%, respectively. The experimental results show that the proposed algorithm effectively reduces the radial error of the navigation results and improves the accuracy of the navigation.

Deep Neural Network-Based Fusion Localization Using Smartphones

Indoor location-based services (LBS) have tremendous practical and social value in intelligent life due to the pervasiveness of smartphones. The magnetic field-based localization method has been an interesting research hotspot because of its temporal stability, ubiquitousness, infrastructure-free nature, and good compatibility with smartphones. However, utilizing discrete magnetic signals may result in ambiguous localization features caused by random noise and similar magnetic signals in complex symmetric and large-scale indoor environments. To address this issue, we propose a deep neural network-based fusion indoor localization system that integrates magnetic and pedestrian dead reckoning (PDR). In this system, we first propose a ResNet-GRU-LSTM neural network model to achieve magnetic localization more accurately. Afterward, we put forward a multifeatured-driven step length estimation. A hierarchy GRU (H-GRU) neural network model is proposed, and a multidimensional dataset using acceleration and a gyroscope is constructed to extract more valid characteristics. Finally, more reliable and accurate pedestrian localization can be achieved under the particle filter framework. Experiments were conducted at two trial sites with two pedestrians and four smartphones. Results demonstrate that the proposed system achieves better accuracy and robustness than other traditional localization algorithms. Moreover, the proposed system exhibits good generality and practicality in real-time localization with low cost and low computational complexity.

A Fusion Positioning Method for Indoor Geomagnetic/Light Intensity/Pedestrian Dead Reckoning Based on Dual-Layer Tent-Atom Search Optimization-Back Propagation

Indoor positioning using smartphones has garnered significant research attention. Geomagnetic and sensor data offer convenient methods for achieving this goal. However, conventional geomagnetic indoor positioning encounters several limitations, including low spatial resolution, poor accuracy, and stability issues. To address these challenges, we propose a fusion positioning approach. This approach integrates geomagnetic data, light intensity measurements, and inertial navigation data, utilizing a hierarchical optimization strategy. We employ a Tent-ASO-BP model that enhances the traditional Back Propagation (BP) algorithm through the integration of chaos mapping and Atom Search Optimization (ASO). In the offline phase, we construct a dual-resolution fingerprint database using Radial Basis Function (RBF) interpolation. This database amalgamates geomagnetic and light intensity data. The fused positioning results are obtained via the first layer of the Tent-ASO-BP model. We add a second Tent-ASO-BP layer and use an improved Pedestrian Dead Reckoning (PDR) method to derive the walking trajectory from smartphone sensors. In PDR, we apply the Biased Kalman Filter-Wavelet Transform (BKF-WT) for optimal heading estimation and set a time threshold to mitigate the effects of false peaks and valleys. The second-layer model combines geomagnetic and light intensity fusion coordinates with PDR coordinates. The experimental results demonstrate that our proposed positioning method not only effectively reduces positioning errors but also improves robustness across different application scenarios.

Step Length Is a More Reliable Measurement Than Walking Speed for Pedestrian Dead-Reckoning

Pedestrian dead reckoning (PDR) relies on the estimation of the length of each step taken by the walker in a path from inertial data (e.g. as recorded by a smartphone). Existing algorithms either estimate step lengths directly, or predict walking speed, which can then be integrated over a step period to obtain step length. We present an analysis, using a common architecture formed by an LSTM followed by four fully connected layers, of the quality of reconstruction when predicting step length vs. walking speed. Our experiments, conducted on a data set collected by twelve participants, strongly suggest that step length can be predicted more reliably than average walking speed over each step.

Accurate Stride-Length Estimation Based on LT-StrideNet for Pedestrian Dead Reckoning Using a Shank-Mounted Sensor

Pedestrian dead reckoning (PDR) is a self-contained positioning technology and has been a significant research topic in recent years. Pedestrian-stride-length estimation is the core part of the PDR system and directly affects the performance of the PDR. The current stride-length-estimation method is difficult to adapt to changes in pedestrian walking speed, which leads to a rapid increase in the error of the PDR. In this paper, a new deep-learning model based on long short-term memory (LSTM) and Transformer, LT-StrideNet, is proposed to estimate pedestrian-stride length. Next, a shank-mounted PDR framework is built based on the proposed stride-length-estimation method. In the PDR framework, the detection of pedestrian stride is achieved by peak detection with a dynamic threshold. An extended Kalman filter (EKF) model is adopted to fuse the gyroscope, accelerometer, and magnetometer. The experimental results show that the proposed stride-length-estimation method can effectively adapt to changes in pedestrian walking speed, and our PDR framework has excellent positioning performance.

A Context-Aware Smartphone-Based 3D Indoor Positioning Using Pedestrian Dead Reckoning

The rise in location-based service (LBS) applications has increased the need for indoor positioning. Various methods are available for indoor positioning, among which pedestrian dead reckoning (PDR) requires no infrastructure. However, with this method, cumulative error increases over time. Moreover, the robustness of the PDR positioning depends on different pedestrian activities, walking speeds and pedestrian characteristics. This paper proposes the adaptive PDR method to overcome these problems by recognizing various phone-carrying modes, including texting, calling and swinging, as well as different pedestrian activities, including ascending and descending stairs and walking. Different walking speeds are also distinguished. By detecting changes in speed during walking, PDR positioning remains accurate and robust despite speed variations. Each motion state is also studied separately based on gender. Using the proposed classification approach consisting of SVM and DTree algorithms, different motion states and walking speeds are identified with an overall accuracy of 97.03% for women and 97.67% for men. The step detection and step length estimation model parameters are also adjusted based on each walking speed, gender and motion state. The relative error values of distance estimation of the proposed method for texting, calling and swinging are 0.87%, 0.66% and 0.92% for women and 1.14%, 0.92% and 0.76% for men, respectively. Accelerometer, gyroscope and magnetometer data are integrated with a GDA filter for heading estimation. Furthermore, pressure sensor measurements are used to detect surface transmission between different floors of a building. Finally, for three phone-carrying modes, including texting, calling and swinging, the mean absolute positioning errors of the proposed method on a trajectory of 159.2 m in a multi-story building are, respectively, 1.28 m, 0.98 m and 1.29 m for women and 1.26 m, 1.17 m and 1.25 m for men.

Step Length Estimation Using the RSSI Method in Walking and Jogging Scenarios

In this paper, human step length was estimated based on wireless channel properties and the received signal strength indicator (RSSI) method. Path loss between two ankles of the person under test was converted from the RSSI, which was measured using our developed wearable transceivers with embedded micro-controllers in four scenarios, namely indoor walking, outdoor walking, indoor jogging, and outdoor jogging. For brevity, we call it on-ankle path loss. The histogram of the on-ankle path loss showed clearly that there were two humps, where the second hump was closely related to the maximum path loss, which, in turn, corresponded to the step length. This histogram can be well approximated by a two-term Gaussian fitting curve model. Based on the histogram of the experimental data and the two-term Gaussian fitting curve, we propose a novel filtering technique to filter out the path loss outliers, which helps set up the upper and lower thresholds of the path loss values used for the step length estimation. In particular, the upper threshold was found to be on the right side of the second Gaussian hump, and its value was a function of the mean value and the standard deviation of the second Gaussian hump. Meanwhile, the lower threshold lied on the left side of the second hump and was determined at the point where the survival rate of the measured data fell to 0.68, i.e., the cumulative distribution function (CDF) approached 0.32. The experimental data showed that the proposed filtering technique resulted in high accuracy in step length estimation with errors of only 10.15 mm for the indoor walking, 4.40 mm for the indoor jogging, 4.81 mm for the outdoor walking, and 10.84 mm for the outdoor jogging scenarios, respectively.

Transductive Learning Models for Accurate Ambulatory Gait Analysis in Elderly Residents of Assisted Living Facilities

Instrumented footwear represents a promising technology for spatiotemporal gait analysis in out-of-the-lab conditions. However, moderate accuracy impacts this technology's ability to capture subtle, but clinically meaningful, changes in gait patterns that may indicate adverse outcomes or underlying neurological conditions. This limitation hampers the use of instrumented footwear to aid functional assessments and clinical decision making. This paper introduces new transductive-learning inference models that substantially reduce measurement errors relative to conventional data processing techniques, without requiring subject-specific labelled data. The proposed models use subject-optimized input features and hyperparameters to adjust the spatiotemporal gait metrics (i.e., stride time, length, and velocity, swing time, and double support time) obtained with conventional techniques, resulting in computationally simpler models compared to end-to-end machine learning approaches. Model validity and reliability were evaluated against a gold-standard electronic walkway during a clinical gait performance test (6-minute walk test) administered to N=95 senior residents of assisted living facilities with diverse levels of gait and balance impairments. Average reductions in absolute errors relative to conventional techniques were -42.0% and -33.5% for spatial and gait-phase parameters, respectively, indicating the potential of transductive learning models for improving the accuracy of instrumented footwear for ambulatory gait analysis.

Model-Based Step Length Estimation Using a Pendant-Integrated Mobility Sensor

The step length is an important parameter in gait analysis. Long-term monitoring applications for gait analysis are often based on inertial measurement units (IMUs) due to their low-cost and unobtrusive nature. Spatial gait parameters, such as step or stride length, are therefore not directly accessible. In this contribution, we focus on model-based algorithms for step length estimation based on a pendant-integrated IMU during slow walking speeds. We present a model-based approach to estimate the step length, which is divided into two successive steps. As the first part of our approach, we present an algorithm for estimation of the vertical displacement of the center of mass (CoM) during gait. Based on this estimate, we present a novel approach to estimate the step length, which we have deduced from a previously published, simplified gait model. The algorithm is compared to a commonly known approach for accelometry-based step length prediction and validated against reference data obtained from a force plate-integrated treadmill for gait analysis during a clinical study with ten healthy subjects. Due to the applicability to gait stability assessment in elderly or gait impaired patients, we focus on slow walking speeds (1-4 km h^-1). The presented algorithms outperform the existing approach and the proposed model calculations provide a more accurate prediction. For the vertical displacement, we achieved a precision of 9.3% (CoV) with an RMSE of 1.5 mm in terms of the trajectory amplitude during normal gait patterns. The step length estimation yields satisfying results with a relative prediction error of lower than 10% for walking speeds of 2-4kmh^-1.

A Deep Learning Approach for Foot Trajectory Estimation in Gait Analysis Using Inertial Sensors

Gait performance is an important marker of motor and cognitive decline in older adults. An instrumented gait analysis resorting to inertial sensors allows the complete evaluation of spatiotemporal gait parameters, offering an alternative to laboratory-based assessments. To estimate gait parameters, foot trajectories are typically obtained by integrating acceleration two times. However, to deal with cumulative integration errors, additional error handling strategies are required. In this study, we propose an alternative approach based on a deep recurrent neural network to estimate heel and toe trajectories. We propose a coordinate frame transformation for stride trajectories that eliminates the dependency from previous strides and external inputs. Predicted trajectories are used to estimate an extensive set of spatiotemporal gait parameters. We evaluate the results in a dataset comprising foot-worn inertial sensor data acquired from a group of young adults, using an optical motion capture system as a reference. Heel and toe trajectories are predicted with low errors, in line with reference trajectories. A good agreement is also achieved between the reference and estimated gait parameters, in particular when turning strides are excluded from the analysis. The performance of the method is shown to be robust to imperfect sensor-foot alignment conditions.

INIM: Inertial Images Construction with Applications to Activity Recognition

Human activity recognition aims to classify the user activity in various applications like healthcare, gesture recognition and indoor navigation. In the latter, smartphone location recognition is gaining more attention as it enhances indoor positioning accuracy. Commonly the smartphone’s inertial sensor readings are used as input to a machine learning algorithm which performs the classification. There are several approaches to tackle such a task: feature based approaches, one dimensional deep learning algorithms, and two dimensional deep learning architectures. When using deep learning approaches, feature engineering is redundant. In addition, while utilizing two-dimensional deep learning approaches enables to utilize methods from the well-established computer vision domain. In this paper, a framework for smartphone location and human activity recognition, based on the smartphone’s inertial sensors, is proposed. The contributions of this work are a novel time series encoding approach, from inertial signals to inertial images, and transfer learning from computer vision domain to the inertial sensors classification problem. Four different datasets are employed to show the benefits of using the proposed approach. In addition, as the proposed framework performs classification on inertial sensors readings, it can be applied for other classification tasks using inertial data. It can also be adopted to handle other types of sensory data collected for a classification task.

Smartphone-Based Inertial Odometry for Blind Walkers

Pedestrian tracking systems implemented in regular smartphones may provide a convenient mechanism for wayfinding and backtracking for people who are blind. However, virtually all existing studies only considered sighted participants, whose gait pattern may be different from that of blind walkers using a long cane or a dog guide. In this contribution, we present a comparative assessment of several algorithms using inertial sensors for pedestrian tracking, as applied to data from WeAllWalk, the only published inertial sensor dataset collected indoors from blind walkers. We consider two situations of interest. In the first situation, a map of the building is not available, in which case we assume that users walk in a network of corridors intersecting at 45° or 90°. We propose a new two-stage turn detector that, combined with an LSTM-based step counter, can robustly reconstruct the path traversed. We compare this with RoNIN, a state-of-the-art algorithm based on deep learning. In the second situation, a map is available, which provides a strong prior on the possible trajectories. For these situations, we experiment with particle filtering, with an additional clustering stage based on mean shift. Our results highlight the importance of training and testing inertial odometry systems for assisted navigation with data from blind walkers.

A Pedestrian Dead Reckoning Method for Head-Mounted Sensors

Pedestrian dead reckoning (PDR) plays an important role in modern life, including localisation and navigation if a Global Positioning System (GPS) is not available. Most previous PDR methods adopted foot-mounted sensors. However, humans have evolved to keep the head steady in space when the body is moving in order to stabilise the visual field. This indicates that sensors that are placed on the head might provide a more suitable alternative for real-world tracking. Emerging wearable technologies that are connected to the head also makes this a growing field of interest. Head-mounted equipment, such as glasses, are already ubiquitous in everyday life. Whilst other wearable gear, such as helmets, masks, or mouthguards, are becoming increasingly more common. Thus, an accurate PDR method that is specifically designed for head-mounted sensors is needed. It could have various applications in sports, emergency rescue, smart home, etc. In this paper, a new PDR method is introduced for head mounted sensors and compared to two established methods. The data were collected by sensors that were placed on glasses and embedded into a mouthguard. The results show that the newly proposed method outperforms the other two techniques in terms of accuracy, with the new method producing an average end-to-end error of 0.88 m and total distance error of 2.10%.

RNN-Aided Human Velocity Estimation from a Single IMU

Pedestrian Dead Reckoning (PDR) uses inertial measurement units (IMUs) and combines velocity and orientation estimates to determine a position. The estimation of the velocity is still challenging, as the integration of noisy acceleration and angular speed signals over a long period of time causes large drifts. Classic approaches to estimate the velocity optimize for specific applications, sensor positions, and types of movement and require extensive parameter tuning. Our novel hybrid filter combines a convolutional neural network (CNN) and a bidirectional recurrent neural network (BLSTM) (that extract spatial features from the sensor signals and track their temporal relationships) with a linear Kalman filter (LKF) that improves the velocity estimates. Our experiments show the robustness against different movement states and changes in orientation, even in highly dynamic situations. We compare the new architecture with conventional, machine, and deep learning methods and show that from a single non-calibrated IMU, our novel architecture outperforms the state-of-the-art in terms of velocity (≤0.16 m/s) and traveled distance (≤3 m/km). It also generalizes well to different and varying movement speeds and provides accurate and precise velocity estimates.

Adaptive Kalman filtering-based pedestrian navigation algorithm for smartphones

Pedestrian navigation with daily smart devices has become a vital issue over the past few years and the accurate heading estimation plays an essential role in it. Compared to the pedestrian dead reckoning (PDR) based solutions, this article constructs a scalable error model based on the inertial navigation system and proposes an adaptive heading estimation algorithm with a novel method of relative static magnetic field detection. To mitigate the impact of magnetic fluctuation, the proposed algorithm applies a two-way Kalman filter process. Firstly, it achieves the historical states with the optimal smoothing algorithm. Secondly, it adjusts the noise parameters adaptively to reestimate current attitudes. Different from the pedestrian dead reckoning-based solution, the error model system in this article contains more state information, which means it is more sensitive and scalable. Moreover, several experiments were conducted, and the experimental results demonstrate that the proposed heading estimation algorithm obtains better performance than previous approaches and our system outperforms the PDR system in terms of flexibility and accuracy.

Differentially Private Mobile Crowd Sensing Considering Sensing Errors

An increasingly popular class of software known as participatory sensing, or mobile crowdsensing, is a means of collecting people’s surrounding information via mobile sensing devices. To avoid potential undesired side effects of this data analysis method, such as privacy violations, considerable research has been conducted over the last decade to develop participatory sensing that looks to preserve privacy while analyzing participants’ surrounding information. To protect privacy, each participant perturbs the sensed data in his or her device, then the perturbed data is reported to the data collector. The data collector estimates the true data distribution from the reported data. As long as the data contains no sensing errors, current methods can accurately evaluate the data distribution. However, there has so far been little analysis of data that contains sensing errors. A more precise analysis that maintains privacy levels can only be achieved when a variety of sensing errors are considered.

Deep Learning-Based Human Activity Real-Time Recognition for Pedestrian Navigation

Several pedestrian navigation solutions have been proposed to date, and most of them are based on smartphones. Real-time recognition of pedestrian mode and smartphone posture is a key issue in navigation. Traditional ML (Machine Learning) classification methods have drawbacks, such as insufficient recognition accuracy and poor timing. This paper presents a real-time recognition scheme for comprehensive human activities, and this scheme combines deep learning algorithms and MEMS (Micro-Electro-Mechanical System) sensors’ measurements. In this study, we performed four main experiments, namely pedestrian motion mode recognition, smartphone posture recognition, real-time comprehensive pedestrian activity recognition, and pedestrian navigation. In the procedure of recognition, we designed and trained deep learning models using LSTM (Long Short-Term Memory) and CNN (Convolutional Neural Network) networks based on Tensorflow framework. The accuracy of traditional ML classification methods was also used for comparison. Test results show that the accuracy of motion mode recognition was improved from 89.9%, which was the highest accuracy and obtained by SVM (Support Vector Machine), to 90.74% (LSTM) and 91.92% (CNN); the accuracy of smartphone posture recognition was improved from 81.60%, which is the highest accuracy and obtained by NN (Neural Network), to 93.69% (LSTM) and 95.55% (CNN). We give a model transformation procedure based on the trained CNN network model, and then obtain the converted .tflite model, which can be run in Android devices for real-time recognition. Real-time recognition experiments were performed in multiple scenes, a recognition model trained by the CNN network was deployed in a Huawei Mate20 smartphone, and the five most used pedestrian activities were designed and verified. The overall accuracy was up to 89.39%. Overall, the improvement of recognition capability based on deep learning algorithms was significant. Therefore, the solution was helpful to recognize comprehensive pedestrian activities during navigation. On the basis of the trained model, a navigation test was performed; mean bias was reduced by more than 1.1 m. Accordingly, the positioning accuracy was improved obviously, which is meaningful to apply DL in the area of pedestrian navigation to make improvements.

Person Re-Identification Using Deep Modeling of Temporally Correlated Inertial Motion Patterns

Person re-identification (re-ID) is among the essential components that play an integral role in constituting an automated surveillance environment. Majorly, the problem is tackled using data acquired from vision sensors using appearance-based features, which are strongly dependent on visual cues such as color, texture, etc., consequently limiting the precise re-identification of an individual. To overcome such strong dependence on visual features, many researchers have tackled the re-identification problem using human gait, which is believed to be unique and provide a distinctive biometric signature that is particularly suitable for re-ID in uncontrolled environments. However, image-based gait analysis often fails to extract quality measurements of an individual’s motion patterns owing to problems related to variations in viewpoint, illumination (daylight), clothing, worn accessories, etc. To this end, in contrast to relying on image-based motion measurement, this paper demonstrates the potential to re-identify an individual using inertial measurements units (IMU) based on two common sensors, namely gyroscope and accelerometer. The experiment was carried out over data acquired using smartphones and wearable IMUs from a total of 86 randomly selected individuals including 49 males and 37 females between the ages of 17 and 72 years. The data signals were first segmented into single steps and strides, which were separately fed to train a sequential deep recurrent neural network to capture implicit arbitrary long-term temporal dependencies. The experimental setup was devised in a fashion to train the network on all the subjects using data related to half of the step and stride sequences only while the inference was performed on the remaining half for the purpose of re-identification. The obtained experimental results demonstrate the potential to reliably and accurately re-identify an individual based on one’s inertial sensor data.

Accurate Ambulatory Gait Analysis in Walking and Running Using Machine Learning Models

Wearable sensors have been proposed as alternatives to traditional laboratory equipment for low-cost and portable real-time gait analysis in unconstrained environments. However, the moderate accuracy of these systems currently limits their widespread use. In this paper, we show that support vector regression (SVR) models can be used to extract accurate estimates of fundamental gait parameters (i.e., stride length, velocity, and foot clearance), from custom-engineered instrumented insoles (SportSole) during walking and running tasks. Additionally, these learning-based models are robust to inter-subject variability, thereby making it unnecessary to collect subject-specific training data. Gait analysis was performed in N=14 healthy subjects during two separate sessions, each including 6-minute bouts of treadmill walking and running at different speeds (i.e., 85% and 115% of each subject's preferred speed). Gait metrics were simultaneously measured with the instrumented insoles and with reference laboratory equipment. SVR models yielded excellent intraclass correlation coefficients (ICC) in all the gait parameters analyzed. Percentage mean absolute errors (MAE%) in stride length, velocity, and foot clearance obtained with SVR models were 1.37%±0.49%, 1.23%±0.27%, and 2.08%±0.72% for walking, 2.59%±0.64%, 2.91%±0.85%, and 5.13%±1.52% for running, respectively. These findings provide evidence that machine learning regression is a promising new approach to improve the accuracy of wearable sensors for gait analysis.

Pedestrian Walking Distance Estimation Based on Smartphone Mode Recognition

Stride length and walking distance estimation are becoming a key aspect of many applications. One of the methods of enhancing the accuracy of pedestrian dead reckoning is to accurately estimate the stride length of pedestrians. Existing stride length estimation (SLE) algorithms present good performance in the cases of walking at normal speed and the fixed smartphone mode (handheld). The mode represents a specific state of the carried smartphone. The error of existing SLE algorithms increases in complex scenes with many mode changes. Considering that stride length estimation is very sensitive to smartphone modes, this paper focused on combining smartphone mode recognition and stride length estimation to provide an accurate walking distance estimation. We combined multiple classification models to recognize five smartphone modes (calling, handheld, pocket, armband, swing). In addition to using a combination of time-domain and frequency-domain features of smartphone built-in accelerometers and gyroscopes during the stride interval, we constructed higher-order features based on the acknowledged studies (Kim, Scarlett, and Weinberg) to model stride length using the regression model of machine learning. In the offline phase, we trained the corresponding stride length estimation model for each mode. In the online prediction stage, we called the corresponding stride length estimation model according to the smartphone mode of a pedestrian. To train and evaluate the performance of our SLE, a dataset with smartphone mode, actual stride length, and total walking distance were collected. We conducted extensive and elaborate experiments to verify the performance of the proposed algorithm and compare it with the state-of-the-art SLE algorithms. Experimental results demonstrated that the proposed walking distance estimation method achieved significant accuracy improvement over existing individual approaches when a pedestrian was walking in both indoor and outdoor complex environments with multiple mode changes.