Scattering Features for Multimodal Gait Recognition

We consider the problem of identifying people on the basis of their walk (gait) pattern. Classical approaches to tackle this problem are based on, e.g., video recordings or piezoelectric sensors embedded in the floor. In this work, we rely on acoustic and vibration measurements, obtained from a microphone and a geophone sensor, respectively. The contribution of this work is twofold. First, we propose a feature extraction method based on an (untrained) shallow scattering network, specially tailored for the gait signals. Second, we demonstrate that fusing the two modalities improves identification in the practically relevant open set scenario.

Recently, various attempts for earthquake detection and seismic observation using low-cost vibration sensors are being actively conducted, and among them, MEMS and Geophone are the most frequently attempted sensors. MEMS and Geophone sensors have different characteristics and strengths and weaknesses, and it is necessary to understand them to select a sensor suitable for the purpose. In this study, MEMS and Geophone sensors were compared and tested for P-wave detection for earthquake early warning and ground and structure earthquake motion observation. For this purpose, the two types of sensors were mounted together on one aluminum plate, vibration table tests were conducted, and earthquake detection results were compared through real-time earthquake detection over several months. Through this, only for the tested sensor models, we could conclude that Geophone is suitable for P-wave detection and pattern extraction for noise classification, and MEMS is suitable for strong vibration measurement.

IntroductionElectromechanically-assisted gait training has been introduced in stroke rehabilitation as a means to enable gait training with a large number of reproducible and symmetrical task repetitions, i.e. steps. However, few studies have evaluated its impact on gait pattern functions. This study includes persons with no independent ambulation function at the start of a 4-week neurorehabilitation period in the sub-acute phase after stroke. The primary aim of the study was to evaluate whether the addition of electromechanically-assisted gait training to conventional training resulted in better gait pattern function than conventional training alone. The secondary aim was to identify correlations between overall gait quality and standardized clinical assessments.Participants and methodsSeventeen patients with no independent ambulation function who participated in a Prospective Randomized Open Blinded End-point study in the sub-acute phase after stroke were randomized into two groups; one group (n = 7) to undergo conventional training only (CONV group) and the other group (n = 10) to undergo conventional training with additional electromechanically-assisted gait training (HAL group). All patients were assessed with 3D gait analysis and clinical assessments after the 4-week intervention period. Overall gait quality as per the Gait Profile Score (GPS), as well as kinematic, and kinetic and other spatiotemporal metrics were collected and compared between intervention groups. Correlations between biomechanical and clinical outcomes were evaluated.ResultsBoth the CONV and HAL groups exhibited similar gait patterns with no significant differences between groups in any kinematic, kinetic parameters or other spatiotemporal metrics. The GPS for the paretic limb had a median (IQR) of 12.9° (7.8°) and 13.4° (4.3°) for the CONV and HAL groups, respectively (p = 0.887). Overall gait quality was correlated with independence in walking, walking speed, movement function and balance. We found no added benefit in gait pattern function from the electromechanically-assisted gait training compared to the conventional training alone.DiscussionThis finding raises new questions about how to best design effective and optimal post-stroke rehabilitation programs in patients with moderate to severe gait impairments to achieve both independent walking and optimal gait pattern function, and about which patients should be in focus in further studies on the efficacy of electromechanically-assisted gait training.Clinical trial registrationThe study was retrospectively registered at ClinicalTrials.gov, identifier (NCT02410915) on April 2015.

Beyond classical aspects related to locomotion (biomechanics), it has been hypothesized that walking pattern is influenced by a combination of distinct computations including online sensory/perceptual sampling and the processing of expectations (neuromechanics). Here, we aimed to explore the potential impact of contrasting scenarios (“risky and potentially dangerous” scenario; “safe and comfortable” scenario) on walking pattern in a group of healthy young adults. Firstly, and consistently with previous literature, we confirmed that the scenario influences gait pattern when it is recalled concurrently to participants’ walking activity (motor interference). More intriguingly, our main result showed that participants’ gait pattern is also influenced by the contextual scenario when it is evoked only before the start of walking activity (motor expectation). This condition was designed to test the impact of expectations (risky scenario vs. safe scenario) on gait pattern, and the stimulation that preceded walking activity served as prior. Noteworthy, we combined statistical and machine learning (Support-Vector Machine classifier) approaches to stratify distinct levels of analyses that explored the multi-facets architecture of walking. In a nutshell, our combined statistical and machine learning analyses converge in suggesting that walking before steps is not just a paradox.

Link of Video Abstract: https://youtu.be/91dLGOg7wjc Background: Treatment that can be done to prevent ulceration in diabetic feet is reducing high pressure in the foot area. One strategy that is currently being developed is biofeedback on gait patterns. Scientific evidence regarding the use of biofeedback in cases of diabetic neuropathy is still sparse, but it is a promising rehabilitation strategy to improve biomechanical deficits in gait patterns and allow subjects to initiate new gait strategies using visual and/or auditory information. Methods: This study is a pre-experimental research design before and after the intervention of giving biofeedback on walking patterns to 20 subjects with diabetic neuropathy who met the inclusion criteria and did not have exclusion criteria, measured plantar pressure at the beginning of the examination, and immediately after giving biofeedback on gait patterns. Results: The initial value of peak plantar pressure in the total area was high (more than 200 kPa). The initial value of peak plantar pressure in the 4 areas analyzed (heel, midfoot, forefoot, and big toe) was found to be in the range of 100-200 kPa, while the initial value of peak plantar pressure in the toes area was found to be <100 kPa. Immediately after administering biofeedback, the walking pattern showed a significant decrease in peak plantar pressure values in the total area, heel, and big toe compared to the initial value (p-value= 0.001, 0.012, 0.00). There was no significant decrease in peak plantar pressure values in the midfoot and toe areas. Conclusions: A significant reduction was achieved in the average plantar pressure in the heel, big toe, and overall area. This supports the recommendation to provide walking pattern biofeedback in diabetic neuropathy patients.

In this paper, we consider the problem of transparently authenticating a user to a local terminal (e.g., a desktop computer) as she approaches towards the terminal. Given its appealing usability, such zero-effort authentication has already been deployed in the real-world where a computer terminal or a vehicle can be unlocked by the mere proximity of an authentication token (e.g., a smartphone). However, existing systems based on a single authentication factor contains one major security weakness - unauthorized physical access to the token, e.g., during lunch-time or upon theft, allows the attacker to have unfettered access to the terminal. We introduce ZEMFA, a zero-effort multi-factor authentication system based on multiple authentication tokens and multi-modal behavioral biometrics. Specifically, ZEMFA utilizes two types of authentication tokens, a smartphone and a smartwatch (or a bracelet) and two types of gait patterns captured by these tokens, mid/lower body movements measured by the phone and wrist/arm movements captured by the watch. Since a user's walking or gait pattern is believed to be unique, only that user (no impostor) would be able to gain access to the terminal even when the impostor is given access to both of the authentication tokens. We present the design and implementation of ZEMFA. We demonstrate that ZEMFA offers a high degree of detection accuracy, based on multi-sensor and multi-device fusion. We also show that ZEMFA can resist active attacks that attempt to mimic a user's walking pattern, especially when multiple devices are used.

The paper presents a method of analysis of electric motors based on acoustic measurements. Acoustic signals were measured by microphones arranged in circular shape array called acoustic camera. Measurements by acoustic camera allow localization of sound source and by that separate sounds of interests from background noise can be obtained. Typically acoustic camera measurements are accompanied by a photography associated with acoustic measurement scene which is used for mapping of sound sources on the monitored area. As a reference, acoustic analysis is compared with vibration measurements. Vibration signals were measured by piezoelectric accelerometers and data collector dedicated for condition monitoring of electric motors. As presented in the result section, acoustic analysis appears as a valuable technique for condition monitoring of electric motors even in noisy industrial environment.

As humans, we constantly change our movement strategies to adapt to changes in physical functions and the external environment. We have to walk very slowly in situations with a high risk of falling, such as walking on slippery ice, carrying an overflowing cup of water, or muscle weakness owing to aging or motor deficit. However, previous studies have shown that a normal gait pattern at low speeds results in reduced efficiency and stability in comparison with those at a normal speed. Another possible strategy is to change the gait pattern from normal to step-to gait, in which the other foot is aligned with the first swing foot. However, the efficiency and stability of the step-to gait pattern at low speeds have not been investigated yet. Therefore, in this study, we compared the efficiency and stability of the normal and step-to gait patterns at intermediate, low, and very low speeds. Eleven healthy participants were asked to walk with a normal gait and step-to gait on a treadmill at five different speeds (i.e., 10, 20, 30, 40, and 60 m/min), ranging from very low to normal walking speed. The efficiency parameters (percent recovery and walk ratio) and stability parameters (center of mass lateral displacement) were analyzed from the motion capture data and then compared for the two gait patterns. The results suggested that step-to gait had a more efficient gait pattern at very low speeds of 10–30 m/min, with a larger percent recovery, and was more stable at 10–60 m/min in comparison with a normal gait. However, the efficiency of the normal gait was better than that of the step-to gait pattern at 60 m/min. Therefore, step-to gait is effective in improving gait efficiency and stability when faced with situations that force us to walk slowly or hinder quick walking because of muscle weakness owing to aging or motor deficit along with a high risk of falling.

In this paper we introduce and elaborate a biologically inspired methodology for robot walking gait pattern self-synchronization using ORCA (Organic Robot Control Architecture). The firefly based pulse coupled biological oscillator concept has been successfully applied for achieving self-organized synchronization of walking robot gait patterns by dynamically prolonging and shortening of robot’s legs stance and swing phases. The results from the experiments done on our hexapod robot demonstrator show the practical usefulness of this biologically inspired approach for run-time self-synchronizing of walking robot gait pattern parameters.Keywordsself-synchronizationorganic robot control architecturedynamically prolongation and shortening of robot’s walking gait patternsemergent robot gait synchronizationfirefly synchronizationdecentralized robot control architecturerobot gait pattern self-synchronization

In this paper, we describe the design procedure of an above-knee powered prosthetic leg and an algorithm to generate appropriate gait patterns that are synchronized with the movement of the user. The developed prosthetic leg has powered knee and ankle joints for transfemoral amputees, and its weight and dimensions were determined on the basis of human body data. In particular, two degrees of freedom (roll and pitch axes) were adopted in the ankle joint to achieve dynamic balance control on uneven ground, and passive toe joints using a crank-rocker mechanism and torsional springs were attached at the foot to increase the walking stability. In addition, we developed a walking pattern simulator that is able to test the walking patterns of the powered prosthetic leg in the air. By attaching two inertial sensors on both thighs of the user and measuring both thigh motions, the per cent of gait cycle is suitably calculated from the proposed algorithm, and smooth gait motions are generated according to the gait cycle percent. Finally, walking patterns of the powered prosthetic leg were successfully generated by synchronizing to the human gait, and the generated gaits were analyzed by comparing them to the human gait.

Renewed interest in optic motion sensors has been prompted by the potential application of such techniques to surface vibration and acoustic intensity measurements. Optical sensors intrinsically possess one highly desirable attribute, that is, no loading of the structure, and possibly several others including high resolution and wide bandwidth. Of the available optic sensors, the fiber optic lever is perhaps the simplest practical approach. The characteristics of the fiber optic lever are such that the principal parameters, i.e, linear range, resolution, bandwidth, and working distance can be optimized to meet the constraints imposed by most surface vibration problems of interest. The key strategy in optimizing the lever is to obtain sufficient linear range to encompass low‐frequency, high‐amplitude vibrations while retaining sufficient resolution for small amplitude high‐frequency motions. When sufficient reflected light is available to meet these requirements, wide bandwidth and excellent phase characteristics can be achieved by presently available electro‐optic devices. The potential for a device permitting hand held “sweeps” of complicated noise radiators exists but is critically dependent on achieving extended linear range. The extent to which these requirements have been achieved in instruments optimized for acoustic intensity measurements close to the vibrating surface is described.

Gait analysis has a significant role in assessing human's walking pattern. It is generally used in sports science for understanding body mechanics, and it is also used to monitor patients' neuro-disorder related gait abnormalities. Traditional marker-based systems are well known for tracking gait parameters for gait analysis, however, it requires long set up time therefore very difficult to be applied in everyday realtime monitoring. Nowadays, there is ever growing of interest in developing portable devices and their supporting software with novel algorithms for gait pattern analysis. The aim of this research is to investigate the possibilities of novel gait pattern detection algorithms for accelerometer-based sensors. In particular, we have used e-AR sensor, an ear-worn sensor which registers body motion via its embedded 3-D accelerom-eter. Gait data was given semantic annotation using pressure mat as well as real-time video recording. Important time stamps within a gait cycle, which are essential for extracting meaningful gait parameters, were identified. Furthermore, advanced signal processing algorithm was applied to perform automatic feature extraction by signal decomposition and reconstruction. Analysis on real-word data has demonstrated the potential for an accelerometer-based sensor system and its ability to extract of meaningful gait parameters.

This paper describes experiments to understand how well individuals can recognize an impaired walking pattern. The gait patterns are generated using a passive dynamic walker (PDW) model to allow a systematic change in gait patterns. The changed gait parameters include gait cadence, knee height asymmetry, step length and step time asymmetry, roll-over-shape (ROS) asymmetry, and knee damping asymmetry. Twenty participants rated twenty-four unimpaired and impaired walking patterns on a 7-point Likert scale. Results revealed that although a walking pattern may deviate dynamically from normal, there is a quantifiable range of gait impairments that may be dismissed as unimpaired. Particularly, we found that a knee height asymmetry range of roughly +10% (up) and −20%, could be dismissed as unimpaired. Findings show that the impairment perception of a damped knee joint can be countered by attaching a mass to the opposite lower limb.

Increasing occupant localization precision through identification of footstep-contact dynamics

Walking is crucial for active and healthy ageing, but changes with age and in the presence of chronic conditions. In the field of chronic obstructive pulmonary disease (COPD) – a highly prevalent chronic condition at older ages - extensive research has been done on physical activity limitation, but little is known about the patterns of walking during daily life (i.e., real-world). We aimed to assess the levels and distribution of real-world walking in people with COPD compared with healthy adults, and across disease severity; and to identify domains of walking in COPD, using digital technology. We studied cross-sectionally 550 older adults with COPD and 19 healthy older adults from ten cities across eight countries. We derived twenty-five digital mobility outcomes (DMOs) that assess different characteristics of walking activity and gait (e.g., walking duration, number of walking bouts of different durations, walking speed, cadence, stride length variability) from a single device worn at subjects’ lower back for one week. Participants also answered validated questionnaires, and conducted lung function and exercise capacity tests. Most DMOs were reduced in older adults with COPD compared to healthy adults, and according to COPD severity. Differences remained statistically significant after adjusting for age, sex, height and comorbidities in multivariable analyses. Factor analysis identified six distinct walking domains: Amount of walking, Patterns of walking, Pace, Rhythm, Pace variability and Rhythm variability. Real-world walking in older adults with COPD can be characterised in multiple dimensions, is impaired in relation to their healthy peers, and worsens as disease progresses.