Walking speed estimation using a shank-mounted inertial measurement unit

We studied the feasibility of estimating walking speed using a shank-mounted inertial measurement unit. Our approach took advantage of the inverted pendulum-like behavior of the stance leg during walking to identify a new method for dividing up walking into individual stride cycles and estimating the initial conditions for the direct integration of the accelerometer and gyroscope signals. To test its accuracy, we compared speed estimates to known values during walking overground and on a treadmill. The speed estimation method worked well across treadmill speeds and slopes yielding a root mean square speed estimation error of only 7%. It also worked well during overground walking with a 4% error in the estimated travel distance. This accuracy is comparable to that achieved from foot-mounted sensors, providing an alternative in sensor positioning for walking speed estimation. Shank mounted sensors may be of great benefit for estimating speed in walking with abnormal foot motion and for the embedded control of knee-mounted devices such as prostheses and energy harvesters.     

Monitoring Escape and Feeding Behaviours of Cruiser Fish by Inertial and Magnetic Sensors

A method was developed and applied for monitoring two types of fast-start locomotion (feeding and escape) of a cruiser fish, Japanese amberjacks Seriola quinqueradiata. A data logger, which incorporated a 3-axis gyroscope, a 3-axis accelerometer and a 3-axis magnetometer, was attached to the five fish. The escape, feeding and routine movements of the fish, which were triggered in tank experiments, were then recorded by the data logger and video cameras. The locomotor variables, calculated based on the high resolution measurements by the data logger (500 Hz), were investigated to accurately detect and classify the types of fast-track behaviour. The results show that fast-start locomotion can be detected with a high precision (0.97) and recall rate (0.96) from the routine movements. Two types of fast-start movements were classified with high accuracy (0.84). Accuracy was greater if the data were obtained from the data logger, which combined an accelerometer, a gyroscope and a magnetometer, than if only an accelerometer (0.80) or a gyroscope (0.66) was used.     

An inertial and magnetic sensor based technique for joint angle measurement

This paper describes the design and evaluation of a miniature kinematic sensor based three dimensional (3D) joint angle measurement technique. The technique uses a combination of rate gyroscope, accelerometer and magnetometer sensor signals. The technique enables 3D inter-segment joint angle measurement and could be of benefit in a variety of applications which require monitoring of joint angles. The technique is not dependent on a fixed reference coordinate system and thus may be suitable for use in a dynamic system such as a moving vehicle. The technique was evaluated by applying it to joint angle measurement of the ankle joint. Experimental results show that accurate measurement of ankle joint angles is achieved by the technique during a variety of lower leg exercises including walking.     

Influence of IMU position and orientation placement errors on ground reaction force estimation

Wearable inertial measurement units (IMU) have been proposed to estimate GRF outside of specialized laboratories, however the precise influence of sensor placement error on accuracy is unknown. We investigated the influence of IMU position and orientation placement errors on GRF estimation accuracy. Methods: Kinematic data from twelve healthy subjects based on marker trajectories were used to simulate 1848 combinations of sensor position placement errors (range ± 100 mm) and orientation placement errors (range ± 25°) across eight body segments (trunk, pelvis, left/right thighs, left/right shanks, and left/right feet) during normal walking trials for baseline cases when a single sensor was misplaced and for the extreme cases when all sensors were simultaneously misplaced. Three machine learning algorithms were used to estimate GRF for each placement error condition and compared with the no placement error condition to evaluate performance. Results: Position placement errors for a single misplaced IMU reduced vertical GRF (VGRF), medio-lateral GRF (MLGRF), and anterior-posterior GRF (APGRF) estimation accuracy by up to 1.1%, 2.0%, and 0.9%, respectively and for all eight simultaneously misplaced IMUs by up to 4.9%, 6.0%, and 4.3%, respectively. Orientation placement errors for a single misplaced IMU reduced VGRF, MLGRF, and APGRF estimation accuracy by up to 4.8%, 7.3%, and 1.5%, respectively and for all eight simultaneously misplaced IMUs by up to 20.8%, 23.4%, and 12.3%, respectively. Conclusion: IMU sensor misplacement, particularly orientation placement errors, can significantly reduce GRF estimation accuracy and thus measures should be taken to account for placement errors in implementations of GRF estimation via wearable IMUs.     

Reconstructing muscle activation during normal walking: a comparison of symbolic and connectionist machine learning techniques

One symbolic (rule-based inductive learning) and one connectionist (neural network) machine learning technique were used to reconstruct muscle activation patterns from kinematic data measured during normal human walking at several speeds. The activation patterns (or desired outputs) consisted of surface electromyographic (EMG) signals from the semitendinosus and vastus medialis muscles. The inputs consisted of flexion and extension angles measured at the hip and knee of the ipsilateral leg, their first and second derivatives, and bilateral foot contact information. The training set consisted of data from six trials, at two different speeds. The testing set consisted of data from two additional trials (one at each speed), which were not in the training set. It was possible to reconstruct the muscular activation at both speeds using both techniques. Timing of the reconstructed signals was accurate. The integrated value of the activation bursts was less accurate. The neural network gave a continuous output, whereas the rule-based inductive learning rule tree gave a quantised activation level. The advantage of rule-based inductive learning was that the rules used were both explicit and comprehensible, whilst the rules used by the neural network were implicit within its structure and not easily comprehended. The neural network was able to reconstruct the activation patterns of both muscles from one network, whereas two separate rule sets were needed for the rule-based technique. It is concluded that machine learning techniques, in comparison to explicit inverse muscular skeletal models, show good promise in modelling nearly cyclic movements such as locomotion at varying walking speeds. However, they do not provide insight into the biomechanics of the system, because they are not based on the biomechanical structure of the system.     

Using input minimization to train a cerebellar model to simulate regulation of smooth pursuit

Cerebellar learning appears to be driven by motor error, but whether or not error signals are provided by climbing fibers (CFs) remains a matter of controversy. Here we show that a model of the cerebellum can be trained to simulate the regulation of smooth pursuit eye movements by minimizing its inputs from parallel fibers (PFs), which carry various signals including error and efference copy. The CF spikes act as “learn now” signals. The model can be trained to simulate the regulation of smooth pursuit of visual objects following circular or complex trajectories and provides insight into how Purkinje cells might encode pursuit parameters. In minimizing both error and efference copy, the model demonstrates how cerebellar learning through PF input minimization (InMin) can make movements more accurate and more efficient. An experimental test is derived that would distinguish InMin from other models of cerebellar learning which assume that CFs carry error signals.     

Assessment of the measurement accuracy of inertial sensors during different tasks of daily living

The low cost and ease of use of inertial measurement units (IMUs) make them an attractive option for motion analysis tasks that cannot be easily measured in a laboratory. To date, only a limited amount of research has been conducted comparing commercial IMU systems to optoelectronic systems, the gold standard, for everyday tasks like stair climbing and inclined walking. In this paper, the 3D joint angles of the lower limbs are determined using both an IMU system and an optoelectronic system for twelve participants during stair ascent and descent, and inclined, declined and level walking. Three different datasets based on different hardware and anatomical models were collected for the same movement in an effort to determine the cause and quantify the errors involved with the analysis. Firstly, to calculate software errors, two different anatomical models were compared for one hardware system. Secondly, to calculate hardware errors, results were compared between two different measurement systems using the same anatomical model. Finally, the overall error between both systems with their native anatomical models was calculated. Statistical analysis was performed using statistical parametric mapping. When both systems were evaluated based on the same anatomical model, the number of trials with significant differences decreased markedly. Thus, the differences in joint angle measurement can mainly be attributed to the variability in the anatomical models used for calculations and not to the IMU hardware.     

IMU: Inertial sensing of vertical CoM movement

The purpose of this study was to use a quaternion rotation matrix in combination with an integration approach to transform translatory accelerations of the centre of mass (CoM) from an inertial measurement unit (IMU) during walking, from the object system onto the global frame. Second, this paper utilises double integration to determine the relative change in position of the CoM from the vertical acceleration data. Five participants were tested in which an IMU, consisting of accelerometers, gyroscopes and magnetometers was attached on the lower spine estimated centre of mass. Participants were asked to walk three times through a calibrated volume at their self-selected walking speed. Synchronized data were collected by an IMU and an optical motion capture system (OMCS); both measured at 100 Hz. Accelerations of the IMU were transposed onto the global frame using a quaternion rotation matrix. Translatory acceleration, speed and relative change in position from the IMU were compared with the derived data from the OMCS. Peak acceleration in vertical axis showed no significant difference (p?0.05). Difference between peak and trough speed showed significant difference (p<0.05) but relative peak-trough position between the IMU and OMCS did not show any significant difference (p?0.05). These results indicate that quaternions, in combination with Simpsons rule integration, can be used in transforming translatory acceleration from the object frame to the global frame and therefore obtain relative change in position, thus offering a solution for using accelerometers in accurate global frame kinematic gait analyses.     

IMU-based ambulatory walking speed estimation in constrained treadmill and overground walking

This study evaluated the performance of a walking speed estimation system based on using an inertial measurement unit (IMU), a combination of accelerometers and gyroscopes. The walking speed estimation algorithm segments the walking sequence into individual stride cycles (two steps) based on the inverted pendulum-like behaviour of the stance leg during walking and it integrates the angular velocity and linear accelerations of the shank to determine the displacement of each stride. The evaluation was performed in both treadmill and overground walking experiments with various constraints on walking speed, step length and step frequency to provide a relatively comprehensive assessment of the system. Promising results were obtained in providing accurate and consistent walking speed/step length estimation in different walking conditions. An overall percentage root mean squared error (%RMSE) of 4.2 and 4.0% was achieved in treadmill and overground walking experiments, respectively. With an increasing interest in understanding human walking biomechanics, the IMU-based ambulatory system could provide a useful walking speed/step length measurement/control tool for constrained walking studies.     

IMU-based ambulatory walking speed estimation in constrained treadmill and overground walking

This study evaluated the performance of a walking speed estimation system based on using an inertial measurement unit (IMU), a combination of accelerometers and gyroscopes. The walking speed estimation algorithm segments the walking sequence into individual stride cycles (two steps) based on the inverted pendulum-like behaviour of the stance leg during walking and it integrates the angular velocity and linear accelerations of the shank to determine the displacement of each stride. The evaluation was performed in both treadmill and overground walking experiments with various constraints on walking speed, step length and step frequency to provide a relatively comprehensive assessment of the system. Promising results were obtained in providing accurate and consistent walking speed/step length estimation in different walking conditions. An overall percentage root mean squared error (%RMSE) of 4.2 and 4.0% was achieved in treadmill and overground walking experiments, respectively. With an increasing interest in understanding human walking biomechanics, the IMU-based ambulatory system could provide a useful walking speed/step length measurement/control tool for constrained walking studies.     

Estimation of temporal parameters during sprint running using a trunk-mounted inertial measurement unit

The purpose of this study was to identify consistent features in the signals supplied by a single inertial measurement unit (IMU), or thereof derived, for the identification of foot-strike and foot-off instants of time and for the estimation of stance and stride duration during the maintenance phase of sprint running. Maximal sprint runs were performed on tartan tracks by five amateur and six elite athletes, and durations derived from the IMU data were validated using force platforms and a high-speed video camera, respectively, for the two groups. The IMU was positioned on the lower back trunk (L1 level) of each athlete. The magnitudes of the acceleration and angular velocity vectors measured by the IMU, as well as their wavelet-mediated first and second derivatives were computed, and features related to foot-strike and foot-off events sought. No consistent features were found on the acceleration signal or on its first and second derivatives. Conversely, the foot-strike and foot-off events could be identified from features exhibited by the second derivative of the angular velocity magnitude. An average absolute difference of 0.005 s was found between IMU and reference estimates, for both stance and stride duration and for both amateur and elite athletes. The 95% limits of agreement of this difference were less than 0.025 s. The results proved that a single, trunk-mounted IMU is suitable to estimate stance and stride duration during sprint running, providing the opportunity to collect information in the field, without constraining or limiting athletes' and coaches' activities.     

Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment

An evaluation method that includes continuous activities in a daily-living environment was developed for Wearable Mobility Monitoring Systems (WMMS) that attempt to recognize user activities. Participants performed a pre-determined set of daily living actions within a continuous test circuit that included mobility activities (walking, standing, sitting, lying, ascending/descending stairs), daily living tasks (combing hair, brushing teeth, preparing food, eating, washing dishes), and subtle environment changes (opening doors, using an elevator, walking on inclines, traversing staircase landings, walking outdoors). To evaluate WMMS performance on this circuit, fifteen able-bodied participants completed the tasks while wearing a smartphone at their right front pelvis. The WMMS application used smartphone accelerometer and gyroscope signals to classify activity states. A gold standard comparison data set was created by video-recording each trial and manually logging activity onset times. Gold standard and WMMS data were analyzed offline. Three classification sets were calculated for each circuit: (i) mobility or immobility, ii) sit, stand, lie, or walking, and (iii) sit, stand, lie, walking, climbing stairs, or small standing movement. Sensitivities, specificities, and F-Scores for activity categorization and changes-of-state were calculated.The mobile versus immobile classification set had a sensitivity of 86.30% ± 7.2% and specificity of 98.96% ± 0.6%, while the second prediction set had a sensitivity of 88.35% ± 7.80% and specificity of 98.51% ± 0.62%. For the third classification set, sensitivity was 84.92% ± 6.38% and specificity was 98.17 ± 0.62. F1 scores for the first, second and third classification sets were 86.17 ± 6.3, 80.19 ± 6.36, and 78.42 ± 5.96, respectively. This demonstrates that WMMS performance depends on the evaluation protocol in addition to the algorithms. The demonstrated protocol can be used and tailored for evaluating human activity recognition systems in rehabilitation medicine where mobility monitoring may be beneficial in clinical decision-making.     

Analysis of pelvic movement in the elderly during walking using a posture monitoring system equipped with a triaxial accelerometer and a gyroscope

The incidence of falls in the elderly is increasing with the aging of society and is becoming a major public health issue. From the viewpoint of prevention of falls, it is important to evaluate the stability of the gait in the elderly people. The pelvic movement, which is a critical factor for walking stability, was analyzed using a posture monitoring system equipped with a triaxial accelerometer and a gyroscope. The subjects were 95 elderly people over 60 years of age. The criteria for instability were open-eye standing on one leg for 15s or less, and 11s or more on 3m timed up and go test. Forty subjects who did not meet both of these criteria comprised the stable group, and the remaining 55 subjects comprised the unstable group. Pelvic movement during walking was compared between the two groups. The angle, angular velocity, and acceleration were analyzed based on the wave shape derived from the device worn around the second sacral. The results indicated that pelvic movement was lower in all three directions in the unstable group compared to the stable group, and the changes in the pelvic movement during walking in unstable elderly people were also reduced. This report is the first to evaluate pelvic movement by both a triaxial accelerometer and a triaxial gyroscope simultaneously. The characteristics of pelvic movement during walking can be applied in screening to identify elderly people with instability, which is the main risk factor associated with falls.     

Feature Selection for Wearable Smartphone-Based Human Activity Recognition with Able bodied,Elderly, and Stroke Patients

Human activity recognition (HAR), using wearable sensors, is a growing area with the potential to provide valuable information on patient mobility to rehabilitation specialists. Smartphones with accelerometer and gyroscope sensors are a convenient, minimally invasive, and low cost approach for mobility monitoring. HAR systems typically pre-process raw signals, segment the signals, and then extract features to be used in a classifier. Feature selection is a crucial step in the process to reduce potentially large data dimensionality and provide viable parameters to enable activity classification. Most HAR systems are customized to an individual research group, including a unique data set, classes, algorithms, and signal features. These data sets are obtained predominantly from able-bodied participants. In this paper, smartphone accelerometer and gyroscope sensor data were collected from populations that can benefit from human activity recognition: able-bodied, elderly, and stroke patients. Data from a consecutive sequence of 41 mobility tasks (18 different tasks) were collected for a total of 44 participants. Seventy-six signal features were calculated and subsets of these features were selected using three filter-based, classifier-independent, feature selection methods (Relief-F, Correlation-based Feature Selection, Fast Correlation Based Filter). The feature subsets were then evaluated using three generic classifiers (Naïve Bayes, Support Vector Machine, j48 Decision Tree). Common features were identified for all three populations, although the stroke population subset had some differences from both able-bodied and elderly sets. Evaluation with the three classifiers showed that the feature subsets produced similar or better accuracies than classification with the entire feature set. Therefore, since these feature subsets are classifier-independent, they should be useful for developing and improving HAR systems across and within populations.     

Classification of broiler behaviours using triaxial accelerometer and machine learning

Understanding broiler behaviours provides important implications for animal well-being and farm management. The objectives of this study were to classify specific broiler behaviours by analysing data from wearable accelerometers using two machine learning models, K-Nearest Neighbour (KNN) and Support Vector Machine (SVM). Lightweight triaxial accelerometers were used to record accelerations of nine 7-week-old broilers at a sampling frequency of 40 Hz. A total of 261.6-min data were labelled for four behaviours – walking, resting, feeding and drinking. Instantaneous motion features including magnitude area, vector magnitude, movement variation, energy, and entropy were extracted and stored in a dataset which was then segmented by one of the six window lengths (1, 3, 5, 7, 10 and 20 s) with 50% overlap between consecutive windows. The mean, variation, SD, minimum and maximum of each instantaneous motion feature and two-way correlations of acceleration data were calculated within each window, yielding a total of 43 statistic features for training and testing of machine learning models. Performance of the models was evaluated using pure behaviour datasets (single behaviour type per dataset) and continuous behaviour datasets (continuous recording that involved multiple behaviour types per dataset). For pure behaviour datasets, both KNN and SVM models showed high sensitivities in classifying broiler resting (87% and 85%, respectively) and walking (99% and 99%, respectively). The accuracies of SVM were higher than KNN in differentiating feeding (88% and 75%, respectively) and drinking (83% and 62%, respectively) behaviours. Sliding window with 1-s length yielded the best performance for classifying continuous behaviour datasets. The performance of classification model generally improved as more birds were included for training. In conclusion, classification of specific broiler behaviours can be achieved by recording bird triaxial accelerations and analysing acceleration data through machine learning. Performances of different machine learning models differ in classifying specific broiler behaviours.     

Measurement of stride parameters using a wearable GPS and inertial measurement unit

Both GPS and inertial measurement units (IMUs) have been extensively used in biomechanical studies. Expensive high accuracy GPS units can provide information about intrastride speed and position, but their application is limited by their size and cost. Single and double integration of acceleration from IMU provides information about short-term fluctuations in speed and position, but suffers from integration error over a longer period of time. The integration of GPS and IMU has been widely used in large and expensive units designed for survey and vehicle navigation. Here we propose a data fusion scheme, which is a Kalman filter based complementary filter and enhances the frequency response of the GPS and IMU used alone. We also report the design of a small (28 g) low cost GPS/IMU unit. Its accuracy after post-processing with the proposed data fusion scheme for determining average speed and intrastride variation was compared to a traditional high cost survey GPS. The low cost unit achieved an accuracy of 0.15 ms⁻¹ (s.d.) for horizontal speed in cycling and human running across a speed range of 3–10 ms⁻¹. The stride frequency and vertical displacement calculated based on measurements from the low cost GPS/IMU units had an s.d. of 0.08 Hz and 0.02 m respectively, compared to measurements from high performance OEM4 GPS units.     

A wearable solution for accurate step detection based on the direct measurement of the inter-foot distance

Accurate step detection is crucial for the estimation of gait spatio-temporal parameters. Although several step detection methods based on the use of inertial measurement units (IMUs) have been successfully proposed, they may not perform adequately when the foot is dragged while walking, when walking aids are used, or when walking at low speed. The aim of this study was to test an original step-detection method, the inter-foot distance step counter (IFOD), based on the direct measurement of the distance between feet. Gait data were recorded using a wearable prototype system (SWING^2DS), which integrates an IMU and two time-of-flight distance sensors (DSs). The system was attached to the medial side of the right foot with one DS positioned close to the forefoot (FORE_DS) and the other close to the rearfoot (REAR_DS). Sixteen healthy adults were asked to walk over ground for two minutes along a loop, including both rectilinear and curvilinear portions, during two experimental sessions. The accuracy of the IFOD step counter was assessed using a stereo-photogrammetric system as gold standard. The best performance was obtained for REAR_DS with an accuracy higher than 99.8% for the instrumented foot step and 88.8% for the non-instrumented foot step during both rectilinear and curvilinear walks. Key features of the IFOD step counter are that it is possible to detect both right and left steps by instrumenting one foot only and that it does not rely on foot impact dynamics. The IFOD step counter can be combined with existing IMU-based methods for increasing step-detection accuracy.     

Estimating daily walking distance of captive African elephants using an accelerometer

Two central concerns for elephant husbandry and management are whether zoological enclosures are appropriately sized and the degree to which naturalistic exercise and activity are observed in such enclosures. In order to address these issues, accurate data on the daily walking distance of elephants both in situ and ex situ are necessary. We used an accelerometer, a pedometer that measures step count and activity level, to estimate walking distance in African elephants (Loxodonta africana) at the San Diego Zoo's Wild Animal Park. The accelerometer was worn simultaneously with a GPS unit that recorded actual walking distance. Estimates of walking distance were extrapolated from the accelerometer and compared with actual distances determined by GPS data. The accelerometer was found to overestimate step count, and subsequently walking distance, by including false counts of steps. Extrapolating walking distance based upon stride length measurements did not match actual GPS walking distance. However, activity level output from the accelerometer significantly correlated with actual GPS walking distance. In addition, we report that the rate of movement is comparable to that reported in other zoological settings. We provide a linear regression equation that can be utilized by other institutions to estimate daily walking distance of elephants in their collection who are outfitted with accelerometers.     

A Self‐Powered Dynamic Displacement Monitoring System Based on Triboelectric Accelerometer

An integrated self‐powered dynamic displacement monitoring system by utilizing a novel triboelectric accelerometer for structural health monitoring is proposed and implemented in this study, which can show the dynamic displacement and transmit the alarming signal by accurately sensing the vibration acceleration. The fabricated triboelectric accelerometer based on the noncontact freestanding triboelectric nanogenerator consists of an outer transparent sleeve tube and an inner cylindrical inertial mass that is suspended by a highly stretchable silicone fiber. One pair of copper film electrodes is deposited by physical vapor deposition on nylon film and adhered on the inner wall of the outer tube, while a fluorinated ethylene propylene film with nanowire structures is adhered on the surface of the inner cylindrical inertial mass. The experimental results show that proposed triboelectric accelerometer can accurately sense the vibration acceleration with a high sensitivity of 0.391 V s² m^?1. In particular, the developed accelerometer has superior performance within the low‐frequency range. One of the most striking features is that the commercial accelerometer using piezoelectric material is strongly dominated by high‐order harmonics, which can cause confusion in computer data analysis. In contrast, the triboelectric accelerometer is only dominated by the base resonance mode.     

Tuning bat LSO neurons to interaural intensity differences through spike-timing dependent plasticity

Bats, like other mammals, are known to use interaural intensity differences (IID) to determine azimuthal position. In the lateral superior olive (LSO) neurons have firing behaviors which vary systematically with IID. Those neurons receive excitatory inputs from the ipsilateral ear and inhibitory inputs from the contralateral one. The IID sensitivity of a LSO neuron is thought to be due to delay differences between the signals coming from both ears, differences due to different synaptic delays and to intensity-dependent delays. In this paper we model the auditory pathway until the LSO. We propose a learning scheme where inputs to LSO neurons start out numerous with different relative delays. Spike timing-dependent plasticity (STDP) is then used to prune those connections. We compare the pruned neuron responses with physiological data and analyse the relationship between IID’s of teacher stimuli and IID sensitivities of trained LSO neurons.