Inertial Motion Unit Research Articles

Action detection fusing multiple Kinects and a WIMU: an application to in-home assistive technology for the elderly

We present a vision-inertial system which combines two RGB-Depth devices together with a wearable inertial movement unit in order to detect activities of the daily living. From multi-view videos, we extract dense trajectories enriched with a histogram of normals description computed from the depth cue and bag them into multi-view codebooks. During the later classification step a multi-class support vector machine with a RBF-\(\mathcal {X}^2\) kernel combines the descriptions at kernel level. In order to perform action detection from the videos, a sliding window approach is utilized. On the other hand, we extract accelerations, rotation angles, and jerk features from the inertial data collected by the wearable placed on the user’s dominant wrist. During gesture spotting, a dynamic time warping is applied and the aligning costs to a set of pre-selected gesture sub-classes are thresholded to determine possible detections. The outputs of the two modules are combined in a late-fusion fashion. The system is validated in a real-case scenario with elderly from an elder home. Learning-based fusion results improve the ones from the single modalities, demonstrating the success of such multimodal approach.

Machine learning algorithms based on signals from a single wearable inertial sensor can detect surface- and age-related differences in walking

The aim of this study was to investigate if a machine learning algorithm utilizing triaxial accelerometer, gyroscope, and magnetometer data from an inertial motion unit (IMU) could detect surface- and age-related differences in walking. Seventeen older (71.5 ± 4.2 years) and eighteen young (27.0 ± 4.7 years) healthy adults walked over flat and uneven brick surfaces wearing an inertial measurement unit (IMU) over the L5 vertebra. IMU data were binned into smaller data segments using 4-s sliding windows with 1-s step lengths. Ninety percent of the data were used as training inputs and the remaining ten percent were saved for testing. A deep learning network with long short-term memory units was used for training (fully supervised), prediction, and implementation. Four models were trained using the following inputs: all nine channels from every sensor in the IMU (fully trained model), accelerometer signals alone, gyroscope signals alone, and magnetometer signals alone. The fully trained models for surface and age outperformed all other models (area under the receiver operator curve, AUC = 0.97 and 0.96, respectively; p ≤ .045). The fully trained models for surface and age had high accuracy (96.3, 94.7%), precision (96.4, 95.2%), recall (96.3, 94.7%), and f1-score (96.3, 94.6%). These results demonstrate that processing the signals of a single IMU device with machine-learning algorithms enables the detection of surface conditions and age-group status from an individual’s walking behavior which, with further learning, may be utilized to facilitate identifying and intervening on fall risk.

Turn cranio-caudal signature assessment from inertial systems for mobility deficit identification in Parkinson’s disease patients

Turning, defined as changing direction while walking, is recognized as a very challenging task for patients with Parkinson’s disease (PD), as it may lead to fall. Optimal turning involves a cranio-caudal sequence (i.e., the head initiates the motion, followed by the trunk and the pelvis) which have been shown to be altered in PD. However, evaluation of this sequence is currently limited to biomechanical measures in the labora- tory. Evaluation of the cranio-caudal kinematic signature (CCKS) with inertial motion units (IMUs) during turning could be a promising avenue to develop markers to assess early mobility deficits in the natural living environment of patients.

Surface gas exchange determined from an aquatic eddy covariance floating platform

AbstractWe present a new approach to quantifying air‐water flux and gas transfer velocity (k600) from underwater eddy covariance (EC) of dissolved oxygen. EC fluxes were measured 35 cm below the air‐water interface using an acoustic Doppler velocimeter (ADV) coupled with a fast‐responding dissolved oxygen optode probe, integrated on a floating platform. A micro‐electro‐mechanical system (MEMS)‐based inertial motion unit integrated with the ADV enabled compensation of measured velocities for platform motion and changing sensor orientation. Deployments of 16 h and 30 h were conducted under low to moderate wind conditions in Sage Lot Pond, a small estuarine embayment. We evaluate air‐sea flux parameterizations based on wind speed, current speed, and turbulent kinetic energy dissipation rate compared to our directly measured EC flux. Total kinetic energy was linearly correlated with EC‐determined k600, indicative of a more direct relationship to near surface turbulence compared to wind and current speed based parameterizations, which were subject to biases attributable to directional differences in fetch and low current speed. Our observations, which encompassed a period of low turbulent energy, suggest that existing parameterizations are not well constrained for these conditions. This new aquatic EC technique is highly advantageous for the determination of air‐water exchange, especially in dynamic near‐shore and inland systems. The system presented here has the potential for wide applicability for lake, riverine, and open‐ocean air‐water exchange and the results can be extrapolated for use with a wide range of biogeochemically relevant gases.

Wearable Glove-Type Driver Stress Detection Using a Motion Sensor

Increased driver stress is generally recognized as one of the major factors leading to road accidents and loss of life. Even though physiological signals are reported as the most reliable means to measure driver stresses, they often require the use of unique and expensive sensors, which produce dynamic and varying readings within individuals. This paper presents a novel means to predict a driver’s stress level by evaluating the movement pattern of the steering wheel. This is accomplished by using an inertial motion unit sensor, which is placed on a glove worn by the driver. The motion sensor selected for this paper was chosen because for its low cost and the fact that it is least affected by environmental factors as compared with a physiological signal. Experiments were conducted in three different environmental scenarios. The scenarios were classified as “urban,” “highway,” and “rural,” and they were chosen to simulate contrasting stress conditions experienced by the driver. In this paper, skin conductance and driver self-reports served as a reference stress to predict the driver’s stress level. Galvanic skin response, a well-known stress indicator, was captured along the driver’s palm and the readings were transmitted to a mobile device via low energy Bluetooth for further processing. The results revealed that indirect measurement of steering wheel movement with an inertial motion sensor could obtain accuracies up to an average rate of 94.78%. This demonstrates the opportunity for inclusion of motion sensors in wireless driver assistance systems for ambulatory monitoring of stress levels.

Measurements of wave damping by a grease ice slick in Svalbard using off-the-shelf sensors and open-source electronics

ABSTRACTVersatile instruments assembled from off-the-shelf sensors and open-source electronics are used to record wave propagation and damping measured by Inertial Motion Units (IMUs) in a grease ice slick near the shore in Adventfjorden, Svalbard. Viscous attenuation of waves due to the grease ice slick is clearly visible by comparing the IMU data recorded by the different instruments. The frequency dependent spatial damping of the waves is computed by comparing the power spectral density obtained from the different IMUs. We model wave attenuation using the one-layer model of Weber from 1987. The best-fit value for the effective viscosity is ν = (0.95 ± 0.05 × 10−2)m2 s−1, and the coefficient of determination is R2 = 0.89. The mean absolute error and RMSE of the damping coefficient are 0.037 and 0.044m−1, respectively. These results provide continued support for improving instrument design for recording wave propagation in ice-covered regions, which is necessary to this area of research as many authors have underlined the need for more field data.

Measurements of Waves in Landfast Ice Using Inertial Motion Units

Inertial motion units (IMUs) are used to perform measurements of waves in water covered by landfast ice close to the ice edge in Svalbard. The effective noise level of the instruments is assessed in controlled wave tank experiments. A set of measurements collected in Tempelfjorden, Svalbard in March 2015 is presented, and the ability of the sensors to operate in the field is validated. Several characteristics of the recorded signals, including correlation between the different sensors, are analyzed. Horizontal and vertical motions are of the same order of magnitude. A clear transition in the signal properties is observed in relation with changes in incoming wave field and the development of cracks in the ice layer. We show that complex physics takes place when waves propagate in landfast ice and that the use of times series containing information on the full three-dimensional linear acceleration, rather than spectra, is required to capture the underlying phenomena.

High-Intensity Events in International Women's Team Handball Matches.

International women's team handball is a physically demanding sport and is intermittent in nature. The aim of the study was to profile high-intensity events (HIEs) in international women's team handball matches with regard to playing positions. Twenty female national-team handball players were equipped with inertial movement units (OptimEye S5, Catapult Sports, Australia) in 9 official international matches. Players were categorized in 4 different playing positions: backs, wings, pivots, and goalkeepers (GKs). PlayerLoad™, accelerations (Acc), changes of direction (CoD), decelerations (Dec), and the sum of the latter 3, HIEs, were extracted from raw-data files using the manufacturer's software. All Acc, Dec, CoD, and HIEs >2.5 m/s were included. Data were log-transformed and differences were standardized for interpretation of magnitudes and reported with effect-size statistics. Mean numbers of events were 0.7 ± 0.4 Acc/min, 2.3 ± 0.9 Dec/min, and 1.0 ± 0.4 CoD/min. Substantial differences between playing positions, ranging from small to very large, were found in the 3 parameters. Backs showed a most likely greater frequency for HIE/min (5.0 ± 1.1 HIE/min) than all other playing positions. Differences between playing positions were also apparent in PlayerLoad/min. HIEs in international women's team handball are position specific, and the overall intensity depends on the positional role within a team. Specific HIE and intensity profiles from match play provide useful information for a better understanding of the overall game demands and for each playing position.

Observations of wave dispersion and attenuation in landfast ice

AbstractObservations of wave propagation in landfast ice were obtained in Tempelfjorden, Svalbard during March 2015. Wave motion was measured near the ice edge using inertial motion units and consisted of a combination of swell from the North Atlantic and wind‐generated waves. The waves were observed to be unidirectional in the ice with comparable magnitudes in the vertical and horizontal displacements. The dispersion relation was calculated from the measured phase difference between two adjacent sensors separated by a distance of approximately 60 m. Deviations from the gravity wave dispersion relation were observed during the growth phase of the waves and were consistent with the presence of flexural waves. This period of wave growth was accompanied by significant wave attenuation in the high frequency portion of the wave spectrum which persisted for 3–5 h.

Classification of Movement of People with Parkinsons Disease Using Wearable Inertial Movement Units and Machine Learning.

In this work, inertial movement units were placed on people with Parkinsons disease (PwPD) who subsequently performed a standard test of walking endurance (six-minute walk test - 6MWT). Five devices were placed on each the limbs and small of the back. These devices captured the acceleration and rotational motion while the person walked as far as they can in six minutes. The wearable devices can objectively indicate the pattern and rhythmicity of limb and body movements. It is possible that this data, when subject to machine learning could provide additional objective measures that may support clinical observations related to the quality of movement. The aim of this work is two fold. First, to identify the most useful features of the captured signals; second, to identify the accuracy of using these features to predict the severity of PD as measured by standard clinical assessment.

Stress Events Detection of Driver by Wearable Glove System

This paper is focused to develop a wearable glove system to detect driver stress events in real time. The driver’s stress is estimated by the use of physiological signals and steering wheel motion analysis. The steering wheel motion is analyzed by driver’s hand moving characteristic. Principally, the sensors on the glove gathered the photoplethysmogram signal via fingertip, and hand motion signal via inertial motion unit. The sensor module readings are transmitted to an end terminal application via a Bluetooth low energy transmission module to compute the driver stress index. The studies are carried out in a simulated driving, which is composed of three distinct driving scenarios to study the subjects’ behaviors that correlate with stress. Twenty-eight subjects are requested to perform three different driving sessions with random scenarios generated while performing various driving maneuvers to assess the dynamic of mental workloads. The stress assessments of driving test subjects are self-reported at pre- and post-stimulus as well as observed through facial expression recorded throughout the whole experiments. Moreover, this paper also aimed to investigate the correlation of stress events with different driving tasks. Stress index is computed by a support vector machine pattern classifier with extracted features from sensors reading. Notably, stress index differences were found among three driving scenarios and driving maneuvers. Results revealed the true accuracy of stress detection is greater than 95% in average.

Robust Head Mounted Wearable Eye Tracking System for Dynamical Calibration

In this work, a new head mounted eye tracking system is presented. Based on computer vision techniques, the system integrates eye images and head movement, in real time, performing a robust gaze point tracking. Nystagmus movements due to vestibulo-ocular reflex are monitored and integrated. The system proposed here is a strongly improved version of a previous platform called HATCAM, which was robust against changes of illumination conditions. The new version, called HAT-Move, is equipped with accurate inertial motion unit to detect the head movement enabling eye gaze even in dynamical conditions. HAT-Move performance is investigated in a group of healthy subjects in both static and dynamic conditions, i.e. when head is kept still or free to move. Evaluation was performed in terms of amplitude of the angular error between the real coordinates of the fixed points and those computed by the system in two experimental setups, specifically, in laboratory settings and in a 3D virtual reality (VR) scenario. The achieved results showed that HAT-Move is able to achieve eye gaze angular error of about 1 degree along both horizontal and vertical directions

Scientific potential of a new 3D multibeam echosounder in fisheries and ecosystem research

SEAPIX is a new multibeam echosounder (MBES) with an original architecture using a steerable symmetric Mills Cross. This configuration allows to image water column and sea bottom in both athwartship and fore-and-aft direction. Furthermore, steering capability in transmit and receive allows a volume coverage of 120°×120° under ship with 1.6°×1.6° beam aperture on the antenna axis. 64 beams are acquired per ping in the frequency range of 145–155kHz using monochromatic or frequency modulated burst. Transmitted beams are stabilized in roll or pitch according with the transmitted mode and receiving beams are motion compensated using an embedded inertial motion unit. This MBES has been designed to be installed on fishing vessel and to help fishing strategies as long as to provide useable data for fish scientist. First part of this paper describes hardware main characteristics of the SEAPIX. We then describe signal processing and functionalities of SEAPIX. Finally, we illustrate first sea results on typical case.

Performance Evaluation of Smartphone Inertial Sensors Measurement for Range of Motion.

Over the years, smartphones have become tools for scientific and clinical research. They can, for instance, be used to assess range of motion and joint angle measurement. In this paper, our aim was to determine if smartphones are reliable and accurate enough for clinical motion research. This work proposes an evaluation of different smartphone sensors performance and different manufacturer algorithm performances with the comparison to the gold standard, an industrial robotic arm with an actual standard use inertial motion unit in clinical measurement, an Xsens product. Both dynamic and static protocols were used to perform these comparisons. Root Mean Square (RMS) mean values results for static protocol are under 0.3° for the different smartphones. RMS mean values results for dynamic protocol are more prone to bias induced by Euler angle representation. Statistical results prove that there are no filter effect on results for both protocols and no hardware effect. Smartphones performance can be compared to the Xsens gold standard for clinical research.

Mapping snow depth from manned aircraft on landscape scales at centimeter resolution using structure-from-motion photogrammetry

Abstract. Airborne photogrammetry is undergoing a renaissance: lower-cost equipment, more powerful software, and simplified methods have significantly lowered the barriers to entry and now allow repeat mapping of cryospheric dynamics at spatial resolutions and temporal frequencies that were previously too expensive to consider. Here we apply these advancements to the measurement of snow depth from manned aircraft. Our main airborne hardware consists of a consumer-grade digital camera directly coupled to a dual-frequency GPS; no inertial motion unit (IMU) or on-board computer is required, such that system hardware and software costs less than USD 30 000, exclusive of aircraft. The photogrammetric processing is done using a commercially available implementation of the structure from motion (SfM) algorithm. The system is simple enough that it can be operated by the pilot without additional assistance and the technique creates directly georeferenced maps without ground control, further reducing overall costs. To map snow depth, we made digital elevation models (DEMs) during snow-free and snow-covered conditions, then subtracted these to create difference DEMs (dDEMs). We assessed the accuracy (real-world geolocation) and precision (repeatability) of our DEMs through comparisons to ground control points and to time series of our own DEMs. We validated these assessments through comparisons to DEMs made by airborne lidar and by a similar photogrammetric system. We empirically determined that our DEMs have a geolocation accuracy of ±30 cm and a repeatability of ±8 cm (both 95 % confidence). We then validated our dDEMs against more than 6000 hand-probed snow depth measurements at 3 separate test areas in Alaska covering a wide-variety of terrain and snow types. These areas ranged from 5 to 40 km2 and had ground sample distances of 6 to 20 cm. We found that depths produced from the dDEMs matched probe depths with a 10 cm standard deviation, and were statistically identical at 95 % confidence. Due to the precision of this technique, other real changes on the ground such as frost heave, vegetative compaction by snow, and even footprints become sources of error in the measurement of thin snow packs (< 20 cm). The ability to directly measure such small changes over entire landscapes eliminates the need to extrapolate limited field measurements. The fact that this mapping can be done at substantially lower costs than current methods may transform the way we approach studying change in the cryosphere.

Drive Analysis Using Vehicle Dynamics and Vision-Based Lane Semantics

Naturalistic driving studies (NDSs) capture large volumes of drive data from multiple sensor modalities, which are analyzed for critical information about driver behavior and driving characteristics that lead to crashes and near crashes. One of the key steps in such studies is data reduction, which is defined as a process by which “trained employees” review segments of driving video and record a taxonomy of variables that provides information regarding the sequence of events leading to crashes. Given the volume of sensor data in NDSs, such manual analysis of the drive data can be time-consuming. In this paper, we introduce “drive analysis” as one of the first steps toward automating the process of extracting midlevel semantic information from raw sensor data. Techniques are proposed to analyze the sensor data from multiple modalities and to extract a set of 23 semantics about lane positions, vehicle localization within lanes, vehicle speed, traffic density, and road curvature. The proposed techniques are demonstrated using real-world test drives comprising over 150 000 frames of visual data, which are also accompanied by vehicle dynamics that are captured from an in-vehicle controller-area-network bus, an inertial motion unit, and a Global Positioning System.

HyperUAS—Imaging Spectroscopy from a Multirotor Unmanned Aircraft System

One of the key advantages of a low‐flying unmanned aircraft system (UAS) is its ability to acquire digital images at an ultrahigh spatial resolution of a few centimeters. Remote sensing of quantitative biochemical and biophysical characteristics of small‐sized spatially fragmented vegetation canopies requires, however, not only high spatial, but also high spectral (i.e., hyperspectral) resolution. In this paper, we describe the design, development, airborne operations, calibration, processing, and interpretation of image data collected with a new hyperspectral unmanned aircraft system (HyperUAS). HyperUAS is a remotely controlled multirotor prototype carrying onboard a lightweight pushbroom spectroradiometer coupled with a dual frequency GPS and an inertial movement unit. The prototype was built to remotely acquire imaging spectroscopy data of 324 spectral bands (162 bands in a spectrally binned mode) with bandwidths between 4 and 5 nm at an ultrahigh spatial resolution of 2–5 cm. Three field airborne experiments, conducted over agricultural crops and over natural ecosystems of Antarctic mosses, proved operability of the system in standard field conditions, but also in a remote and harsh, low‐temperature environment of East Antarctica. Experimental results demonstrate that HyperUAS is capable of delivering georeferenced maps of quantitative biochemical and biophysical variables of vegetation and of actual vegetation health state at an unprecedented spatial resolution of 5 cm.

Portable Inertial Motion Unit for Continuous Assessment of In-shoe Foot Movement

ObjectiveThe validity and reliability of a new prototype (PT) to measure tibio-calcaneal eversion/inversion, internal/external rotation, and plantar/dorsiflexion angle during running were investigated. Design. Test-retest measurements by a 3D accelerometer and gyroscope inertial motion unit (IMU) were compared to a motion capture (MC) based system both capturing in-shoe rearfoot and tibia kinematics. Background. Laboratory running tests may not reflect movement characteristics experienced during outdoor training. Lower extremity running kinematics have not been obtained by 3D IMU measurement within the shoe previously. Methods. 3D motion of two IMUs attached to the tibia and calcaneus, as well as retroreflective markers through windows in the running shoes, were determined during running. Intersegmental motion was extracted by a complementary filter fusing accelerometer data with integrated gyroscope angular velocity. PT measurements for motion along main anatomical axes were similar to MC derived curves based on coefficients of multiple correlation. Intraclass correlation correlations (ICC) showed a low correlation between PT and MC in three out of four parameters, but high repeatability for PT between test and retest measures. PT underestimated angular motion with a root mean square error (RMSe) of 182.5°/s and bias of 176.3°/s. Eversion was underestimated by PT with RMSe of 6.3°. The PT method was found inadequate to determine valid tibio-calcaneal motion while it was reliable for repeated tests and may allow for intra-subject comparisons of different footwear or inserts.

Standing balance tests for screening people with vestibular impairments

To improve the test standards for a version of the Romberg test and to determine whether measuring kinematic variables improved its utility for screening. Healthy controls and patients with benign paroxysmal positional vertigo, postoperative acoustic neuroma resection, and chronic peripheral unilateral weakness were compared. Subjects wore Bluetooth-enabled inertial motion units while standing on the floor or medium-density, compliant foam, with eyes open or closed, with head still or moving in pitch or yaw. Dependent measures were time to perform each test condition, number of head movements made, and kinematic variables. Patients and controls did not differ significantly with eyes open or with eyes closed while on the floor. With eyes closed, on foam, some significant differences were found between patients and controls, especially for subjects older than 59 years. Head movement conditions were more challenging than with the head still. Significantly fewer patients than controls could make enough head movements to obtain kinematic measures. Kinematics indicated that lateral balance control is significantly reduced in these patients compared to controls. Receiver operator characteristics and sensitivity/specificity analyses showed moderately good differences with older subjects. Tests on foam with eyes closed, with head still or moving, may be useful as part of a screening battery for vestibular impairments, especially for older people. 3b.

Sharpening the Tandem Walking Test for Screening Peripheral Neuropathy

Few tests of functional motor behavior are useful for rapidly screening people for lower extremity peripheral neuropathy. The goal of this study was to improve the widely used tandem walking (TW) test. We tested "normal" (control) adult and ambulatory patients with peripheral neuropathy (PN) with their eyes open and eyes closed while they performed TW on industrial carpeting in sock-covered feet. Each subject wore a torso-mounted inertial motion unit to measure kinematic data. The data of subjects with PN also were compared with historical data on patients with vestibular impairments. The normal and PN groups differed significantly on TW and on the number of steps completed. PN and vestibular impairments data also differed significantly on both visual conditions. Kinematic data showed that patients with PN were more unstable than normal patients in the group. For the number of steps taken during the eyes open condition, receiver operating characteristic (ROC) values were only 0.81 and for the number of steps taken during the eyes closed condition, ROC values were 0.88. Although not optimal, this ROC value is better. Sensitivity and specificity at a cutoff of two steps were 0.81 and 0.92, respectively, and at a cutoff of three steps were 0.86 and 0.75, respectively. ROC values for kinematic data were <0.8, and when combined with the ROC value for the number of steps, the total ROC value did not improve appreciably. Although not ideal for screening patients who may have PN, counting the number of steps during TW is a quick and useful clinical test. TW is most sensitive to patients with PN when they are tested with eyes closed.