Edgar Charry

Measuring toe clearance using a wireless inertial sensing device

Tripping and slipping falls are serious health concerns for the elderly because they incur high medical costs and can result in life threatening injuries i.e., internal bleeding. Recently there has been an increased interest in the development of portable inertial sensing devices for measuring human movement. These devices employ acceleration and angular data to obtain limb motion such as velocity and displacement through integration of acceleration values. In this paper, we investigate the performance of a wireless inertial sensing device for monitoring vertical toe clearance during walking. It has been demonstrated that variability in minimum toe clearance (MTC) which occurs during the swing phase is a sensitive falls risk predictor. Our device is composed of a tri-axis accelerometer and dual-axis gyroscope wirelessly connected through the use of Crossbow motes. The vertical toe clearance is obtained through double integration of the resultant acceleration in the vertical direction. We have performed zero base, treadmill and intersubject experiments to investigate device performance to environmental variations and compared the calculated toe clearance against measurements made by an Optotrak motion system. It was found that device outputs were approximately independent of small ambient temperature variations, had a reliable range of 20 m indoors and 50 m outdoors and a maximum transmission rate of 20 packets/s. Toe clearance measurements coincided with Optotrak measurement trends but suffered from cumulative integration errors over time. The identification, quantification and reduction of these errors are intricate but important further research issues.

A prototype wireless inertial-sensing device for measuring toe clearance

Tripping and slipping are serious health concerns for the elderly because they result in life threatening injuries i.e., fractures and high medical costs. Our recent work in detection of tripping gait patterns has demonstrated that minimum toe clearance (MTC) is a sensitive falls risk predictor. MTC measurement has previously been done in gait laboratories and on treadmills which potentially imposes controlled walking conditions. In this paper, we describe a prototype design of a wireless device for monitoring vertical toe clearance. The sensors consists of a tri-axis accelerometer and dual-axis gyroscope connected to Crossbow sensor motes for wireless data transmission. Sensor data are transmitted to a laptop and displayed on a Matlab graphic user interface (GUI). We have performed zero base and treadmill experiments to investigate sensor performance to environmental variations and compared the calculated toe clearance against measurements made by an Optotrak motion system. It was found that device outputs were approximately independent of small ambient temperature variations, had a reliable range of 20m indoors and 50m outdoors and a maximum transmission rate of 20 packets/s. Toe clearance measurements were found to follow the Optotrak measurement trend but could be improved further by dealing with double integration errors and improving data transmission rates.

Study on estimation of peak Ground Reaction Forces using tibial accelerations in running

Ground Reaction Forces (GRF) are exerted by a surface as a reaction to a person standing, walking or running on the ground. In elite and recreational sports, GRFs are measured and studied to facilitate performance improvement and enhance injury management. Although, GRFs can be measured accurately using force platforms, such a hardware can only operate in a constrained laboratory environment and hence may limit and potentially alter a subjects natural walking or running pattern. Alternatively, a system that can measure GRFs in a more natural environment with less constraints can provide valuable insights of how humans move naturally given different gait patterns, terrain conditions and shoe types. In this regard, inertial Micro-Electrical-Mechanical-Sensors (MEMS), such as accelerometers and gyroscopes, are a promising alternative to laboratory constrained data collection systems. Kinematics of various body parts, such as their accelerations and angular velocities, can be quantified by attaching these sensors at points of interest on human body. In this paper, we investigate the relationship between the vertical GRF peaks measured by an OR6 series AMTI force plate, and accelerations along the tibial axis measured by a MEMS sensor. Our measuring system consists of two low-power wireless inertial units (ViPerform), containing one tri-axis accelerometer placed on the medial tibia of each leg.We investigate the accuracy of the measured and estimated GRF peak in 3 subjects, by means of the Root Mean Square Error (RMSE). The RMSE achieved across the speeds of 6, 9, 12, 15, 18, 21km/h and sprinting were 157N and 151N for subject 1, 106N and 153N for subject 2, and 130N and 162N for subject 3 for the left and right legs respectively. We achieved normalized errors of 6.1%, 5.9% and 5.4% for all the subjects.

A machine learning approach to k-step look-ahead prediction of gait variables from acceleration data

This paper investigates the use of machine learning to predict a sensitive gait parameter based on acceleration information from previous gait cycles. We investigate a k-step look-ahead prediction which attempts to predict gait variable values based on acceleration information in the current gait cycle. The variable is the minimum toe clearance which has been demonstrated to be a sensitive falls risk predictor. Toe clearance data was collected under normal walking conditions and 9 features consisting of peak acceleration and their normalized occurrences times were extracted. A standard least squares estimator, a generalized regression neural network (GRNN) and a support vector regressor (SVR) were trained using 60% of the data to estimate the minimum toe clearance and the remaining 40% was used to validate the model. It was found that when the training data contained data from all subjects (inter-subject) the best GRNN model provided a root mean square error (RMSE) of 2.8mm, the best SVR had RMSE of 2.7mm while the standard least squares linear regression method obtained 3.3mm. When the training and test data consisted of different subject examples (inter-subject) data, the linear SVR demonstrated superior generalization capability (RMSE=3.3mm) compared to other competing models. Validation accuracies up to 5-step look-ahead predictions revealed robust performances for both GRNN and SVR models with no clear degradation in prediction accuracy.

Frequency analysis of inertial sensor data for measuring toe clearance

The use of sensors in studies of human movement and gait is now fast gaining attention as a promising alternative to video capture systems in gait laboratories. Inertial sensors such as accelerometers and gyroscopes have been deployed to detect acceleration and angular velocities of body landmarks, from which secondary quantities such as velocity, displacement and joint angles could be inferred. The major challenge with secondary quantity derivation has been the cumulative errors arising from the integration used. While this has often been attributed to sensor measurement errors, quantification of these errors have so far been limited. In this paper, we performed frequency analysis of accelerometer and gyroscope signals to investigate the effect of sensor errors on the frequency components in motion measurements. We employ the Fast Fourier Transform (FFT) to obtain the Fourier frequency spectra of the inertial sensor signals. Comparisons were made against the frequency spectra of the same signals obtained from the Optotrak video capture system. Our analysis revealed that sensor signals contained a large DC signal which could be filtered out using a bandpass Butterworth filter with a 1-20Hz passband. This allowed us to obtain better estimates of toe displacement data without the use of strapdown integration techniques. It was also discovered that inertial sensors possessed low frequency errors which co-existed in the frequency band of gait motion making them difficult to remove with static filters. Future work will focus on adaptive filtering methods to detect and remove these sensor errors in an effort to improve displacement measurements.

Design and validation of an ambulatory inertial system for 3-D measurements of low back movements

In studies of human movement, inertial sensors (accelerometers and gyroscopes) are gaining attention as a promising alternative to laboratory-constrained video capture systems. Kinematics of various body parts and joints can be quantified by attaching inertial sensors at points of interest and integrating the observed acceleration and angular velocity signals. It is broadly accepted that this measurement procedure is significantly influenced by cumulative errors arising from sensor noise, non-linearities, asymmetries, sensitivity variations and bias drifts. In addition, it is also known that linear acceleration superimposed to the gravity acceleration introduces errors when calculating tilt angles. Recently, newer techniques using sensor fusion methods have shown error reduction in orientation measurements, but require additional hardware and consume more energy. In this paper, we assess the accuracy of a low-power wireless inertial system (ViMove) that measures Low Back (lumbar spine) orientation in three dimensions. The system consists of two inertial units (sensor), with each sensor containing one tri-axis accelerometer and one single-axis gyroscope. We investigate the accuracy of 1D, 2D and 3D simultaneous movements by means of root mean square error (RMSE) computed in comparison with NDI Optotrak, an optical tracking system. The RMSE achieved for one dimensional movements in the Flexion, Lateral Flexion and Twist planes were 1.0°, 0.5° and 2.4° respectively, and 2.1°, 2.4° and 4.6° for three dimensional movements.

An inertial sensor system for measurements of tibia angle with applications to knee valgus/varus detection

Accurate measurement of knee motion during dynamic movements is the key to detect and highlight deficiencies in peripheral muscles and ligaments of the knee and hence to predict the risk of injury. Miniature inertial sensors are increasingly becoming a viable option for human movement measurement, given their small size, low cost and relatively good accuracy compared with traditional optical measurements. A system capable of measuring tibia angle using a shank mounted wireless inertial sensor is proposed. The system employs a simple setup with only one skin-mounted triaxial accelerometer and gyroscope module attached to the tibia segment, and an algorithm to estimate the tibia angle. The accuracy of the system was assessed by an optical tracking system (Optotrak Certus) during dynamic movements performed by three subjects by evaluating Root-Mean-Square Error (RMSE) of tibia-flexion and tibia-adduction angles over the period of motion. We achieve an RMSE of 1.6±1.1 and2.5±1.6 degrees in tibia-flexion and tibia-adduction angles, respectively. It is argued that tibia angle can be reliably used to detect valgus or varus movement of the knee and hence the proposed system provides a simple and useful assessment tool for performance enhancement and rehabilitation.

A study on band-pass filtering for calculating foot displacements from accelerometer and gyroscope sensors.

As a promising alternative to laboratory-constrained video capture systems in studies of human movement, inertial sensors (accelerometers and gyroscopes) are recently gaining popularity. Secondary quantities such as velocity, displacement and joint angles can be calculated through integration of acceleration and angular velocities. It is broadly accepted that this procedure is significantly influenced by accumulative errors due to integration, arising from sensor noise, non-linearities, asymmetries, sensitivity variations and bias drifts. In this paper, we assess the effectiveness of applying band-pass filtering to raw inertial sensor data under the assumption that sensor drift errors occur in the low frequency spectrum. The normalized correlation coefficient rho of the Fast Fourier Transform (FFT) spectra corresponding to vertical toe acceleration from inertial sensors and from a video capture system as a function of digital band-pass filter parameters is compared. The Root Mean Square Error (RMSE) of the vertical toe displacement for 30 second walking windows is calculated for 2 healthy subjects over a range of 4 walking speeds. The lowest RMSE and highest cross correlation achieved for the slowest walking speed of 2.5Km/h was 3.06cm and 0.871 respectively, and 2.96cm and 0.952 for the fastest speed of 5.5Km/h.

Filtering techniques using frequency analysis for inertial sensors in gait measurements

Inertial sensors (accelerometers and gyroscopes) have recently gained attention as a promising alternative to video capture systems in studies of human movement. Secondary quantities such as velocity and displacement are calculated through integration of acceleration and angular velocities. It is now widely accepted that this technique is greatly influenced by sensor noise, non-linear and asymmetrical sensitivity/ offset signals and bias drifts. In this paper, we compare the Root Mean Square Error (RMSE) of vertical toe acceleration obtained by these sensors and derived by double differentiation from a video capture system, i.e. Optotrak Certus, NDI. Spectral analysis was performed on both accelerations using the Fast Fourier Transform to compute the correlation coefficient ρ as a function of various static band-pass filter parameters and over a range of 3 different walking speeds. RMSE and cross correlation achieved for the slowest walking speed of 2.5Km/h was 2.92m/s2 and 0.668 respectively, and 4.38m/s2 and 0.984 for the fastest speed of 4.5Km/h.