Toward goal-oriented robotic gait training: The effect of gait speed and stride length on lower extremity joint torques

Abstract

Robot-assisted gait training is becoming increasingly common to support recovery of walking function after neurological injury. How to formulate controllers capable of promoting desired features in gait, i.e. goals, is complicated by the limited understanding of the human response to robotic input. A possible method to formulate controllers for goal-oriented gait training is based on the analysis of the joint torques applied by healthy subjects to modulate such goals. The objective of this work is to understand how sagittal plane joint torque is affected by two important gait parameters: gait speed (GS) and stride length (SL). We here present the results obtained from healthy subjects walking on a treadmill at different speeds, and asked to modulate stride length via visual feedback. Via principal component analysis, we extracted the global effects of the two factors on the peak-to-peak amplitude of joint torques. Next, we used a torque pulse approximation analysis to determine optimal timing and amplitude of torque pulses that approximate the SL-specific difference in joint torque profiles measured at different values of GS. Our results show a strong effect of GS on the torque profiles in all joints considered. In contrast, SL mostly affects the torque produced at the knee joint at early and late stance, with smaller effects on the hip and ankle joints. Our analysis generated a set of torque assistance profiles that will be experimentally tested using gait training robots.

A Controlled Slip: Training Propulsion via Acceleration of the Trailing Limb

“…A ramp-down procedure was then conducted with the treadmill starting at the subjects FC-WS and decreased in increments of 0.02 m/s until the subject indicated their SS-WS had been reached. The ramp-up and ramp-down procedures to determine SS-WS were each repeated twice, and the average of the four measured velocities was taken as the subjects SS-WS [10].…”

“…1) Data Preprocessing: EMG, kinematic marker data, and gait speed data were acquired on 3 separate systems and time synced via a common force-plate data signal. VICON marker position data were fed into a standard Visual3D preprocessing pipeline, which included i) manual labelling of markers; ii) interpolation of missing marker data with a third order polynomial fit for a maximum gap size of five samples; and iii) low-pass filtering at 6 Hz with a 4th order zero-shift Butterworth filter [10]. Force-plate data were filtered with a 4th order zero-shift low-pass Butterworth filter at 25 Hz [10].…”

“…VICON marker position data were fed into a standard Visual3D preprocessing pipeline, which included i) manual labelling of markers; ii) interpolation of missing marker data with a third order polynomial fit for a maximum gap size of five samples; and iii) low-pass filtering at 6 Hz with a 4th order zero-shift Butterworth filter [10]. Force-plate data were filtered with a 4th order zero-shift low-pass Butterworth filter at 25 Hz [10]. EMG data were bandpass filtered at 20-500 Hz, rectified, and the envelope was taken via a lowpass zero-shift 4th order Butterworth filter with a 10 Hz cut off frequency [5].…”

Training propulsion: Locomotor adaptation to accelerations of the trailing limb

“…The same procedure was repeated with the treadmill beginning at 1.8 m/s and decreasing in increments of 0.03 m/s until the subject indicated their SS-WS had been reached. Each procedure was repeated three times and subjects SS-WS was calculated as the average between the six measured speed values [6].…”

“…A robotic exoskeleton that increases ankle plantarflexion torque during push off has been FS (corresponding author -fabs@udel.edu), AJF , RM, and ML are with the Human Robotics Laboratory, Department of Biomedical Engineering, University of Delaware, Newark DE, 19713 USA. shown to be metabolically advantageous during treadmill and overground walking, and is currently used during robotassisted gait training for post-stroke individuals [3], [4]. Our group is currently working on developing robotic multijoint assistance algorithms that modulate the posture of the trailing limb at push-off to increase propulsion [5], [6]. Unfortunately, wearable exoskeletons are not always feasible for gait training of stroke survivors, due to the cost of such active devices, the burden required to wear external structures, and the technical complexity of operating such devices.…”