When applying exoskeletons for walking assistance, one important consideration is to ensure that the users retain full control over the exoskeleton-provided assistance, which is quite limited in existing exoskeletons due to the absence of a suitable control system. In this article, a time-independent exoskeleton control system is developed based on a novel assistance profile generation method and an iterative force control method to enable continuous assistance adjustment. The assistance profile is formulated as a Gaussian function with a human state variable and can be updated online to adapt to different users. The proposed profile continuously self-adjusts along the movement of the user's leg, especially when users change their walking patterns. The proposed control system iteratively compensates for the force control lag and amplitude attenuation to enable precise tracking of the assistance profile during natural human walking. Experiments have been conducted using a soft exoskeleton on subjects with and without prior experience using an exoskeleton. The experimental results have shown the effectiveness of the proposed control system compared with a common time-dependent control system.

A Time-Independent Control System for Natural Human Gait Assistance With a Soft Exoskeleton

When applying exoskeletons for walking assistance, one important consideration is to ensure that the users retain full control over the exoskeleton-provided assistance, which is quite limited in existing exoskeletons due to the absence of a suitable control system. In this article, a time-independent exoskeleton control system is developed based on a novel assistance profile generation method and an iterative force control method to enable continuous assistance adjustment. The assistance profile is formulated as a Gaussian function with a human state variable and can be updated online to adapt to different users. The proposed profile continuously self-adjusts along the movement of the user's leg, especially when users change their walking patterns. The proposed control system iteratively compensates for the force control lag and amplitude attenuation to enable precise tracking of the assistance profile during natural human walking. Experiments have been conducted using a soft exoskeleton on subjects with and without prior experience using an exoskeleton. The experimental results have shown the effectiveness of the proposed control system compared with a common time-dependent control system. 

—Powered prosthetic legs must anticipate the user’s intent when switching between different locomotion modes (e.g., level walking, stair ascent/descent, ramp ascent/descent). Numerous data-driven classiﬁcation techniques have demon-strated promising results for predicting user intent, but the performance of these intent prediction models on novel subjects remains undesirable. In other domains (e.g., image classiﬁca- tion), transfer learning has improved classiﬁcation accuracy by using previously learned features from a large dataset (i.e., pre-trained models) and then transferring this learned model to a new task where a smaller dataset is available. In this paper, we develop a deep convolutional neural network with intra-subject (subject-dependent) and inter-subject (subject- independent) validations based on a human locomotion dataset. We then apply transfer learning for the subject-independent model using a small portion (10%) of the data from the left-out subject. We compare the performance of these three models. Our results indicate that the transfer learning (TL) model outperforms the subject-independent (IND) model and is comparable to the subject-dependent (DEP) model (DEP Error: 0.74 ± 0.002%, IND Error: 11.59 ± 0.076%, TL Error: 3.57 ± 0.02% with 10% data). Moreover, as expected, transfer learning accuracy increases with the availability of more data from the left-out subject. We also evaluate the performance of the intent prediction system in various sensor conﬁgurations that may be available in a prosthetic leg application. Our results suggest that a thigh IMU on the the prosthesis is sufﬁcient to predict locomotion intent in practice.

Deep Convolutional Neural Network and Transfer Learning for Locomotion Intent Prediction

Human ambulation is typically characterized during steady-state isolated tasks (e.g., walking, running, stair ambulation). However, general human locomotion comprises continuous adaptation to the varied terrains encountered during activities of daily life. To fill an important gap in knowledge that may lead to improved therapeutic and device interventions for mobility-impaired individuals, it is vital to identify how the mechanics of individuals change as they transition between different ambulatory tasks, and as they encounter terrains of differing severity. In this work, we study lower-limb joint kinematics during the transitions between level walking and stair ascent and descent over a range of stair inclination angles. Using statistical parametric mapping, we identify where and when the kinematics of transitions are unique from the adjacent steady-state tasks. Results show unique transition kinematics primarily in the swing phase, which are sensitive to stair inclination. We also train Gaussian process regression models for each joint to predict joint angles given the gait phase, stair inclination, and ambulation context (transition type, ascent/descent), demonstrating a mathematical modeling approach that successfully incorporates terrain transitions and severity. The results of this work further our understanding of transitory human biomechanics and motivate the incorporation of transition-specific control models into mobility-assistive technology. 

/pdf/there-are-unique-kinematics-during-locomotor-transitions-2t9lmlv4.pdf

There are unique kinematics during locomotor transitions between level ground and stair ambulation that persist with increasing stair grade

Abstract Human ambulation is typically characterized during steady-state isolated tasks (e.g., walking, running, stair ambulation). However, general human locomotion comprises continuous adaptation to the varied terrains encountered during activities of daily life. To fill an important gap in knowledge that may lead to improved therapeutic and device interventions for mobility-impaired individuals, it is vital to identify how the mechanics of individuals change as they transition between different ambulatory tasks, and as they encounter terrains of differing severity. In this work, we study lower-limb joint kinematics during the transitions between level walking and stair ascent and descent over a range of stair inclination angles. Using statistical parametric mapping, we identify where and when the kinematics of transitions are unique from the adjacent steady-state tasks. Results show unique transition kinematics primarily in the swing phase, which are sensitive to stair inclination. We also train Gaussian process regression models for each joint to predict joint angles given the gait phase, stair inclination, and ambulation context (transition type, ascent/descent), demonstrating a mathematical modeling approach that successfully incorporates terrain transitions and severity. The results of this work further our understanding of transitory human biomechanics and motivate the incorporation of transition-specific control models into mobility-assistive technology. 

Although emerging powered prostheses can enable people with lower-limb amputation to walk and climb stairs over different task conditions (e.g., speeds and inclines), the control architecture typically uses a finite-state machine to switch between activity-specific controllers. Because these controllers focus on steady-state locomotion, powered prostheses abruptly switch between controllers during gait transitions rather than continuously adjusting leg biomechanics in synchrony with the users. This paper introduces a new framework for powered prosthesis control by modeling the lower-limb joint kinematics over a continuum of variable-incline walking and stair climbing, including steady-state and transitional gaits. Steady-state models for walking and stair climbing represent joint kinematics as continuous functions of gait phase, forward speed, and incline. Transition models interpolate kinematics as convex combinations of the two steady-state models, with an additional term to account for kinematics that fall outside their convex hull. The coefficients of this convex combination denote the similarity of the transitional kinematics to each steady-state mode, providing insight into how able-bodied individuals continuously transition between ambulation modes. Cross-validation demonstrates that the model predictions of untrained kinematics have errors within the range of physiological variability for all joints. Simulation results demonstrate the model’s robustness to incline estimation and mode classification errors.

S. Cheng

Papers

Modeling the Transitional Kinematics Between Variable-Incline Walking and Stair Climbing

Deep Convolutional Neural Network and Transfer Learning for Locomotion Intent Prediction