UA

institutional

repository

Communities and Collections

Select a community to browse its collections.

Recent Submissions

No Thumbnail Available
Item
Ultrasound Imaging and Surface Electromyography-based Voluntary Effort Prediction and Control Strategies for Ankle Assistance
(North Carolina State University, 2021) Zhang, Qiang
The ankle joint plays a significant role in standing balance, walking, running, ascending, or descending stairs. However, impairments due to neuromuscular disorders, e.g., spinal cord injury, stroke, and multiple sclerosis, limit voluntary ankle power and severely undermine activities of daily living. These individuals usually have leftover or residual voluntary movement in their ankle muscles, which may not be enough to produce a normal gait. In these cases, powered ankle exoskeleton and functional electrical stimulation (FES) are potential interventions to assist ankle movements externally. Typically, control of these assistive devices is open-loop or trigger-based and largely ignores the residual ability. The key to restoring the normal function in the impaired ankle is to cooperatively control the assistive intervention with the user’s residual ankle effort. Therefore, incorporating physiological measurements of users’ residual ability is essential for the control of assistive devices. Mechanical sensors and surface electromyography (sEMG) signals are currently used to measure a user’s ankle joint volitional effort or motion intent. However, these sensors may not measure the ankle effort accurately due to several reasons, including frame misalignment, susceptibility to signal interference from neighboring muscles, and the inability to measure muscle contractions at deeper levels. Accurate estimation of the residual effort is critical for stable and effective control of desired ankle assistance. Therefore, a new reliable biological interface is required to predict residual ability more accurately and enable safe and therapeutically effective control of assistive devices. Recently, ultrasound (US) imaging has been proposed as a new sensing modality to visualize muscle contractions during voluntary limb movements directly. US imaging can potentially provide supplementary architectural-type information while sEMG signals provide electrical information for the same muscle contraction activity and thus to be used to predict the residual effort with much higher accuracy. However, US imaging features that are suitable in the prediction model are currently not fully developed, and closed-loop design to accommodate US imaging-derived measurements is non-existent. Furthermore, due to the high computational cost of analyzing US images, US-derived signals are unavailable in real-time or, at the most, are available at low sampling rates. Thus, its integration in real-time closed-loop control remains challenging. The first goal of this dissertation is to develop voluntary ankle effort prediction models that incorporate US imaging-derived signals and sEMG signals. The second goal of this dissertation is to design a new class of closed-loop controllers that incorporate low sampled US imaging-derived measurements signals of voluntary or FES-evoked muscle activity. For the first goal, I extracted multiple architectural and functional features from a high-dimensional US image to fuse with a low-dimensional sEMG signal. Then, I developed a modified Hill-type neuromuscular model and machine learning approaches to map neuromuscular signals (fused sEMG and US imaging features) to ankle joint kinetics and kinematics. For the second goal, I first derived an extended Kalman filter-type multi-rate observer to fuse high-sampled sEMG signal and low-sampled US imaging signal for the real-time closed-loop control of a cable-driven ankle exoskeleton. The dissertation also describes the design of a new bidirectional cable-driven ankle exoskeleton. The ankle exoskeleton assistance framework merges the multi-rate observer of sEMG and US imaging measurements with a newly derived adaptive impedance control that is based on the assist as-needed control principle. The controller is derived using a Lyapunov-based stability approach, which also guarantees its stability. Lastly, in the second goal, a sampled-data observer (SDO) is designed to transform the US imaging-derived measures of FES-evoked muscle activation, sampled at a low rate, to control FES of ankle dorsiflexion continuously. A dynamic surface control technique is derived to use the SDO transformed continuous signals and achieve the ankle joint trajectory tracking task. The combined controller and observer framework uses a Lyapunov-Krasovskii-based stability approach for its design and to guarantee the framework’s stability. The first outcome of this dissertation is the development of voluntary ankle effort prediction approaches that used fused features from sEMG and US imaging. The second outcome of this dissertation is the derivation of the US imaging-based closed-loop control frameworks for assisting the ankle joint with either an ankle exoskeleton or FES during walking. The control framework incorporates the voluntary or FES-evoked ankle torque in the control of powered ankle exoskeleton or FES. Multiple seated and treadmill walking experiments conducted on non-disabled participants validate the volitional effort prediction approaches and proposed control frameworks. These outcomes show a promising use of US imaging to predict residual ability when fused with existing sensing modalities such as sEMG and its potential clinical translation in the control of assistive devices such as FES and powered exoskeletons.
No Thumbnail Available
Item
No Thumbnail Available
Item
Shared Control of a Powered Exoskeleton and Functional Electrical Stimulation Using Iterative Learning
(Frontiers, 2021) Molazadeh, Vahidreza; Zhang, Qiang; Bao, Xuefeng; Sharma, Nitin
A hybrid exoskeleton comprising a powered exoskeleton and functional electrical stimulation (FES) is a promising technology for restoration of standing and walking functions after a neurological injury. Its shared control remains challenging due to the need to optimally distribute joint torques among FES and the powered exoskeleton while compensating for the FES-induced muscle fatigue and ensuring performance despite highly nonlinear and uncertain skeletal muscle behavior. This study develops a bi-level hierarchical control design for shared control of a powered exoskeleton and FES to overcome these challenges. A higher-level neural network–based iterative learning controller (NNILC) is derived to generate torques needed to drive the hybrid system. Then, a low-level model predictive control (MPC)-based allocation strategy optimally distributes the torque contributions between FES and the exoskeleton’s knee motors based on the muscle fatigue and recovery characteristics of a participant’s quadriceps muscles. A Lyapunov-like stability analysis proves global asymptotic tracking of state-dependent desired joint trajectories. The experimental results on four non-disabled participants validate the effectiveness of the proposed NNILC-MPC framework. The root mean square error (RMSE) of the knee joint and the hip joint was reduced by 71.96 and 74.57%, respectively, in the fourth iteration compared to the RMSE in the 1st sit-to-stand iteration.
Thumbnail Image
Item
Thumbnail Image
Item
Personalized Fusion of Ultrasound and Electromyography-derived Neuromuscular Features Increases Prediction Accuracy of Ankle Moment during Plantarflexion
(Elsevier, 2022) Zhang, Qiang; Clark, William H.; Franz, Jason R.; Sharma, Nitin
Objective: Compared to mechanical signals that are used for estimating human limb motion intention, non-invasive surface electromyography (sEMG) is a preferred signal in human-robotic systems. However, noise interference, crosstalk from adjacent muscle groups, and an inability to measure deeper muscle tissues are disadvantageous to sEMG’s reliable use. In this work, we hypothesize that a fusion between sEMG and in vivo ultrasound (US) imaging will result in more accurate detection of ankle movement intention. Methods: Nine young able-bodied participants were included to volitionally perform isometric plantarflexion tasks with different fixed-end ankle postures, while the sEMG and US imaging data of plantarflexors were synchronously collected. We created three dominant feature sets, sole sEMG feature set, sole US feature set, and sEMG-US feature fusion set, to calibrate and validate a support vector machine regression model (SVR) and a feedforward neural network model (FFNN) with labeled net moment measurements. Results: The results showed that, compared to the sole sEMG feature set, the sEMG-US fusion set reduced the average net moment prediction error by 35.7% (p<0.05), when using SVR, and by 21.5% (p<0.05), when using FFNN. In SVR, the sole US feature set reduced the prediction error by 24.9% (p<0.05) when compared to the sole sEMG feature set. In FFNN, the sEMG-US fusion set reduced the prediction error by 28.2% (p<0.05) when compared to the sole US feature set. Conclusion: These findings indicate that the combination of sEMG signals and US imaging is a superior sensing modality for predicting human plantarflexion intention and can enable future clinical rehabilitation devices.