Browsing by Author "Arami, Arash"

Now showing 1 - 9 of 9

Application of Machine Learning Modeling in Establishing the Process, Structure, and Property Relationships of the Cast-Forged AZ80 Magnesium Alloy
(University of Waterloo, 2024-05-27) Azqadan, Erfan; Jahed, Hamid; Arami, Arash
The cast-forging process is a novel hybrid manufacturing paradigm that leverages the cost effectiveness of cast product while alleviating its structural and durability weaknesses through forging. Therefore, the cast-forging process is a promising candidate for production of AZ80 magnesium alloy structural components with potential use in automotive and aerospace industries. In this novel method, the low formability of magnesium alloys at room temperature for near net shape forming is removed by elevated temperature forging of magnesium products. Also, the flexibility of casting in producing complex shapes is leveraged, and its low mechanical properties is enhanced through considerable deformation of AZ80 alloy offered by forging. The cast-forging manufacturing method takes advantage of possible microstructure variations induced in the material via different cast geometries and/or processing parameters. Therefore, this novel method can produce reliable lightweight magnesium structural components. Currently, there is limited knowledge of the effects of initial cast microstructure on the hot deformation behavior of AZ80 alloy. The current study aims to establish a link between casting process parameters that controls the microstructure of cast material and their effects on forging process. Due to the complexity of the relationship between process parameters, microstructure, and properties of cast-forged AZ80 magnesium alloys, advanced characterization methods and data-driven models are used to establish this link. In this work, it is shown that casting cooling rate controls the matrix grain size and the morphology and distribution of intermetallic particles formed during and after solidification. These microstructural features influence dynamic recrystallization (DRX) during the forging process that affects further formability of the material. Also, the x-ray computed tomography (XCT) analysis of cast material shows the role of casting process parameters on the formation of porosities and their effect on mechanical properties. Moreover, several different morphologies of the Mg17Al12 intermetallic compound forms during the casting and forging processes. The evolution of the Mg17Al12 intermetallic during casting, pre-forging heat treatment, and forging process occurs due to breakage, dissolution, and precipitation of this phase. Different Mg17Al12 intermetallic morphologies affect DRX phenomenon. Since the final microstructure and mechanical properties of the cast-forged component is controlled by occurrence of DRX, a detailed investigation of the interactions between the Mg17Al12 intermetallic and DRX is conducted. This study shows, as previously suggested by the literature, the eutectic, lamellar, and discontinuous morphologies of the Mg17Al12 phase promote DRX through particle stimulated nucleation (PSN) mechanism. However, in contrast to the literature, this study finds that the continuous Mg17Al12 morphology when broken as a result of severe plastic deformation can also promote DRX occurrence. The combination of casting and forging process parameters can result in a wide range of deformation behaviors and consequently varied microstructure and final mechanical properties. An informed search for optimal manufacturing route based on desired final mechanical properties requires modeling of materials evolution to accelerate research. Therefore, the application of data-driven modeling methods to establish the process-structure-property relationships of this system is studied. In this regard, an artificial neural network (ANN)-based screening tool is developed using more than 800 hardness measurements conducted on 12 different cast-forged components. This process-to-property model takes casting cooling rate and forging temperature to predict the hardness distribution of the cast-forged components. The hardness of materials correlates with several other properties such as tensile strength and resistance to deformation. This model, which requires no characterization of the material, can be used to find the most optimal combination of the process parameters that might satisfy the mechanical properties requirements. The application of this model for unseen combination of the process parameters is investigated and shows the robustness of the model as a screening tool. In order to develop a mesoscale microstructure model predicting the microstructure evolution based on process parameters, image generative machine learning models, namely generative adversarial network (GAN) and denoising diffusion probabilistic model (DDPM) are implemented. Capitalizing on the enhanced data distribution capturing characteristic of DDPM, it is utilized for the final model. This process-to-microstructure model, trained on 434 high-resolution SEM images from 27 cast-forged samples, takes parameters like the casting geometry, casting cooling rate, pre-forging heat treatment, pre-forging soaking process, forging temperature, metallography extraction location, and image magnification, to produces convincing high-resolution synthesized SEM images for seen and unseen process parameter combinations. To evaluate the predictive capabilities of this proposed approach, computer vision and morphological feature metrics are analyzed for the real and synthesized images, revealing the model’s ability to capture underlying physical relationships, such as grain size, Mg17Al12 morphology and area fraction, distribution of morphological features, and DRX percentage within cast-forged AZ80 SEM images. To the best of our knowledge, this represents the most comprehensive study of machine learning image generative models aimed at producing high-resolution microstructure images. The establishment of the relationship between process parameters, microstructure, and mechanical properties in this work aims to facilitate the search for optimum processing route for production of the AZ80 cast-forged front lower control arm (FLCA) component with superior mechanical properties compared to previous attempts and the aluminum alloy-based benchmark. In this regard, an Image-based machine learning model is also developed, based on 377 SEM images and tensile test results of 27 cast-forged components, to predict the yield strength, ultimate tensile strength, and elongation to failure of the cast-forged AZ80 alloy directly from the SEM microstructure images. The proposed process-to-microstructure and microstructure-to-property machine learning models provide an end-to-end framework to explore the possible microstructure and property spaces of this system. This framework is implemented using internally developed, custom-built Python scripts and leverages the PyTorch library.
Coordinated human-exoskeleton locomotion emerges from regulating virtual energy
(Public Library of Science (PLOS), 2025) Nasiri, Rezvan; Dinovitzer, Hannah; Manohara, Nirosh; Arami, Arash
Lower-limb exoskeletons have demonstrated great potential for gait rehabilitation in individuals with motor impairments; however, maintaining human-exoskeleton coordination remains a challenge. The coordination problem, referred to as any mismatch or asynchrony between the user's intended trajectories and exoskeleton desired trajectories, leads to sub-optimal gait performance, particularly for individuals with residual motor ability. Here, we investigate the virtual energy regulator (VER)'s ability to generate coordinated locomotion in lower limb exoskeleton. Contribution: (1) In this paper, we experimented VER on a group of nine healthy individuals at different speeds (0.6m/s-0.85m/s) to study the resultant gait coordination and naturalness on a large group of users. (2) The resultant assisted gait is compared to the natural and passive (zero-torque exoskeleton) walking conditions in terms of muscle activities, kinematic, spatiotemporal and kinetic measures, and questionnaires. (3) Moreover, we presented the VER's convergence proof considering the user contribution to the gait and introduced a metric to measure the user's contribution to gait. (4) We also compared VER performance with the phase-based path controller in terms of muscle effort reduction and joint kinematics using three able-bodied individuals. Results: (1) The results from the VER demonstrate the emergence of natural, coordinated locomotion, resulting in an average muscle effort reduction ranging from 13.1% to 17.7% at different speeds compared to passive walking. (2) The results from VER revealed improvements in all indicators towards natural gait when compared to walking with a zero-torque exoskeleton, for instance, an enhancement in average knee extension ranging from 3.9 to 4.1 degrees. All indicators suggest that the VER preserves natural gait variability and user engagement in locomotion control. (3) Using VER also yields in 13.9%, 15.1%, and 7.0% average muscle effort reduction when compared to the phase-based path controller. (4) Finally, using our proposed metric, we demonstrated that the resultant locomotion limit cycle is a linear combination of human=intended limit cycle and the VER's limit cycle. These findings may have implications for understanding how the central nervous system controls our locomotion.
Cuffless Blood Pressure Monitoring: Estimation of the Waveform and its Prediction Interval
(University of Waterloo, 2021-12-21) Landry, Cederick; Peterson, Sean D.; Arami, Arash
Cuffless blood pressure (BP) estimation devices are receiving considerable attention as tools for improving the management of hypertension, a condition that affects 1.13 billion people worldwide. It is an approach that can provide continuous BP monitoring, which is not possible with existing non-invasive tools. Therefore, it yields a more comprehensive picture of the patient’s state. Cuffless BP monitoring relies on surrogate models of BP and the information encoded in alternative physiological measures, such as photoplethysmography (PPG) or electrocardiography (ECG), to continuously estimate BP. Existing models have typically relied upon pulse-wave delay between two arterial segments or other pulse waveform features in the estimation process. However, the models available in the literature (1) provide an estimation of the systolic BP (SBP), diastolic BP (DBP), and mean BP (MAP) only, (2) are validated solely in controlled environments, and (3) do not assign a confidence metric to the estimates. At this point, cuffless methods are not used by clinicians due to their inaccuracy, the validation inadequacy, and/or the unevaluated uncertainty of the existing methods. The first objective of this thesis is to develop a cuffless modeling approach to estimate the BP waveform from ECG and PPG, and extract important BP features, such as the SBP, DBP, and MAP. Access to the full waveform has significant advantages over previous cuffless BP estimation tools in terms of accuracy and access to additional cardiovascular health markers (e.g., cardiac output), as well as potentially providing arterial stiffness. The second objective of this thesis is to validate cuffless BP estimation during activities of daily living, an uncontrolled environment, but also in more challenging physiological conditions such as during exercise. Such validation is important to increase confidence in cuffless BP monitoring, it also helps understand the limitation of the method and how they would affect clinical outcomes. Finally, in an effort to improve confidence in the cuffless BP estimation framework (third objective), a prediction interval (PI) estimation method is introduced. For potential clinical uses, it is imperative to assess the uncertainty of the BP estimate for acute outcome evaluation and it is even more so if cuffless BP is to be employed outside of the clinic. In this thesis, user-specific nonlinear autoregressive models with exogenous inputs (NARX) are implemented using an artificial neural network (ANN) to predict the BP waveforms using ECG and/or PPG signals as inputs. To validate the NARX-based BP estimation framework during activities of daily living, data were collected during six-hours testing phase wherein the participants go about their normal daily living activities. Data are further collected at four-month and six-month time points to validate long-term performance. To broaden the range of BP in the training data, subjects followed a short procedure consisting of sitting, standing, walking, Valsalva maneuvers, and static handgrip exercises. To evaluate the uncertainty of the BP estimates, one-class support vector machines (OCSVM) models are trained to cluster data in terms of the percentage of outliers. New BP estimates are then assigned to a cluster using the OCSVMs hyperplanes, and the PIs are estimated using the BP error standard deviation associated with different training data clusters. The OCSVM is used to estimate the PI for three BP model architectures: NARX models, feedforward ANN models, and pulse arrival time (PAT models). The three BP estimations from the models are fused using the covariance intersection fusion algorithm, which improves BP and PI estimates in comparison with individual model performance. The proposed method models the BP as a dynamical system leading to better accuracy in the estimation of SBP, DBP and MAP when compared to the PAT model. Moreover, the NARX model, with its ability to provide the BP waveform, yields more insight into patient health. The NARX model demonstrates superior accuracy and correlation with “ground truth” SBP and DBP measures compared to the PAT models and a clear advantage in estimating the large range of BP. Preliminary results show that the NARX models can accurately estimate BP even months apart from the training. Preliminary testing suggests that it is robust against variabilities due to sensor placement. The employed model fusion architecture establishes a method for cuffless BP estimation and its PI during activities of daily living that can be used for continuous monitoring and acute hypotension and hypertension detection. The NARX model, with its capacity to estimate a large range of BP, is next tested during moderate and heavy intensity exercise. Participants performed three cycling exercises: a ramp-incremental exercise test to exhaustion, a moderate and a heavy pseudorandom binary sequence exercise tests on an electronically braked cycle ergometer. Subject-specific and population-based NARX models are compared with feedforward ANN models and PAT (and heart rate) models. Population-based NARX models, when trained on 11 participants’ three cycling tests (tested on the participant left out of training), perform better than the other models and show good capability at estimating large changes in MAP. A limitation of the approach is the incapability of the models to track consistent decreases in BP during the exercise caused by a decrease in peripheral resistance since this information is apparently not encoded in either the forehead PPG or ECG signals. Nevertheless, the NARX model shows good precision during the whole 21 minutes testing window, a precision that is increased when using a shorter evaluation time window, and that can potentially be even further increased if trained on more data. The validation protocols and the use of a confidence metric developed in this thesis is of great value for such health monitoring application. Through such methodology, it is hoped that cuffless BP estimation becomes, one day, a well-established BP measurement method.
Enhancement of Human-Robot Physical Interaction in Lowerlimb Exoskeletons
(University of Waterloo, 2024-05-24) Shushtari, Mohammad; Arami, Arash
Many people face mobility challenges due to spinal cord injury, stroke, and aging. Therapeutic interventions using assistive exoskeletons have emerged as promising tools to enhance their quality of life. The efficacy of exoskeletons requires a delicate balance between assistance and allowing users to regain control of their movements. This needs the exoskeleton to continuously alternate between follower and leader roles to assist the user only when needed. My PhD research focused on proposing a solution for this challenge by optimizing the human-exoskeleton physical interaction. I developed an innovative method to optimize interaction torques, enabling the exoskeleton to adapt its assistance based on the user's motor capacity. Using musculoskeletal modelling and simulation tools such as OpenSim and MATLAB, I integrated human and exoskeleton models and simulated individuals with varying levels of injuries. I implemented an adaptive approach to determine an efficient exoskeleton trajectory, resulting in improved gait stability and spatiotemporal parameters by decreasing the physical disagreement between the user and the exoskeleton, which is expected to increase the user comfort level. I further extended the optimization formulation to adapt to the changes in gait speed, transitions, and pathological gait patterns by developing a data-driven gait phase estimator using a rich dataset collected from 14 participants, offering superior performance in estimating gait phases under diverse conditions. Moreover, I tackled the issue of measuring interaction torques in practice, where direct measurements are impractical due to the complex nature of human-exoskeleton interaction. I introduced an innovative excitation approach to capture the dynamics of the exoskeleton in all regimes (i.e., swing, stance, and double support) with a single model. This method allows researchers to estimate interaction torques throughout the entire gait phase, ensuring accurate monitoring of the human-exoskeleton interaction dynamics. Leveraging these contributions, I implemented my optimization method in practical settings, and validated its effectiveness in experiments on 15 participants during treadmill and overground walking. Finally, I developed an adaptive feedforward torque controller capable of learning the user desired joint trajectory and accordingly generating an appropriate feedforward torque based on the exoskeleton dynamical model. Comparative assessments on 9 participants against current methods demonstrated that my controller reduces metabolic costs, physical interaction, and enhances the overall user experience compared to a recently developed stated-dependent feedforward controller. As a part of this assessment, I proposed a new method of evaluating human-exoskeleton interaction based on co-analysis of the user muscular effort and the interaction torques called Interaction Portrait. I showed that the distribution of the interaction portrait can determine different regimes of human-exoskeleton physical interaction. In conclusion, the methodologies I introduced contributed to the advancement of assistive robotics. By focusing on optimizing interaction torques, I addressed a key limitation in contemporary exoskeleton designs, ensuring the device intelligently adapts to the user's unique motor capacities. By successfully addressing real-world challenges, such as adapting to diverse gait patterns and accurately estimating interaction torques, my research offers a tangible and significant improvement in exoskeleton performance. The practical implementation and subsequent evaluations underscore the potential of my approach to not only enhance mobility but also elevate the user experience. My research lays a strong foundation for future endeavours aimed at bridging the gap between robotic assistance and human motor impairments.
Enhancing Lower Limb Blood Flow by Optimizing Intermittent Pneumatic Compression Timing
(University of Waterloo, 2023-11-03) Santelices, Iara; Arami, Arash; Peterson, Sean
Intermittent pneumatic compression (IPC) is used in the management of vascular disorders. IPC systems apply external pressure to the circumference of the leg to enhance blood velocity (BV). Although IPC device performance can be evaluated by the induced change in mean blood velocity (a representation of the device’s ability to vacate venous blood from the legs), commercial IPC systems are often designed with the main considerations of cost and size instead of performance. Previous studies have demonstrated that the application of cardiac-gated compression further enhanced blood velocity (BV) compared to fixed compression timing (CT). However, optimal CT, as a function of the cardiac cycle, is not constant across individuals and may change over time. Additionally, IPC is often being used over the course of days and weeks to promote blood flow. Consequently, at the start of a therapy session, the optimal CT for an individual is unknown. Learning the optimal CT for an individual will maximize BV and make the therapy session more efficient. Current CT modelling methods for IPC are limited to predictions for a single day and one heartbeat ahead, which is not sufficient for typical IPC usage since typical IPC sessions last for 60 minutes and IPC is used over the course of days and weeks. In this thesis a data-driven predictive model to simulate the blood flow response to compression timing across two days was built. Furthermore, a deep reinforcement learning agent learns the CT that maximizes blood flow for each heartbeat in a custom-built simulated environment. Finally, a deep reinforcement learning agent optimizes compression timing with a human-in-the-loop. In this thesis, six participants wore a custom IPC system and experienced random cardiac-gated CTs for 1.5 hours per day for two days. Six user-specific nonlinear autoregressive models with exogenous inputs (NARX) were implemented using an artificial neural network (ANN) to estimate the BV response to a CT. The BV response to a CT is estimated using past inputs of BV, electrocardiography (ECG), and CT. The NARX model is trained on the first session. This predictive model can be used online to estimate the optimal cardiac gated CT for a therapy session. The mean R2 for this model across participants on the second session was 0.74 ± 0.09 and the mean absolute error was approximately 3%, which is a reduction of only 11% compared to the first sessions, for both metrics. This study is the first to show that BV across IPC sessions can be predicted using a pre-trained model. A deep reinforcement learning (DRL) algorithm learns the optimal compression timing to maximize its cumulative reward, mean beat-to-beat mean blood velocity (BBMV). In this work, participant-specific simulated lower limb environments were built, using the aforementioned NARX models, for six participants and show that DRL can optimize CT, across 19 CTs, in an IPC framework. We show that the DRL agent can adapt to changes in the physiological state unlike previous methods which employ fixed dynamics. The DRL agent can learn an optimal policy in 15 minutes ± 2 on average. The proposed DRL agent can be implemented in IPC systems to rapidly learn the optimal CT with a human-in-the-loop. This is particularly valuable as the ideal CT is unknown at the outset of a therapy session, and learning it maximizes the benefit of therapy. It is shown that a DRL agent can optimize IPC CT with a human-in-the-loop within a 60-minute therapy session. This study included 10 participants. To observe learning within 60 minutes, the DRL agent optimizes across two cardiac gated CTs, diastole and systole. The DRL agent state is composed of previous heart rate and BBMV information. The utilized reward function incentivizes constant increases in BBMV to achieve agent convergence. The DRL agent converged to diastole CT at each heartbeat across 10 participants, selecting diastolic compression 83% ± 9 of the time. The mean difference in blood velocity between the DRL CT and cardiac-gated diastolic CT was -0.09 cm/s ± 0.09 (-4% ± 4). This DRL agent is the first of its kind to optimize IPC CT with a human-in-the-loop.
A hybrid inverse dynamic-neural network approach to lower limb exoskeleton control
(University of Waterloo, 2022-08-22) Dinovitzer, Hannah; Arami, Arash
A powered machine that is wearable over all or part of the human body can be referred to as a powered exoskeleton. The role of powered exoskeletons is usually to provide ergonomic structural support while using motor power to synchronize to and assist with intended movements. One specific category of exoskeletons is the lower limb exoskeleton. There is a variety of applications for lower limb exoskeletons, including assistance, rehabilitation, and augmentation. A challenge in developing any of these forms of exoskeletons is the design of controllers which are able to perform well under a variety of scenarios, such as change in speed while walking or stair climbing, as well as with a variety of users. There are many different controllers that have been developed accordingly. One of these approaches involves estimating joint torques and applying these directly as control torques. This can be done in one of two ways: estimating the torques based on a few subjects and applying these prescribed torques to everyone, or estimating and applying joint torques in real-time. Many existing controllers which estimate and apply joint torques in real-time, only do so with a portion of the joint torques. For instance, this can be done by computing and applying joint torques which result from gravity only. The challenge with the estimating and applying joint torques in real-time is developing an accurate model to represent the dynamics of the system and accurately measuring all required state signals. The signals which are most problematic for measurement is ground contact force measurements. As forceplates are not useful for continuous overground measurements and instrumented insoles can be unreliable, an alternative approach is required. To fill this gap and generate a robust real-time joint torque estimator, a hybrid inverse dynamic-neural network model is proposed. In addition, a data-driven solution is proposed and comprises of an end-to-end neural network for direct joint torque estimation. The hybrid model computes joint torques with the use of kinematic information only. Eliminating the need for kinetic measurements allows ease with implementation in scenarios where forceplates are not available; this is done with a neural network for ground contact force estimation. The hybrid model was validated with 11 subjects during treadmill walking, including several different gait patterns. In comparison to the end-to-end direct torque estimator, the hybrid model has slightly worse performance at the knee and hip joints during treadmill walking which includes speed changes, asymmetrical walking, and start-stops. However, when testing these two approaches with a participant wearing an exoskeleton, the hybrid model outperforms the end-to-end network. This validates the versatility of the hybrid model to generalize to many different conditions and subjects. The hybrid model was then implemented as a controller in a lower limb exoskeleton. A second pre-defined direct torque controller was also developed. The pre-defined torques are recorded from the response of a feedback controller used on one participant. These torques are then applied as the direct torque control as a function of walking speed and gait phase. Both these controllers can be considered to be feedforward control approaches as the applied torques are not explicitly encoded by feedback errors. These controllers were tested individually and in combination for treadmill and overground walking with nine participants. A combination of the two controllers, with more contributions from the hybrid control, produces the overall best results in terms of spatiotemporal metrics. At a joint-level, all the tested controllers have similar performance in terms of range of motion and joint angle correlation to natural walking. The controller consisting of a combination of both hybrid and direct torque control, with more weight on the hybrid model, was also able to decrease the activation of four out of six muscles measured in the lower limbs, which includes knee flexors and extensors, and ankle dorsi- and plantarflexors, on average when compared to walking with the exoskeleton in passive mode. The decrease in muscle activity indicates that this control approach is able to provide assistance as well as improve the spatiotemporal performance. As the joint-level performance was not meaningfully improved by this controller consisting of a combination of both approaches, this control alone would be insufficient for users who require assistive as well as corrective torques from the exoskeleton. For example, those who have suffered from an incomplete spinal cord injury or post-stroke hemiparesis do not have the ability to walk with a natural gait, therefore can benefit from corrections from the exoskeleton to achieve a natural gait. The addition of corrective torques in the form of a position feedback control (FB) to the previously defined feedforward control (FF) is designed to provide both assistive and corrective torques to the user. In a pilot study with two participants for both treadmill and overground walking, the feedforward control alone has the best spatiotemporal performance while the feedback alone has the best joint-level performance. A combination of the two controllers will produce a balance of these two characteristics. All three of these controllers (FF, FB, FF-FB) were able to produce some reduction in muscle activation of the knee extensor and ankle dorsiflexor muscles, compared to passive exoskeleton walking. This indicates that all the controllers provide some level of assistance. However further testing is required to validate this hypothesis as well as optimize the method for combining these two control approaches. This thesis demonstrated that the application of biological joint torques as an exoskeleton controller can be further improved with the addition of other control strategies. It is possible that combining biological torques with other control approaches, including those not explored in this thesis, will be more suitable for those suffering from physical impairments such as hemiparesis or severe muscle weakness.
Modeling and Simulation of Lower Limb Spasticity in Motor-Impaired Individuals
(University of Waterloo, 2021-06-29) Cha, Yesung; Arami, Arash
Spasticity is a symptom that impairs the ability to freely move and control one’s limbs through increased tone and involuntary activations in the muscles. It can cause pain and discomfort and interfere with daily life and activities such as walking. Spasticity is a result of upper motor neuron lesions and is seen commonly in survivors of stroke and brain trauma, and individuals with cerebral palsy, multiple sclerosis, and spinal cord injuries. Despite its ubiquity the phenomena is not well understood. However, the most referred to definition describes spasticity as “a velocity-dependent increase in tonic stretch reflexes with exaggerated tendon jerks, resulting from hyper-excitability of the stretch reflexes.” Qualitative, subjective measures are commonly used in the clinical setting to assess spasticity, most notably the Modified Ashworth score, which has been shown to have inconsistent reliability, relying heavily on the examiner’s experience, and is inaccurate for the lower limbs. Furthermore, these subjective scores do not account for the velocity-dependence of spasticity, which is a key differentiator against other symptoms such as rigidity. Consequently, there is a need for an objective measure of spasticity that can provide a more accurate and reliable alternative or supplement to the current clinical practice, in order to improve the evaluation of treatment and rehabilitation for spasticity. To address this need, a system was developed, validated and applied for modeling the spasticity in the lower-limbs of an affected individual. An experimental setup consisted of a brace-handle system with integrated force sensors for passive actuation of each leg segment, stretching spastic muscles to assess the severity of the condition. The setup included wearable sensors sEMG and IMUs – recording muscular activity and limb segment kinematics respectively during these motions. From the data, onsets of muscular activity and subsequently the trigger points of spastic reflexes were identified, which were mapped onto the calculated joint kinematics. Based on threshold-control theory, stretch reflex threshold (SRT) models of spasticity were created for each muscle by plotting the joint velocities and positions and using regression analysis to create a dynamic threshold in the kinematic space that divided the regimes of spastic and non-spastic motion. These muscle-specific models were combined by muscle groups, leading to the creation of a novel, data-based measure that characterizes the severity of spasticity of a group of muscles. The models and measures were found to agree with the expected changes from different conditions of muscle stretch, and different levels of spasticity in the included subjects, but required more data for statistical validation. The muscle-specific models were then implemented in a spasticity controller developed for use in neuromuscular simulations, in addition to further modeling of spastic reflex characteristics. The controller was applied in a scenario simulation of the same passive movement spasticity assessments used to collect the original data, which provided additional validation of the methodology and results of the modeling. The spasticity controller was also applied in a previously developed reinforcement-learning walking agent, to see the effects of spasticity on simulated gait. Following modification and training of the new agents, the spatio-temporal parameters of gait were analyzed to determine the differences in healthy and spastic gait, which agreed with expectations and further validated the spasticity modeling. This thesis presents a system to accurately and reliably model spasticity, establishing a novel, objective measure to better characterize spasticity, validating it through demonstrations of its use that may be extended in future work to accomplish better understanding of spasticity and provide invaluable improvements to the lives of affected individuals through practical applications.
Recovering Optimal Cost Functions for Natural Walking: From Musculoskeletal Simulation to Exoskeleton Control
(University of Waterloo, 2022-05-03) Weng, Jiacheng; Arami, Arash; Hashemi, Ehsan
Human movement studies have contributed to our understanding of how the central nervous system's (CNS) interactions with our body results in rich and complex motor behaviours, such as human gait. Such understanding is particularly important for human-centered engineering such as lower-limb exoskeletons. Assuming the emerged natural gait patterns are the result of some optimization done by CNS, researchers modelled the walking simulation problem as an optimization problem that recast the walking task into a cost function. However, accurately capturing the CNS goal within the cost function is challenging. Cost functions in existing studies were often assumed a priori which either did not lead to natural gait behaviour, or were manually tuned based on the researcher's knowledge which is time consuming. Some studies attempted to tune the cost function algorithmically using inverse optimal control (IOC), but suffered from expensive computation. These limitations hinder the use of IOC for personalized cost function tuning and, by extension, exoskeleton controller design. To address this issue, computationally efficient tuning methods of the cost function were designed and validated in two optimization frameworks: deep reinforcement learning (DRL) and predictive simulation. For DRL, a novel learning method, which generates a control policy with close-to-natural walking behaviour, was developed. The proposed neuromechanically-inspired cost function contributed to the effective learning of the realistic gait by the DRL agent. The nature-inspired curriculum learning scheme led to efficient convergence to natural and bilateral symmetric gait by adaptively tuning the cost function weights while maintaining the agent's walking capability. To further improve the cost function tuning efficiency, an efficient IOC algorithm named Adaptive Reference IOC (AR-IOC) was proposed that used direct collocation for solving optimal gait trajectories and gradient-based weight optimization. We showcased the efficiency of the proposed algorithm in tuning cost functions and matching gait trajectories using both synthetic data and experimental data which outperformed the Genetic Algorithm by more than 80\% in computational time. With the AR-IOC, the correlation between the walking tasks and the cost function weights were studied which revealed a change in cost function compositions with respect to walking speed. With the efficient AR-IOC algorithm, we explored the potential of using predictive simulation to generate physics-informed reference trajectories for lower-limb exoskeleton tracking controllers. First, an accurate human-exoskeleton system was developed. By combining the optimal human cost function obtained using AR-IOC and the exoskeleton cost function, we obtained the optimal gait trajectories for the human-exoskeleton system which were different from the unassisted natural walking trajectories. These optimal trajectories were then tested in real exoskeleton systems using a time-dependent proportional-derivative (PD) controller and their performances in reducing muscle activities were compared to the unassisted natural walking trajectories. The likely limitations of the controller design were also discussed. With the proposed simulation frameworks and the efficient cost function tuning methods, this thesis serves as a catalyst for enabling personalized rehabilitation design based on detailed musculoskeletal simulation. The presented framework, that covers from data collection and post-processing, to simulation and experiments, serves as a guidance and reference to future developments in this field, such as extending the musculoskeletal simulation to impaired subjects with different locomotion tasks, and different control systems for lower-limb rehabilitation.
The Relationship Between Embodiment Perception and Motor Learning in Virtual Reality-based Interventions
(University of Waterloo, 2024-05-10) Ajami, Sahand; Arami, Arash; Schneider, Oliver
Virtual reality (VR) is a rapidly evolving technology that offers immersive experiences by simulating realistic environments and interactions. In the context of motor learning and rehabilitation, VR has emerged as a promising tool because of its ability to provide controlled, customizable and engaging training scenarios. A key factor in the effectiveness of VR-based interventions is the sense of embodiment, which refers to the user's perception of being present in the virtual environment and having control over a virtual body. This thesis investigates the influence of different sensory feedback modalities on the sense of embodiment and task performance in VR-based motor learning. Through two studies, we examine how the combination of visual and tactile feedback affects embodiment perception and motor task performance in VR environment. In the first study, we explore the effects of pressure feedback on task performance and embodiment in VR-based mirror therapy. Twenty-two able-bodied participants were divided into two groups, with one group receiving the pressure feedback on their thumb and index fingertips during a pick-and-place task. The results indicate that the group with haptic feedback achieved a 15.07% higher task success rate and reported a 12.80% higher embodiment perception compared to the control group. The second study extends the investigation to the impact of vibrotactile feedback in a ball-and-beam control task. Nineteen participants were exposed to four conditions: No Feedback, Vibrotactile Feedback only, Visual Feedback Only and Both Vibrotactile and Visual Feedback. The condition with Both Vibrotactile and Visual Feedback had a 14.52% improvement in task performance and a higher embodiment perception compared to other conditions. Overall, this thesis contributes to the understanding of how sensory feedback modalities can be effectively integrated into VR systems to enhance embodiment and motor learning, suggesting that incorporating haptic feedback into a visual interaction may be associated with higher embodiment and improved motor task performance. These insights have implications for the design of more effective VR-based interventions for training and rehabilitation purposes, emphasizing the value of multisensory feedback in these contexts.