Development And Assessment Of A Control Approach For A Lower Limb Exoskeleton For Use In Gait Rehabilitation Post Stroke

Stroke is the second highest cause of death worldwide and the third leading cause of adult disability across all age brackets. Recovering gait following stroke is a major goal of patients, and hence rehabilitation, as it is central to many activities of daily living. Of the different treatment modalities, robotic assisted gait training is growing in popularity, but is still considered complementary to, and not substitute for conventional therapies comprising physiotherapy, overground walking and body weight supported treadmill training. The potential advantages that lower limb robotics bring to neurorehabilitation over conventional therapies include, higher dosage, specificity, improved consistency, and duration, though these benefits have been slow to manifest. Exoskeletons are well placed to provide these benefits, as well as environmental variation and task salience if they can be used away from outpatient settings. Control strategies that may be enhancing of recovery are often confined to stationary exoskeletons, and the control of mobile exoskeletons is only loosely related to gait, if at all, which limits rehabilitation outcomes. -- The primary aim of this PhD thesis was to develop an adaptive, user-initiated gait Controller that aims to target a novel neural recovery pathway. The Controller would use a robotic exoskeleton, with the intention of developing novel neuroplasticity that is beneficial for gait and would be permissive of simultaneous control of hip and knee posture. A theoretical framework based on the principles of neuroplasticity was proposed that seeks to bring higher engagement, task variance, and volition to gait rehabilitation. This framework considers stroke and rehabilitation timelines and the interaction of the proposal with existing theory, how beneficial neuroplasticity may manifest, and how the proposal may be detrimental. A comprehensive survey of candidate lower limb devices followed (164 devices), to understand exactly what features are compatible, complementary, or contradictory to the proposed control method, and to understand the implications the various specifications have. Specifically, it was found that ambulating exoskeletons that can move around the environment were preferred for their ability to be used in the community and the home, and that extended joint range of motion will be permissive of activities that are supportive of gait such as sit-to-stand and stair ascent/descent. Of the various control systems that have been implemented with exoskeleton devices, trajectory control, where motion is enforced on the limb by the exoskeleton, is preferred. -- The method of control was assessed for suitability as a gait controller through a participant study (n = 21). Participants were asked to reproduce the motion required for the controller, and with minor modification to participant motion it was shown that reliable control signals can be obtained. The remainder of the thesis applies the learnings of the previous stages in the development of the Controller and an accompanying Sensor. The custom Sensor was designed with a small form factor to be applied on the Controller. The thesis concludes with an implementation of the Controller and a successful demonstration of the proposed concept, where the control signals are reproduced on a scale lower limb exoskeleton. The full technical detail and specification of the Controller, and the custom position Sensor developed specific for this application, are presented as part of this work. -- This work has added a new theoretical framework for gait control following stroke and has added technological capability to implement the proposal. It is the primary recommendation of this PhD that the novel control method be tested further with participant studies and that the component hardware be developed further. Therapies targeting novel recovery mechanisms breathe fresh air into rehabilitation and may inspire other new treatments, and future funded work originating from this PhD will see the concept tested with a chronic stroke population, using an ambulating exoskeleton and the Controller.

Several genetic, developmental and neurological disorders can cause various levels of gait impairment in the pediatric population. Powered lower limb orthoses, or exoskeletons, have recently been used to address gait impairment and afford therapists alternative solutions and strategies for gait therapy. Most exoskeleton research has focused on the adult population while the pediatric population remains underserved. The limitations of current pediatric exoskeletons make them impractical for use in both community and clinical settings. Furthermore, exoskeleton controllers suitable for these environments should promote human volitional control while guiding the subject towards a dynamically stable healthy gait pattern. This dissertation presents the design of a pediatric lower limb exoskeleton and the application of a virtual constraint-based controller on the device. First, a small and lightweight exoskeleton joint actuator capable of delivering the torque and power requirements needed to assist and guide the hip and knee joints was developed. Testing and in-air gait tracking of a model leg in a provisional orthosis demonstrated that the joint actuators were suitable for use in a pediatric exoskeleton. Second, an adjustable exoskeleton frame was designed and fabricated, and a human factors assessment of the fully assembled pediatric lower limb exoskeleton demonstrated that the device was lightweight, comfortable, easily adjustable and suitable for children. Third, a virtual constraint-based controller was applied on an underactuated adult exoskeleton. This initial investigation demonstrated that virtual constraint-based control guided the subject towards a dynamically stable gait in a time-invariant manner, provided greater volitional control to the subject and promoted active participation in the walking exercise. Finally, this dissertation research concluded with the application of a virtual constraint-based controller on the pediatric lower limb exoskeleton in treadmill walking experiments. The results showed that virtual constraint-based control reduced gait variability and the amount of robotic intervention applied relative to proportional-derivative control. Subject feedback also indicated that the virtual constraint-based controller was easier to use compared to time-based proportional-derivative control. This dissertation research demonstrates that the developed exoskeleton is suitable as an investigative platform for pediatric exoskeleton controllers and that virtual constraint-based controllers have potential for the rehabilitation and guidance of pediatric gait.

The field of robot-assisted physical rehabilitation and robotics technology for providing support to the elderly population is rapidly evolving. Lower limb robot aided rehabilitation and assistive technology have been a focus for the engineering community over the last three decades as several robotic lower limb exoskeletons have been proposed in the literature as well as some being commercially available. One of the most important aspects of developing exoskeletons is the selection of the appropriate material. Strength to weight ratio is the most important factor to be considered before selection of a manufacturing material. The material selection strongly influences the overall weight and performance of the exoskeleton robot. In addition to material selection the type of mechanism and the actuation strongly effect the overall weight of the lower limb robotic exoskeleton. Most of the lower limb exoskeleton provided in the literature use a parallel mechanism, are properly actuated and either use aluminium or steel as their manufacturing materials. All these factors significantly increase the weight of the lower limb robot exoskeleton and make the device heavy, bulky, and uncomfortable for the wearer. Furthermore, an increase in weight contributes to a decrease of energy efficiency, reduces the energy efficiency of the final product, and increase the running cost of the designed robot devices. This thesis explores the wide-ranging potential of lower limb robot exoskeletons in the context of physical rehabilitation. Implementation and testing of a lightweight and high strength material without effecting the reliability was the main research objective of the present work. In this research, a linkage based under-actuated mechanism was used for the development of a lightweight design. Structural and mechanical load analysis of the mechanism was performed by using an advanced approach of finite element analysis. Three materials, namely structural steel, aluminium, and carbon reinforced fibre were compared as the manufacturing materials of the modelled mechanism. After that, a weight estimation was carried out for all three materials and the material which exhibits the best response under mechanical load analysis was selected. From the weight comparison, the carbon reinforced fibre provided the least weight for the digital twin of a lower limb exoskeleton. After material selection, the next step was the topology optimisation to further decrease the mass of the designed prototype without effecting the mechanical performance. The optimisation was carried out by using a multi-mode single objective genetic algorithm (GA) and a reduction of 30 % in the weight of the designed prototype was obtained. The selected material, which is carbon fibre, is a type of polymer material that is highly anisotropic, meaning it shows different material behaviour in different orientations of applied force. The next stage of the research work was the material characterization of the manufacturing material, which was carried out both analytically and experimentally. For defining the optimal criteria for fiber orientation, Hashin's Failure Criteria is considered, and experimental work is performed to determine the most suitable fibre orientation. The material monotonic tensile properties were experimentally determined by experimental work and used to select a suitable orientation to manufacture a physical prototype model of the lower limb robot exoskeleton. After that the manufacturing process was carried out which is divided into three main steps. The first step was the use of the suitable lightweight and high strength material, which was selected by weight comparison in the design stage. The second step was the use of a single actuator to actuate the whole mechanical system and the final step was the use fabrication method to get a strong and reliable structure. Shaping of the different exoskeleton parts was carried out by CNC milling and parts were assembled to build a robotic prototype. A DC motor was used to actuate the complete prototype, which includes hip, knee, and ankle joints. In the end, a reliability analysis was carried out by using a machine learning based approach. A machine learning framework was developed for time-dependent reliability analysis of the developed robot. A neural network algorithm was designed to estimate the time-dependent reliability of the joint displacement and the positions of the end-effector first. From the above methodology, a lightweight and high strength lower limb robot exoskeleton was just not only conceptualized but a significant work was done to get a physical model starting from the material selection and concluding with the fabrication of a physical prototype. The reliability analysis gives an overview of the mechanism safety as a function of joint displacement. The designed prototype of carbon reinforced fibre was four times lighter in weight as compared to steel and three times lighter than aluminium, which is expected to give the wearer a comfortable wearing experience and improves the overall physical rehabilitation experience for the patients.

This conference proceeding gather a selection of peer-reviewed papers presented at the 1st International Conference on Artificial Intelligence for Smart Community (AISC 2020), held as a virtual conference on 17–18 December 2020, with the theme Re-imagining Artificial Intelligence (AI) for Smart Community to apply computational intelligence for biomedical instruments, automation & control, and smart community to develop suitable solution for various real-world application. The conference virtually brought together researchers, scientists, engineers, industrial professionals, and students presenting important results in the related field of healthcare technology, soft computing technologies, IoT, evolutionary computations, automation and control, smart manufacturing and smart cities. Researchers and scientist working in the allied domain of Artificial Intelligence and others will find the book useful as it will contain some latest computational intelligence methodologies and applications.

The book provides readers with a comprehensive overview of the state of the art in the field of gait and balance rehabilitation. It describes technologies and devices together with the requirements and factors to be considered during their application in clinical settings. The book covers physiological and pathophysiological basis of locomotion and posture control, describes integrated approaches for the treatment of neurological diseases and spinal cord injury, as well as important principles for designing appropriate clinical studies. It presents computer and robotic technologies currently used in rehabilitation, such as exoskeleton devices, functional electrical stimulation, virtual reality and many more, highlighting the main advantages and challenges both from the clinical and engineering perspective. Written in an easy-to-understand style, the book is intended for people with different background and expertise, including medical and engineering students, clinicians and physiotherapists, as well as technical developers of rehabilitation systems and their corresponding human-compute interfaces. It aims at fostering an increased awareness of available technologies for balance and gait rehabilitation, as well as a better communication and collaboration between their users and developers.

Robotic rehabilitation for physical therapy has several advantages over conventional manual rehabilitation, especially in the aspects of accuracy and repeatability. Initial attempts at robotic rehabilitation focused on training muscles by moving limbs in a fixed repetitive pattern. Later it was realized that such an approach could be suboptimal. Better approach would be the use of 'assist-as-needed' paradigm, where an orthotic device provides just enough assistance to enable the patient to move his leg under his own control. However, at this time, lower extremity devices which can apply appropriate forces to implement this paradigm are still in research and not commercially available. The goal of this work is to develop lower extremity orthotic devices using assist-as-needed paradigm for robotic rehabilitation. To achieve this goal two orthotic devices were developed. They are Gravity Balancing leg Orthosis (GBO) and Active Leg EXoskeleton (ALEX). GBO assists persons with hemiparesis to walk by reducing or eliminating the effects of gravity on the affected limb. The amount of assistance provided can be tuned by the therapist from 0% to 100% gravity balancing. For a quantitative evaluation of the performance of the device several experiments were conducted. These experiments were performed on five healthy subjects and three stroke patients. The results showed that with the GBO set to 100% balancing the EMG activity from the rectus femoris and hamstring muscles was reduced by 75%, during static hip and knee flexion, respectively. For leg-raising tasks the average torque for static positioning reduced by 66.8% at hip joint and 47.3% at knee joint, however if transient portion of the leg raising task is included, the average torque at hip reduced by 61.3% and at knee increased by 2.7% at knee joints. In the walking experiment there was a positive impact on the range of movement at the hip and knee joints, especially for stroke patients, the range of movement increased by more than 57% at hip joint and by more than 73% at the knee joint. These results show that the GBO provides assistance which can be used for rehabilitation. An intensive training of a stroke patient was performed to study the long term effects of GBO, the training lasting for six weeks. The training started out with maximum assistance of 100% gravity balancing and gradually reduced to 0% by the end of training. Patient is also shown visual display of his gait pattern in real time and summary performance after individual sessions. Some of the effects of the training were, increase in patients preferred speed of treadmill walking from 2.72 km/h to 3.04 km/h, patient's preferred overground speed increased from 3.38 km/h to 3.86 km/h by the last evaluation. An improvement of gait pattern was seen where the patients gait pattern became more like a healthy subject's pattern. The patient was able to increase weight bearing on the hemiparetic leg and was more symmetric in his walk. ALEX, on the other hand, is a motorized orthotic device. To achieve the goal of 'assist-as-needed' paradigm for ALEX, Force-Field controller was developed. This controller generates "virtual walls'' in the plane containing human thigh and shank segments. These virtual walls guide and assist the subject's foot along the prescribed trajectory. Linear actuators were used at hip and knee joints of the device. To make the actuators back-drivable, friction compensation was used. Gait training studies with healthy subjects were conducted to measure the effectiveness of ALEX in retraining modified gait pattern. The results show that a healthy human leg muscles can be trained in about 45 to 60 minutes to a modified pattern of foot trajectory. A 15-day long gait training was conducted with a stroke patient using ALEX, the results indicate that using ALEX and force-field controller, the patient's gait pattern improved significantly in many aspects. His gait speed improved both on treadmill from 1.45 km/h to 2.57 km/h and overground from 1.82 km/h to 2.50 km/h. His foot trajectory increased and got about 85% closer to a healthy subject's foot trajectory. Knee flexion increased from 27.2 deg to 47.5 deg and ankle dorsi-flexion increased from 1.9 deg to 5.9 deg by the end of the training. All these results indicate that by using these devices suitably and implementing a long term gait training can help patients with walking disability in a speedy recovery.

The concepts represented in this textbook are explored for the first time in assistive and rehabilitation robotics, which is the combination of physical, cognitive, and social human-robot interaction to empower gait rehabilitation and assist human mobility. The aim is to consolidate the methodologies, modules, and technologies implemented in lower-limb exoskeletons, smart walkers, and social robots when human gait assistance and rehabilitation are the primary targets. This book presents the combination of emergent technologies in healthcare applications and robotics science, such as soft robotics, force control, novel sensing methods, brain-computer interfaces, serious games, automatic learning, and motion planning. From the clinical perspective, case studies are presented for testing and evaluating how those robots interact with humans, analyzing acceptance, perception, biomechanics factors, and physiological mechanisms of recovery during the robotic assistance or therapy. Interfacing Humans and Robots for Gait Assistance and Rehabilitation will enable undergraduate and graduate students of biomedical engineering, rehabilitation engineering, robotics, and health sciences to understand the clinical needs, technology, and science of human-robot interaction behind robotic devices for rehabilitation, and the evidence and implications related to the implementation of those devices in actual therapy and daily life applications.

Lower-limb stroke rehabilitation is physically demanding on therapists and requires the concerted effort of multiple staff members. Researchers have accordingly begun investigating the use of lower-limb exoskeletons for rehabilitation. Unfortunately, if the exoskeleton ensures the correct trajectory regardless of whether or not the user contributes effort, rehabilitation can be ineffective as the patient can begin to slack. Recent research suggests using assistance-as-needed control to facilitate functional motor recovery by only applying torques if the patient deviates too far from the desired trajectory. Assistance-as-needed control has been difficult to employ in lower-limb exoskeletons, however, due to the need to ensure stability. This work demonstrates how virtual constraint control—a method used in prostheses and assistive exoskeleton control with robust stability properties—can be combined with a velocity-modulated deadzone to ensure stability. The simulations suggest the method can accommodate a large deadzone while remaining stable across a range of gait pathologies.