[pt] No contexto de robos colaborativos, há um crescente interesse em Atuadores Elásticos em Série impulsionado pela necessidade de garantir segurança e funcionalidade. No entanto, as não linearidades inerentes a esses atuadores, como atrito, folga nas engrenagens e ruído, aumentam significativamente o desafio de controlar e modelar tais dispositivos. Além disso, um elemento elástico adiciona uma nova não linearidade. Visando essas características, este trabalho propõe um extenso trabalho de identificação do sistema para obter um modelo para um atuador elástico em série baseado em elastômero de baixo custo e original. As metodologias propostas investigam diferentes características do sistema. A primeira se concentra em modelar as não linearidades da junta elástica por meio de um modelo híbrido. A segunda contribuição visa examinar a precisão de redes neurais informadas por física para identificação de caixa cinza de parâmetros de atrito. Por último, é proposto uma metodologia para obter os estados da montagem usando vídeo. A partir dessas estimativas, é proposta uma identificação de caixa cinza usando vídeo. Todos os três estudos utilizam os dados da montagem do atuador. As duas primeiras contribuições obtiveram resultados importantes indicando a eficiência das metodologias propostas. A terceira contribuição mostrou o potencial da nova abordagem de identificação baseada em vídeo. / [en] In the context of human interactive robotics, there is a growing interest in Series Elastic Actuators (SEA), driven by the critical need to ensure safety and functionality. Moreover, a precise model is required to obtain optimal control. However, the inherent nonlinearities of those actuators, such as friction, gear backlash, and noise, greatly increase the challenge of controlling and modeling such devices. Furthermore, a compliant element adds a new nonlinearity, making the modeling task more challenging. Aiming to tackle these issues, this work proposes extensive system identification to obtain mathematical models characterizing the dynamics of an original low-cost elastomer-based SEA. The proposed methodologies investigate different characteristics of the system. The first focuses on modeling the elastic joint s nonlinearities through a hybrid model. The second contribution aims to examine the accuracy of physics-informed neural networks for gray-box identification of friction parameters. Lastly, a framework to obtain the states of the assembly using video is proposed. From these estimations, a gray-box identification using video is proposed. All three studies use the data from the actuator assembly. The first two contributions obtained important results indicating the efficiency of the proposed methodologies. The third contribution showed the potential of the novel video-based identification approach.

This thesis presents a new mechanical design for an exoskeleton actuator to power the sagittal plane motion in the human hip. The device uses a DC motor to drive a Scotch yoke mechanism and series elasticity to take advantage of the cyclic nature of human gait and to reduce the maximum power and control requirements of the exoskeleton. The Scotch yoke actuator creates a position-dependent transmission that varies between 4:1 and infinity, with the peak transmission ratio aligned to the peak torque periods of the human gait cycle. Simulation results show that both the peak and average motor torque can be reduced using this mechanism, potentially allowing a less powerful motor to be used. Furthermore, the motor never needs to reverse direction even when the hip joint does. Preliminary testing shows the exoskeleton can provide an assistive torque and is capable of accurate position tracking at speeds covering the range of human walking. This thesis provides a detailed analysis of how the dynamic nature of human walking can be leveraged, how the hip actuator was designed, and shows how the exoskeleton performed during preliminary human trials.

[pt] Questões de segurança têm sido fatores cruciais para que robôs se tornem aptos a trabalhar em colaboração com seres humanos. Esse esforço envolve um controle de força mais refinado e uma certa flexibilidade nas juntas para que a adaptação dos robôs ao ambiente real e às atividades comuns dos seres humanos seja efetiva. Uma das tecnologias com esse objetivo é o Atuador Elástico em Série (SEA - Series Elastic Actuator), que apresenta um bom desempenho para controle de força, tolerância a impactos causados por agentes externos, baixa impedância, e a redução de vibrações mecânicas. Em um SEA, um elemento elástico passivo é adicionado entre o motor e o elo acionado, a fim de gerar flexibilidade. Este elemento pode ser uma mola, ou outro elemento deformável com flexibilidade caracterizada por sua geometria e pela elasticidade do material utilizado. Esta tese propõe um Atuador Elástico em Série Baseado em Elastômero (eSEA), cuja flexibilidade é obtida a partir de um elastômero depositado entre dois elementos metálicos: um interno acoplado ao atuador, e o outro externo acoplado ao elo. O eSEA foi projetado e avaliado por software de CAD e Elementos Finitos, com o intuito de obter a flexibilidade desejada para a aplicação. Foram produzidas duas versões do eSEA, com duas durezas diferentes: 10 e 55 Shore A. Testes estáticos com células de carga foram executados para caracterizar a rigidez dos eSEA. Os eSEA foram instalados em manipuladores robóticos especialmente desenvolvidos para essa tese. Experimentos compararam o desempenho das técnicas de controle com e sem a influência dos eSEA, mostrando que o uso dos eSEA diminuiu os erros de posicionamento do manipulador e possibilitou o controle de força sem a necessidade de sensores específicos. A fim de criar um modelo para que a estimativa do torque seja mais precisa a partir do eSEA, foram realizadas técnicas de identificação para estimar uma função de transferência que melhor representa o alongamento da borracha. E combinados com modelos NARX e NARMAX do erro de estimativa, gerou-se um modelo híbrido para o elemento elástico no qual soma-se a função de transferência com o erro modelado. / [en] Safety issues have been crucial factors for robots to become able to work in collaboration with humans. This effort involves more refined force control and a certain flexibility at the joints, for the robots to better adapt to real environments and common human tasks. A technology with this objective is the Series Elastic Actuator (SEA), which presents good performance for force control, tolerance to impacts caused by external agents, low impedance, and dampening of mechanical vibrations. In an SEA, a passive elastic element is added between the motor and the driven link, in order to generate a desired flexibility. This element can be a spring, or else another deformable element with flexibility characterized by its geometry and material elasticity. This thesis proposes an Elastomer-Based Series Elastic Actuator (eSEA), whose flexibility is obtained from an elastomer deposited between two metallic elements: an internal element attached to the actuator, and an external element attached to the link. The eSEA was designed and evaluated by CAD and Finite Element software, in order to obtain the desired flexibility for the application. Two versions of the eSEA were produced, with two different hardnesses: 10 and 55 Shore A. Static tests with load cells were then executed to characterize the stiffness of the eSEA. The eSEA elements were installed on robotic manipulators especially developed for this thesis. Experiments compared the performance of control techniques with and without the influence of eSEA, showing that the use of the eSEA reduced manipulator positioning errors and enabled force control without the need for specific sensors. In order to create a model for more accurate torque estimation from eSEA, identification techniques were performed to estimate a transfer function that best represents the rubber elongation. And combined with NARX and NARMAX models of the estimation error, a hybrid model was generated for the elastic element in which the transfer function is added together with the modeled error.

The implications of gait disorder, muscle weakness, and spinal cord injuries for work and age-related mobility degradation have increased the need for rehabilitation exoskeletons. Specifically, the hip rehabilitation exoskeletons due to a high percentage of the mechanical power is generated by this join during the gait cycle. Additionally, the prolonged hospitalisation after hip replacement and acetabular surgeries that affect human mobility, the social-economic impacts and the quality of life. For these reasons, a hip rehabilitation exoskeleton was our focus in this research, as it will contribute being a sustainable solution to take over the burden of physiotherapy and let patients perform their rehabilitation at home or outdoors.  This thesis details an approach of creating a hip rehabilitation exoskeleton, starting with modelling, simulating, and controlling the rehabilitation hip joint in a based-simulation environment. The mathematical model and the reason for using a series elastic actuator in the hip joint to execute the movement in a sagittal plane are more detailed. Because trajectory tracking is commonly used for controlling rehabilitation exoskeletons to ensure safe and reliable motion tracking methods; therefore, two desired torque signals were tested and analysed with the optimal linear quadratic regulator (LQR). The experiments were performed using two torque signals of a healthy hip joint—representing the sit-to-stand (STS) and the walking activity for their importance in lower limb movements. However, the mathematical model used as a basis of the optimal control strategy is usually influenced by multiple sources of uncertainties. Therefore, four case studies of various optimal control strategies were tested for a twofold reason: to choose the most optimal control strategy, and to test the energy consumption of these cases during the STS and walking movements, because the long-term goal is to produce a lightweight and reliable rehabilitation hip exoskeleton. The research showed compelling evidence that tuning the control strategy will not influence the robustness of an optimal controller only, but affect the energy consumption during the STS and walking activity, which needs to be considered in exoskeleton control design regarding its applications. / Behovet av exoskelett för rehabilitering har ökat p.g.a. komplikationer som uppstår vid arbete och åldersrelaterad försämring. Komplikationerna består bland annat av gångstörning, muskelsvaghet och ryggmärgsskador. Speciellt höftexoskelett avsett för rehabilitering är extra intressant på grund av att rehabilitering inom detta område omfattar långvarig sjukhusvistelse efter höftprotes- och acetabulära operationer. Höftleden är en av de leder som utsätts för relativt höga mekaniska påfrestningar och minskad rörelseförmåga leder inte sällan till socioekonomiska effekter och minskad livskvalité. Av denna anledning kommer höftexoskelett för rehabilitering vara det primära området i denna avhandling då det kommer att vara en lösning för att minska belastningen inom sjukvård och låta patienter utföra sin rehabilitering hemma på egen hand. Denna avhandling beskriver en metod för att skapa ett höftexoskelett avsett för rehabilitering med början i modellering, simulering och kontroll av en höftled av exoskelett i en simuleringsmiljö. Genom att använda ett serieelastiskt manöverdon för att utföra en höftledsrörelse i ett sagittalt så uppnås en mer detaljerad matematisk modell. Genom att använda banspårning, som vanligtvis används för att kontrollera exoskelett för rehabilitering för att säkerställa säkra och pålitliga rörelsespårningsmetoder, så analyserades två vridmomentssignaler mot en linjär kvadratisk regulator (LQR). Simuleringarna utfördes med hjälp av två vridmomentsignaler som representerar sitt-till-stå (STS) och gångaktivitet hos en frisk höftled. Den matematiska modellen som används för att hitta den optimala kontrollstrategin påverkas vanligtvis av flera osäkerhetskällor. Därför testades fyra fallstudier av olika optimala kontrollstrategier för två skäl: den ena för att välja den mest optimala kontrollstrategin emellan och den andra för att mäta energiförbrukningen för dessa STS och gångrörelse så att vi kan producera ett lätt och pålitligt höftexoskelett avsett för rehabilitering. Forskningen visar övertygande bevis för att inställning av styrstrategin inte bara kommer att påverka robustheten hos en optimal styrenhet utan även påverkar energiförbrukningen under STS och gångaktivitet vilket måste beaktas vid design av exoskelett.