Design enhancements and control of a lower extremity exoskeleton

Abstract

Exoskeleton robots, powered by electric, pneumatic, or hydraulic actuators, support human bones and muscles externally, enhancing movement and strength. They are indispensable in fields such as medical, industrial, military, and safety, aiding both healthy and disabled individuals. Full-body exoskeletons support the entire body, while lower-body exoskeletons target the hip, knee, and ankle joints to assist with locomotion and reduce physical stress. The goal of enabling independent walking with exoskeletons has significant potential for neurological dysfunctional rehabilitation and could improve the quality of life of people with disabilities. Over the last two decades, exoskeleton research and development have grown substantially, and some prototypes are now commercially available. However, significant challenges in design, modeling, and control persist. Some of these challenges include managing complex kinematics and dynamics, dealing with intricate geometries, addressing holonomic loop constraints, and overcoming workspace limitations inherent in parallel or series-parallel hybrid mechanisms. The robotics research community employs model-based and model-free control methods to enhance performance, aiming to develop robust, safe, and reliable exoskeletons for assistive, rehabilitative, and power augmentation purposes. Despite these efforts, achieving the desired levels of adaptability, user comfort, and seamless integration remain an ongoing challenge, requiring continuous innovation and interdisciplinary collaboration. This thesis addresses key challenges inherent in the Almost Spherical Parallel Mechanism (ASPM) device, which are crucial to exoskeleton research and development. The ASPM is a 3-DOF module representing the ankle and hip joint modules of the Recupera-Reha full-body exoskeleton. It has three principal movements: dorsiflexion-plantarflexion (DF-PF), eversion - inversion (EV-IN), and adduction-abduction (AD-AB). However, the joint modules are faced with workspace limitations due to inadequate alignment in the exoskeleton leg. We address these problems using the rotative inverse geometric method to align the principal human joint axes with the axes that optimize the ASPM device’s range of motion in the exoskeleton leg. The study demonstrated that rotating the mechanism along the AD-AB yaw axis and selecting appropriate ball and socket joints for the actuators can significantly enhance the usable workspace for the exoskeleton wearer. Furthermore, we present advancements in the functionality of the Recupera-Reha lower extremity exoskeleton robot, which features a series-parallel hybrid design with multiple kinematic loops, resulting in 148 degrees of freedom for all the spanning tree joints and 102 independent loop closure constraints. These complexities pose significant challenges for modeling and control. We tackled these challenges by utilizing the Hybrid Robot Dynamic (HyRoDyn) software framework to address the kinematic loop constraints. We then employed an optimal control approach to generate feasible trajectories for movements such as sitting, standing, and static walking, which are tested on the exoskeleton robot. To generate initial´trajectories, the method efficiently solves the optimal control problem using a serial abstraction of the model. We then use the full series-parallel hybrid model, which accounts for all kinematic loop constraints, to generate the final actuator commands. Finally, experimental results confirm the effectiveness of this approach in achieving the desired motions for the exoskeleton and advancing its functional capabilities.