Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Bipedal locomotion is a remarkable feature of humans. This skill is necessary for the activities of daily living. Unfortunately, many people only partially benefit from it or miss it entirely because of a disability. This may result in a slower gait, less endurance before pain or exhaustion, inability to climb stairs, or even the permanent use of a wheelchair. This may make reaching remote places difficult, prevents from practicing some sportive activities, and limit the chances of social interaction. Robotic exoskeletons have been introduced recently to assist the gait. The control of such devices should be addressed with care, because of the interaction with the human user, which makes part of the system unpredictable or at least difficult to model. This PhD thesis report explores the broad range of existing control methods that are currently disseminated in the literature. All these references have been sorted with a novel classification that categorizes the subparts of an algorithm, instead of considering the entire one. This rational overview contribute to understanding what was already tried, and what is missing to the state of the art. Then, the developed modular control platformis described. This platform allowed sharing electronicmodules and software between all the devices and provided a consistent and efficient development experience. It could handle all eight exoskeletons of the laboratory, even though their hardware was never the same. Fully embedded and with a convenient development interface, it allowed testing efficiently in the field. The case of the fullmobilization for paraplegic users was tackled with the exoskeleton TWIICE, intended for personal use. A controller based on operating modes was implemented and allowed its user to walk again and overcome many types of obstacles of daily living, autonomously. This controller made TWIICE the current best performer for the "Powered Exoskeletons Race" category of the CYBATHLON competition, able to overcome all the six obstacles. A variant, WIITE, enabled another paraplegic user to practice ski touring in snowy mountains. This showed a use-case with a similar handicap, but a totally different need. TWIICE is sagitally constrained, is actuated at the hip and knee joints only and has a locked ankle, to be lightweight and cost-effective. Crutches are then required to maintain the user’s balance, which prevents using the arms for interaction with the surrounding environment. A controller performing dynamic balance with such hardware would be too complicated and probably not robust enough. A suggested acceptable trade-off was to let the user use freely his arms for interactions only while standing still and use normally the crutches while walking. The controller was biologically inspired, from healthy participants trying to keep standing dynamically in a passive constraining exoskeleton, called INSPIIRE. Dynamic standing balance control was successfully assessed with a complete paraplegic user wearing TWIICE. Finally, partial assistance techniques have been investigated. First, in running using SPRIINT, a unilateral hip orthosis to assist a transfemoral amputeed user. Second, the endurance augmentation in walking was also studied with HiBSO, a bilateral hip exoskeleton. A novel assistance controller was developed, able to immediately assist, from the beginning of the first step.
Auke Ijspeert, Mohamed Bouri, Ali Reza Manzoori, Romain Pierre François Baud
Auke Ijspeert, Mohamed Bouri, Ali Reza Manzoori, Coline Lugaz, Tian Ye, Davide Malatesta