DTA

Archivio Digitale delle Tesi e degli elaborati finali elettronici

 

Tesi etd-01122018-102058

Tipo di tesi
Perfezionamento
Autore
PARRI, ANDREA
URN
etd-01122018-102058
Titolo
Design, development and validation of control architectures for lower-limb wearable robots
Settore scientifico disciplinare
ING-IND/34
Corso di studi
INGEGNERIA - Biorobotics
Commissione
relatore Prof. VITIELLO, NICOLA
Parole chiave
  • Active pelvis orthosis
  • Gait phase estimation
  • Human-robot interaction
  • Intention decoding
  • Wearable robotics
Data inizio appello
20/04/2018;
Disponibilità
completa
Riassunto analitico
Old age dependency ratio is expected to drastically increase worldwide in the next years. Since ageing is strongly related with the emergence of gait disorders, an ageing society could threaten the sustainability of our welfare and economic system. In most of the cases, restoration of the locomotion abilities is limited and patients remain affected by mild lower-limb impairments which result in poor mobility and lack of adequate quality of life.
Design and development of wearable robotic technologies for restoring or empowering gait capabilities of people affected by lower-limb impairments is a compelling topic of both academic and industrial research activities. Wearability implies a close mechanical and cognitive interaction between the robot and its human operator. Symbiosis between the two entities can be achieved only if some critical aspects of the human-robot interface are satisfied. Ergonomic and effective mechanical power transfer, as well as stable and intuitive flow of information, are fundamental for the wearer's comfort and the success of the interpretation of the user intention.
Within this framework, a major open challenge is the design and development of simple, intuitive yet reliable control architectures that are able to decode: (i) the intention of the movement of the user and (ii) the phase of completion of the performed task. The goal of this process is to intend the voluntary motion of the user and cooperate with him/her in a safe and natural fashion along its accomplishment. Acceptability of the interface driving the information from the user to the robot is another important requirements. Ideally, no active action should be requested by the user but the movement intention itself; the responsibility of the robot is then to track and extrapolate information from the movement thanks to wearable sensors measuring human biomechanics.
Moreover, a smart controller should satisfy dependability requirements such as robustness to usability-related issues by: (i) minimizing the complexity of the sensory system and (ii) achieving subject-independency for the sake of a wear-and-play device that guarantees its performance independently on the user or environmental conditions.
The objective of this dissertation is to present the design, development and validation of smart algorithms for the control of lower-limb wearable robots that are capable of: (i) online tracking and providing a continuous gait phase reference signal, and (ii) to online decoding the user’s intention. The distinctive features of the proposed methods are the dependency on a non-invasive, reliable and stable wearable sensory system and subject-independency that guarantees high performance for different users overcoming intra- and inter-variability issues.
Three cases of study are presented in the dissertation. The first one is an adaptive oscillators-based gait-phase estimator used to track the human biomechanics, provide a continuous gait-phase reference that is locked with a phase-invariant biomechanical event, such as the heel strike. As a result, the reset of phase reference for the stride cycle is locked with the same biomechanical event independently on the environmental conditions, e.g. cadence, terrain or even user.
The second one is a method for intention decoding and control of an assistive lower-limb wearable robot that seamlessly detects locomotion-related activities of daily living in real-time. A subject-independent approach relying on a simple wearable sensory apparatus can recognize ground-level walking, stair ascending, stair descending, sitting, standing, sit-to-stand, and stand-to-sit. Its robustness was validated in a human in-the-loop application, i.e. when assistive mechanical power is transferred from the robot to the user on top of the decoding process.
The third one is a decoding method for an active robotic transfemoral prosthesis. The approach relies on a distributed non-invasive wearable interface constituted of a set of IMUs and a pair of sensitive instrumented shoes, decodes on the basis of a simple set of heuristic transition rules the intention of the user and closes the loop on the human counterpart by commanding the proper setpoints to the low-level controllers of the active knee and ankle joints.
The three methods were validated in experimental activities involving active wearable robots and their end-users achieving performance that advances the current state of the art. Robustness and reliability of the applications showed also promising perspectives for their usability in real-life scenarios.
File