Iterative Learning Control Methods for Hybrid Wearable RobotsMolazadeh, Vahidreza (2020) Iterative Learning Control Methods for Hybrid Wearable Robots. Doctoral Dissertation, University of Pittsburgh. (Unpublished)
AbstractIn this dissertation, iterative learning control methods for a hybrid exoskeleton to produce sitting-to-standing and walking in people with paraplegia are investigated. The hybrid exoskeleton combines a lower limb powered exoskeleton and functional electrical stimulation (FES). Limited research has been done to design control methods that provide shared modulation of FES and the powered exoskeleton. A major technical challenge to the implementation of control algorithms is their need to identify a user's musculoskeletal dynamics. Further, currently, setting desired regulation points or desired limb trajectories during sitting-to-standing and walking movements is a daunting task as it requires separate and coordinated design for each lower-limb. An inaccurate regulation of set-points or desired trajectories can possibly cause uncoordinated standing-up movements, potentially destabilizing the user. Goal: The goal of this research is to design robust and adaptive control algorithms for hybrid exoskeletons that overcome the difficulty in model identification, can dynamically allocate the shared use of FES and the powered exoskeleton, and produce coordinated joint movements. Objectives: The primary objective of this research is to develop robust control methods that iteratively learn modeling uncertainties in the hybrid exoskeleton (i.e., addressing model identification), while facilitating allocation of FES and motor input (i.e., resolving actuator redundancy) in the hybrid exoskeleton. The proposed control methods are experimentally validated for a sitting to standing task with the hybrid exoskeleton. The experiments are performed on human participants with no disabilities and a participant with spinal cord injury. The tasks that are accomplished to achieve the objectives are listed as: 1- Design and implement time-invariant desired joint trajectories by using virtual constraints for sitting-to-standing and walking motion 2- Derive and experimentally validate a robust control method that uses an arbitrarily switched allocation strategy to coordinate motor and FES. 3- Derive a control method that iteratively learns the system nonlinear dynamics and control gains. 4- Using an optimal and cooperative model predictive control method, instead of switched control, to allocate between motors and FES. Share
Details
MetricsMonthly Views for the past 3 yearsPlum AnalyticsActions (login required)
|
![[feed]](/images/feed-icon-32x32.png){kind=icon}