A procedure to optimize the geometric and dynamic designs of assistive upper limb exoskeletons

Abstract

The need for upper limb assistive and wearable exoskeletons is growing in various fields, e.g. either to support patients with neuromuscular disabilities or to reduce the effort strains on workers. These exoskeletons should reduce the efforts required by the user during functional tasks (dynamic consideration) and should fit the user’s size (geometric consideration). This is a tedious task, due to the 3D human-exoskeleton interactions, and to the complex and interdependent selection of the power transmission characteristics, i.e. motors or passive elements. There are still few guidelines and few clear procedures to support geometric and dynamic syntheses of these exoskeletons. The objective of this study is to develop a procedure for geometric and dynamic syntheses of assistive upper limb exoskeletons, to serve as a tool to optimize their design. Firstly, a geometric optimization of the exoskeleton dimensions enabled to satisfy the kinematic loop closures between the exoskeleton and the user for a maximum of positions while carrying out specific functional tasks and avoiding collisions with the body segments. Secondly, through an optimal control problem, the dynamic characteristics of the exoskeleton were obtained by minimizing the user’s joint torques for the functional tasks. Closing the kinematic loops of the exoskeletons with optimized dimensions was achieved for all positions of the user while carrying out the functional tasks, which was 10.8% more than with a visual identification of the dimensions. The resulting dynamic parameters could reduce the user’s joint torque to less than 10.6% of the human-only simulations for nearly all joints and tasks. These results showed that the geometric and dynamic synthesis procedures were successful. This is important, as it can enable the development of dedicated exoskeletons, such as lighter and smaller exoskeletons. The future perspectives will be to build an optimization framework, where the geometric and dynamic parameters could be optimized together, and to minimize the user’s muscle forces instead of joint torques for specific design purposes.