Compliant Gravity Balancing Orthosis
Flexion-synergy is a stereotypical movement pattern that inhibits independent joint control for those who have been affected by stroke; this abnormal co-activation of elbow flexors with shoulder abductors significantly reduces range of motion when reaching against gravity. While wearable orthoses based around compliant mechanisms have been shown to accurately compensate for the arm at the shoulder, it is unclear if accurate compensation can also be achieved while minimizing device bulk.
In this work, we present a novel, multi-objective simulation-optimization framework towards the goal of designing practical gravity-balancing orthoses for the upper-limb. Our framework includes a custom built VB.NET application to run nonlinear finite element simulations in SolidWorks, and interfaces with a MATLAB-based particle swarm optimizer modified for multiple objectives. The framework is able to identify a set of Pareto-optimal compliant mechanism designs, confirming that compensation accuracy and protrusion minimization are indeed conflicting design objectives.
The preliminary execution of the simulation-optimization framework demonstrates a capability of achieving designs that compensate for almost 90% of the arm’s gravity or that exhibit an average protrusion of less than 5% of the arm length, with different trade-offs between these two objectives.
The system to be balanced is simplified by constraining rotation to the sagittal plane and enforcing the elbow to be fully extended (A). A compliant beam is used to compensate for gravity at the shoulder (B). Perfect compensation occurs when the reaction profile is sinusoidal (C).
UDiffWrist
The complexity of the human wrist hampers the kinematic compatibility achievable when interacting with a wrist exoskeleton. Past studies have shown that the implementation of passive degrees of freedom (DOFs) between the human and robotic kinematic chain may compensate for any joint misalignments between user and robot. However, the additional inertia and friction associated with these new DOFs affect the transparency of the device. As such, there is a need to investigate the trade-off between two critical objectives of wrist exoskeleton design: ensuring low endpoint impedance; and minimizing the effects of misalignments.
To investigate this trade-off, we developed a low-impedance 2-DOF wrist exoskeleton featuring a cable-differential transmission: the UDiffWrist (UDW). We developed two versions of the robot to investigate how the implementation of passive DOFs affect system transparency, torque transfer to the wrist, and system robustness to misalignments. The UDiffWrist-Colocated (UDW-C) places the user’s wrist inside the transmission and assumes perfect alignment between user and robot; while the UDiffWrist-NonColocated (UDW-NC) forgoes this assumption, and implements a compensation mechanisms comprised of passive joints.
Through dynamic characterization, we have concluded that while the UDW-NC is significantly more robust to misalignments, it is significantly less transparent than the UDW-C. Further, the torque transmission capability of the UDW-NC is significantly worse than the UDW-C. This suggests that while the use of passive degrees of freedom as a mechanism to compensate for misalignments is valid, the benefits can be outweighed by the negative affects associated with their inertia and friction. We have thus concluded that for small wrist movements (~10 degree movements in the flexion/extension and radial/ulnar deviation axes) it is not necessary to compensate for misalignments in our 2-DOF wrist exoskeleton, and we can instead assume alignment.
Shape encoding is performed in two stages: 1) a centroid axis is determined as a spline defined off a control polygon (Centroid Axis); 2) Bidirectional thickness is defined off the centroid axis by creating a thickness profile using methods similar to those used for the centroid axis (Thickness Profile).
Publications on this topic
H. A. Chishty, F. Sergi, “A Multi-objective Simulation-Optimization Framework for the Design of a Compliant Gravity Balancing Orthosis”, doi: 10.1101/2024.02.16.580745, 2024, pre-print.
H. A. Chishty, A. Zonnino, A. J. Farrens, F. Sergi, “Kinematic compatibility of a wrist robot with cable differential actuation: effects of misalignment compensation via passive joints”, IEEE Transactions on Medical Robotics and Bionics, vol. 3, pp. 970-979, October 2021, doi: 10.1109/TMRB.2021.3123528, available online, pre-print.
A. Zonnino, F. Sergi, “Optimal design of cable differential actuation for 2-DOF wrist robots: effect of joint misalignments on interaction force”, IEEE Engineering in Medicine and Biology Conference (EMBC), 2016, available online.
