Enhanced Optimization Strategies to Design an Underactuated Hand Exoskeleton

TL;DR

This work addresses safe and effective design of a complex underactuated hand exoskeleton (U-HEx) by posing design optimization as both a single-objective problem and a subsequent multi-objective problem that includes torque balance and actuator-displacement considerations. It benchmarks evolutionary algorithms (GA and BBBC) and introduces two BBBC-based MOEAs (NS-BBBC and SP-BBBC), revealing that BBBC generally offers faster, more consistent convergence, while MOOP exposes diverse trade-offs among force transmission, torque balance, and displacement. The results show that constraining actuator displacement reduces optimal force transmission, but turning constraints into objectives yields richer Pareto fronts and design guidelines. The study provides design insights for safe, efficient exoskeletons and offers practical guidance on choosing optimization strategies based on user needs and design priorities. It also suggests avenues for future work, including rank-partitioning MOOP and real-human testing to assess usability and interaction forces.

Abstract

Exoskeletons can boost human strength and provide assistance to individuals with physical disabilities. However, ensuring safety and optimal performance in their design poses substantial challenges. This study presents the design process for an underactuated hand exoskeleton (U-HEx), first including a single objective (maximizing force transmission), then expanding into multi objective (also minimizing torque variance and actuator displacement). The optimization relies on a Genetic Algorithm, the Big Bang-Big Crunch Algorithm, and their versions for multi-objective optimization. Analyses revealed that using Big Bang-Big Crunch provides high and more consistent results in terms of optimality with lower convergence time. In addition, adding more objectives offers a variety of trade-off solutions to the designers, who might later set priorities for the objectives without repeating the process - at the cost of complicating the optimization algorithm and computational burden. These findings underline the importance of performing proper optimization while designing exoskeletons, as well as providing a significant improvement to this specific robotic design.

Figures (8)

• Figure 1: The optimized device for the index finger component of U-HEx.
• Figure 2: Underactuation concept of U-HEx during a grasping task: (left) initial pose with all finger joints extended, (mid) actuator force moves the MCP joint until the first finger phalange gets in touch with the object, (right) actuator force is transmitted to move the PIP joint when the first finger phalange touches the object and finally grasping task is completed when both phalanges touch the object
• Figure 3: Optimization process for U-HEx: for a given set of link lengths, a Simulink model is executed to compute numerical inverse kinematics and analytical statics as the finger joints are iterated from fully open to fully closed; while the optimization algorithms are implemented in MATLAB.
• Figure 4: Force transmission (optimality) and convergence time comparisons between experiments (with a different number of variable link lengths): with the constraint to the maximum actuator displacement ($con_7 = L_x \leq 50$) and without akbas2024impact. Plots report the mean and the standard error.
• Figure 5: A zoomed-in representation of force transmission (optimality) with the linear actuation constraint ($con_7$) as $L_x \leq 50$. The plots show the median, interquartile range, and outliers of the data.