Date of Award

Summer 8-22-2025

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science in Mechanical Engineering (MSME)

Department

Mechanical Engineering

First Committee Advisor

Babak Hejrati

Second Committee Member

Andrew Goupee

Third Committee Member

Vincent Caccese

Abstract

Gait speed is a critical indicator of health and independence in older adults, yet it often declines with age. This decline is strongly correlated with an increased risk of falls and is primarily characterized by a decrease in stride length. Wearable haptic feedback has emerged as a promising tool for gait retraining to address this deficit by encouraging a longer stride, which is associated with faster, safer walking. However, the efficacy of such systems depends on setting appropriate and personalized goals. Our previous work utilized static feedback targets derived from a user’s maximum performance during a short fast-walking trial, an approach that proved to be a key limitation. In some cases, targets were often unsustainable for longer periods of walking, while in others, targets were not sufficiently challenging. This study, therefore, investigated the feasibility and effectiveness of a novel human-in-the-loop (HIL) optimization algorithm designed to overcome these shortcomings by dynamically personalizing peak thigh extension (PTE) targets during gait training. To evaluate this new approach, ten community-dwelling older adults participated in a single-session study. Participants were equipped with a wearable system consisting of thigh-worn inertial measurement units (IMUs) and haptic feedback modules. The system was controlled by the HIL algorithm, which monitored real-time gait performance to incrementally adjust the PTE target for each leg independently. The target was increased by a fixed 2◦ increment when a user exceeded their current target on at least 80% of the preceding 20 strides, and decreased by the same 2◦ if their success rate fell below 20%. The algorithm also incorporated a critical safeguard to monitor cadence, providing corrective anterior feedback if it detected an abnormally slow, elongated gait pattern. This ensured that improvements in stride length did not come at the expense of a natural walking rhythm. The results demonstrated that the HIL algorithm is a viable and highly effective method for gait retraining. The adaptive feedback led to significant increases in participants’ PTE, stride length, and overall walking speed, all achieved without a significant change in cadence. The central finding was the algorithm’s success in personalizing the intervention; it converged on final targets that were variably higher, lower, or similar to what would have been set by our previous static method, confirming its ability to adapt to individual capabilities in real-time. A key secondary finding related to gait symmetry. While on a group level the feedback did not significantly alter symmetry, considerable individual variability was observed. This was attributed to two factors: the algorithm’s design, which allowed targets for each leg to adapt independently, and the biomechanical demands of the oval track used for testing, which naturally required greater extension of the outer leg during turns. In conclusion, this study validates the HIL algorithm as a superior approach for personalizing haptic gait training, overcoming the critical limitations of static target-setting. The adaptive system effectively tailored challenges to the individual, successfully improving key spatiotemporal gait parameters. However, the findings also highlight that without constraints, such a system may not inherently improve, and could in some cases worsen gait symmetry. This work underscores the need for future research to focus on refining the algorithm to manage and promote symmetric movement explicitly. Future iterations should explore unified targets and develop context-awareness to create a more robust clinical tool. This study represents a significant step toward developing more intelligent and effective wearable technologies for gait rehabilitation.

Download

Share