Testing and Advancement of a Variable Stiffness Transverse Plane Adapter for Use in a Lower Limb Prosthesis

Thesis (Ph.D.)--University of Washington, 2017-06Abstract Testing and Advancement of a Variable Stiffness Transverse Plane Adapter for Use in a Lower Limb Prosthesis Corey Pew Chair of the Supervisory Committee: Glenn Klute, Ph.D. Affiliate Professor Department of Mechanical Engineering Background The goal of a lower limb prosthesis is to restore the abilities of the intact limb for an individual with lower limb amputation. Daily ambulation includes many maneuvers such as turning, and twisting, which require a component of transverse plane mobility. It has been shown that the inclusion of a transverse plane adapter could reduce peak torsional loads on the residual limb and may alleviate soft tissue damage, increase comfort, and improve mobility level for a lower limb amputee. However, currently available transverse plane adaptors only allow for a single stiffness setting and do not allow for variation to accommodate the maneuvers of everyday ambulation. The specific aims of this research were to determine the transverse plane stiffness that minimizes the transverse plane moment applied to the residual limb of lower limb amputees during different ambulatory activities and identify a user’s preferred transverse plane stiffness during different ambulatory activities at different speeds. Three tasks were performed to achieve these aims. First, proof of concept human subject performance tests were conducted using the first-generation prototype (VSTA I). Next, a second-generation prototype (VSTA II) was developed using more refined design criteria gleaned from the testing of the VSTA I. Third, the VSTA II was used to perform preference testing with human subjects during different ambulatory activities at varying walking speeds. First Generation VSTA Design (VSTA I) Design of the VSTA I focused primarily on existing transverse rotation adapter devices and the capabilities of other prototypes in recent literature. The VSTA I is a device that allows for variable stiffness in transverse plane about its central axis. Variation of the VSTA I is accomplished by an electric motor adjusting the position of a lever arm via an ACME lead screw. The change in position of the lever arm in relation to a central torsion spring effectively varies the mechanical advantage by modifying the lever arm length. This allows for infinitely variable stiffness between 0.10 Nm/° and fully locked. VSTA I Mechanical Testing Mechanical bench testing was performed on the VSTA I via a servo-hydraulic material test machine. The VSTA I was run through its full range of motion (+/- 30°) at both 0.5 °/s and 60°/s at four static stiffness settings: 0.10, 0.33, 0.64, and 1.17 Nm/°. It was found that the VSTA I was not direction or rate dependent and was capable of stiffness variation between 0.12-0.91 Nm/°, with a rotational range of ±30°. VSTA I Human Subject Testing The first generation VSTA (VSTA I) was used to completed pilot testing on human subject individuals with lower limb amputation. These tests conducted participants through simulated activities of daily living that focused on turning and twisting maneuvers. These included straight walking, 90° turns (spin and step), 180° turns, standing reach, and the L-Test of Functional Mobility. Testing was performed at three constant settings of the VSTA I (compliant: 0.30 Nm/°, intermediate: 0.57 Nm/°, stiff: 0.91 Nm/°). Evaluation of the testing focused primarily on peak transverse plane moments during each maneuver. It was hypothesized that reduced transverse plane loading on the residual limb relates to decreased transverse plane stiffness. Additionally, the self-selected walking speed and L-Test time of each subject was used as an indicator of mobility. Results indicated that activities requiring high levels of transverse plane motion (90° spin and 180° turns) had significantly reduced peak transverse plane moments at the socket when walking with the compliant transverse plane stiffness as compared to the stiff setting. Additionally, use of the VSTA resulted in no measurable loss of mobility at self-selected walking speeds between the three settings. These preliminary results indicate that a transverse rotation adapter with variable stiffness capability could be useful for a lower limb amputee to help reduce stresses at the socket-limb interface. Testing of the VSTA I also resulted in an updated set of design requirements. It was found that the VSTA I was too heavy, causing fatigue, and too tall, restricting subject population. It also lacked an onboard controller, and suffered from internal deflections that limited its stiffness capabilities. Second Generation VSTA Design (VSTA II) Given the lessons learned during the testing of the VSTA I, a new and improved VSTA II has been designed. The VSTA II features a 42% reduction in height and 51% reduction in mass compared to the VSTA I with an intended finite stiffness variability from 0.30 Nm/° to 1.25 Nm/°, in five 0.25 Nm/° increments. Stiffness variation is enabled by five independent spring subunits that can be combined in parallel to create different, linear, stiffness settings. It also features an onboard controller that can control the stiffness states, and record rotational displacement and stiffness setting during use. VSTA II Mechanical Testing Mechanical bench testing performed on the VSTA II mirrored that of the VSTA I. Deflection testing performed on a servo-hydraulic material test machine through its full range of motion (+/- 30°) at both 0.5 °/s and 60°/s. Additionally, tests were performed to determine the VSTA II’s ability to modulate stiffness in real time and determine power consumption during operation. The VSTA II was found to be capable of five discrete stiffness settings (0.31, 0.56, 0.83, 1.08, 1.29 Nm/°) with ±30° of motion in addition to fully locked operation. Stiffness settings were found to be independent of rotation rate and direction. Stiffness selection performed via small locking mechanisms operated by electro-mechanical solenoids can vary the stiffness of the VSTA II in 0.029 ± 0.008 seconds while the device is unloaded during the swing phase of gait. Lastly, the 2200 mAh battery was found to last 105 minutes under full electrical load. Turn Intent Prediction Daily use of the VSTA would require an autonomous control algorithm to determine optimal stiffness for a given activities and modulate the device accordingly via the onboard controller. Ideally, IMU signals from the shank of the lower limb prosthesis could be used to identify upcoming changes in activity with adequate time to modulate stiffness before the foot is loaded during a turn (changes in VSTA stiffness performed under no load during swing). Three classifiers were trained and tested: support vector machine (SVM), K nearest neighbors (KNN), and a bagged decision tree ensemble (Ensemble). Training for an individual gave superior results over training on a pooled set of five individuals. Coupled with a simple control scheme, the SVM, KNN, and Ensemble classifiers could attain 96%, 93%, and 91% accuracy (no significant difference), respectively, predicting an upcoming turn 400 ± 70 ms prior to the heel strike of the turn. However, classification of straight walking transition steps varied between classifiers at 85%, 82%, 97% (Ensemble significantly different, p = 0.002), respectively. The Ensemble model produced the best result overall, however, depending the priority of identifying turning vs transition steps and processor requirements, the SVM or KNN might still be considered. VSTA II Human Subject Testing Testing to determine ideal settings for control of the VSTA II consisted of paired preference testing at varying activity and walking speed combinations. Subjects performed walking trials for three separate activities (straight walking, prosthesis inside turning, prosthesis outside turning), at their self-selected walking speed and at speeds 20% faster and slower than self-selected. At each activity-speed combination subjects compared stiffness settings and indicated their preference for one of three settings (Compliant: A (0.31 Nm/deg), Moderate: B (0.83 Nm/deg), Stiff: C (1.29 Nm/deg). Results indicated correlation between increased peak transverse plane moment with increasing walking speed when turning, but not when walking straight. However, contrary to the outcome of previous findings, no significant relation was found between peak transverse plane moment and the stiffness of the VSTA II. Lastly, results indicated no significant trend for stiffness preference between speeds or activities, while subjects did qualitatively prefer lower stiffness when turning vs straight walking. Findings may indicate that no global ideal stiffness settings may be available given activity and speeds variations, but that VSTA II settings may have to be tailored to individual users. Conclusions Testing with the VSTA I and VSTA II prototypes identified that reduced transverse plane stiffness in the shank of a lower limb prosthesis can significantly reduce peak loading on the residual limb. Moreover, that transverse plane stiffness could be reduced beyond what is available in current, single stiffness, transverse rotation adapters. Reductions in stiffness also have no effect on the user’s mobility when walking at their self-selected speed. Initial attempts were made to determine a global control scheme that could indicate when and to what level stiffness should be adjusted, however, no overall trend could be found. It was determined that reduced transverse plane stiffness produces the most significant reductions in limb loading when turning compared to straight walking. Additionally, walking speed played a significant role in transverse plane limb loading such that both factors should be considered when determining the optimal transverse plane stiffness for a given activity and speed combination. Overall a control scheme that optimizes when and to what level to adjust transverse plane stiffness should be customized to an individual’s specific needs and preferences