Session C: 1:45PM – 3:15PM
Engineering Session C- Oral Presentations, Den, Union
SESSION C (1:45PM – 3:15PM)
Location: Den, A. Ray Olpin University Union
Mobile Base for Physical Human-Robot Interaction and Co-manipulation
Nate Ludlow, Brigham Young University
Faculty Mentor Marc Killpack, Brigham Young University
SESSION C 1:45-2:00PM
Den, Union
Engineering
There is a lack of high-payload, low-cost mobile robot bases with high enough levels of mobility to perform effective human-robot manipulation. This research aims to provide an open-source option for a mobile base that can aid humans in carrying large and heavy objects. We designed, constructed, and developed code for an omnidirectional mobile base to enable research in human-robot co-manipulation applied to search and rescue situations. The base was designed with four individually steerable caster wheels to facilitate omnidirectional movement, allowing it to move in any direction to dynamically match the motion of a human. These casters are mounted on differential rocker arms to allow the base to move over small to mid-sized obstacles and objects on the ground. This should allow the base to be better suited for traversing rough terrain often found in search and rescue operations. The base is equipped with a pneumatically actuated soft continuum robot arm attached to a rotating turret to interact with the object being carried. This research aims to extend or build on the results of Freeman [BYU Scholars Archive, 9433 (2022)] and Jensen [Frontiers in Neurorobotics, 15, 626074 (2021)] by providing a testing platform for human-robot co-manipulation experiments using the human-human data co-manipulation data gathered. The mobile base can achieve speeds of greater than 3 m/s (6.7 mph) in any direction and can carry a payload of greater than 68kg (150lbs) making it suitable for human-robot co-manipulation of large objects and heavy loads. It was additionally able to run on battery power for over an hour and can traverse uneven ground. The future trials run with this mobile base will provide greater insight into the parameters involved in effective co-manipulation of heavy objects and will be used to for insight into robot assisted search and rescue tasks.
Engineering Outreach with Soft-Robotics: How to Design a Lesson in Design
Haylee Sevy, Brigham Young University
Faculty Mentor Marc Killpack, Brigham Young University
SESSION C 2:05-2:20PM
Den, Union
Engineering
Starting in the Fall of 2021, the BYU Robotics and Dynamics lab started working on a soft robotics outreach curriculum originally developed by the Fabratory at Yale. The activity the curriculum is centered around involves making a pneumatically driven soft-robot gripper out of silicone. The basic construction involves assembling a 3d printed mold to define the shape designed by the student(s), filling the mold with silicone to create the body of the gripper, and attaching the silicone gripper to a piece of fabric to create closed air-channels. The project itself was shown to have a high attention retention rate with students, as seen from data collected from surveys taken by students who participated, but the grippers that they created did not have a high success rate in terms of longevity or functionality. We explored many design iterations to increase the rate of successful outcomes with the outreach activity. Variables that were changed included the design of the mold, the type of silicone used, and the way the instructions are presented. We were able to simplify and shorten many of the steps in the manufacturing process that students would need to go through. We also reduced the prep work needed beforehand. Overall, these changes will make the activity easier for K-12 teachers to conduct in their classrooms, as well as give young students a more positive interaction and association with engineering and robotics.
An Automated, Parameterized Model of Maize Stalk Strength via Machine Learning
Ryan Hall, Brigham Young University
Faculty Mentor Douglas Cook, Brigham Young University
SESSION C 2:25-2:40PM
Den, Union
Engineering
A fully parameterized model of the maize stalk morphology was created using machine learning techniques. A database of 1000 CT scans of maize stalks served as the training data. The model consists of over 50 geometric parameters and 14 physical material properties. The parameterization scheme allows independent control of each physical feature of the stalk. This was accomplished by linking key landmarks with empirical eigenfunctions to capture morphological patterns in the transverse and axial directions. The parameterized model was validated by comparing results of models based on actual maize stalk shapes with parameterized counterparts in multiple loading scenarios: axial, torsion, bending, transverse compression, flexural stiffness, and ultimate bending strength. The resulting model accurately captures behavior of actual stalks, can be “fit” to any specimen, and can be used to perform sensitivity and optimization studies. The model creation, validation, and preliminary sensitivity results will be presented.
Case Study: Powered Hip Exoskeleton Reduces Metabolic Cost of Walking in Individual with Hemiparesis
Kai Pruyn, University of Utah
Faculty Mentor Tommaso Lenzi, University of Utah
SESSION C 2:45-3:00PM
Den, Union
Engineering
Every year about 795,000 Americans suffer a stroke. Though the mortality rate has decreased, stroke commonly results in physical disability due to hemiparesis, muscle weakness in one side of the body. Hemiparesis causes stroke survivors to struggle with mobility as it results in poor balance, reduced range of motion, and early onset of fatigue. Powered exoskeletons have been proposed as a potential solution to this problem. Powered exoskeletons are wearable devices that support the movements of the user by providing assistance from electric motors. Exoskeletons have been successful in assisting healthy subjects who have a consistent and symmetric gait. However, the ability of a powered exoskeleton to assist the wearer depends on the capability of the exoskeleton controller to synchronize with the biological movements of the user. Hemiparetic subjects have asymmetric gait patterns that make it challenging for the exoskeleton controller to synchronize with the user. Therefore, developing assistive controllers based on the hemiparetic gait is necessary for powered exoskeletons to effectively assist stroke subjects. The goal of my undergraduate research is to address the problem of human-robot coordination in the context of assistive powered exoskeletons for individuals with hemiparesis. To this end, I have developed a controller that provides flexion and extension assistance to one or both of the user’s hip joints. The assistive torque profile and timing is adjusted based on each subject’s gait cycle and where they need the most support. The controller tracks the wearer’s previous steps to predict their next ones, providing consistent and reliable exoskeleton assistance. Each side can be controlled independently to deliver different amounts of assistance with different timing to a subject’s hemiparetic side and unaffected side. To preliminarily verify the viability of this assistive control approach, I conducted a case study with one individual with hemiparesis walking with and without a powered hip exoskeleton while we measured metabolic cost. The participant walked on a treadmill at 0.9 ms-1 for a series of 6-minute-walk-tests (6MWT): one without the exoskeleton and three with the exoskeleton. The exoskeleton provided bilateral flexion and extension assistive torque to both hip joints. Our results show that the powered hip exoskeleton reduced the metabolic cost of walking by 34%. This reduction is equivalent to removing a 58-lbs backpack for a healthy individual. The results of this study will inform future metabolic and clinical study designs, focusing on assisting stroke survivor ambulation with the powered hip exoskeleton.
