Research
Current Projects
Jogging Stroller Mechanics
Running is a popular form of physical activity due to its cardiovascular and musculoskeletal benefits. Many parents of young children can maintain their running by using “jogging strollers,” which have improved tremendously in the last two decades, with one fixed front wheel and two rear wheels.
Studies have shown that metrics of performance, such as running economy and ratings of perceived exertion, do increase while running with a stroller. One study explored some kinematic measures but did not consider the ground reaction force (GRF) while pushing a stroller. Metrics from the GRF can be used to assess performance and injury risk.
To date, no studies have compared popular running stroller brands. This project aims to compare GRF parameters during stroller running using three different strollers, at three different price levels. It is hypothesized that metrics of impact loading may be reduced while metrics of propulsion may be higher when pushing a stroller.
See conference presentions:
Footstrike Pattern Identification
While running-related overuse injuries are multifactorial, many common injuries have been associated with high-impact loading. High-impact loading, characterized by a high vertical loading rate, is influenced by the manner with which the foot strikes the ground. While most runners strike the ground with the heel first (RFS), evidence suggests that these runners are injured more frequently than those who strike the ground with a more anterior footstrike pattern. Anterior footstrike patterns can be characterized as either midfoot (MFS) or forefoot (FFS) strike patterns, where either the foot lands flat or on the ball of the foot, respectively. Thus, in the interest of reducing overuse injuries, it may be helpful to use wearable technology to identify FSP in this population. Many new types of wearable technology use accelerometery to characterize biomechanical parameters in runners. This study proposes to apply ML paradigms on tibial accelerometry data to characterize FSP in runners. Given the success of ML in other gait-classification studies, it is hypothesized this method will successfully characterize FSP using acceleration data alone.
Footstrike Pattern Recognition Using Machine Learning on Tibial Accelerometry
Results from the first version of the ML analysis will be in a forthcoming paper.
Treadmill Retrofit
The ability to track the time of heelstrikes and toe-offs as well as the location of the center of pressure (CoP) of a user on a treadmill would be beneficial for Biomechanics research. However, turn-key instrumented treadmills can cost over $100,000. The project retrofits a treadmill with loadcells to measure the force on each leg. A PCB was designed to interface the load cells to a Vicon data acquisition box. Software was developed to convert the captured voltages from the four legs into the CoP location in post-processing.
Immersive Virtual Reality System
This project combines the motion tracking and balance board to create an immersive virtual reality environment for one user. Real-time visual information will be relayed via an Oculus Rift. The user will be able to “walk” continuously in a plane.
Step Counting and Evaluation
Activity tracking has become a popular interest among the health conscious. However, current consumer products are often inaccurate — especially at low and high speeds — and may not differentiate between different movement activities. Here, we began by developing simple open source algorithms with low-cost IMUs to count steps and validate their accuracy. Next, the quality of the step was investigated: was the step during a walk or a run? We are currently expanding on this detailed investigation. Now, we want to see if a running step hit with the forefoot or rearfoot using a light, wearable device.
Associated Papers:
Past Projects
Virtual Shuffleboard and Air Hockey
Analysis of the performance of repeated, skilled tasks can elicit information about the underlying motor control system. The stability measurement of the control system may correlate with the health of the user. These measurements may be tracked over time to monitor degradation due to a degenerative illness or monitor improvement after an intervention. Within a session, we may be able to detect when learning ends and when fatigue begins to set in.
Shuffleboard Mark I
Shuffleboard Mark II
Early Prototype Version (Summer 2016): Video courtesy MA Bianco
Thesis-Ready Mark I (Spring 2018)
More technical information is available from the ASB 2018 Regional conference and "A low-cost, open-source virtual air hockey table for human motion applications"
Inexpensive & Portable 2D Motion Capture
Tracking of lower-body joint angles during walking shows the range of motion during a gait cycle. Applying markers to limbs and using motion analysis techniques automates the process of identifying the limbs and their angle in a plane (or space). Most mainstream commercially-available systems cost over $100,000 for the camera and software system. For this project, we developed a low cost (<$1000 (without Matlab)) system for 2D motion tracking. This was employed in conjunction with a project to track the range of motion of the ankle and knee when a subject is wearing an Ankle-Foot Orthosis (AFO). The project is being developed to work in real time and extended into 3D motion tracking using multiple cameras.
More technical information available from the ASB 2015 conference. The detection algorithm was adapted for the aforementioned step counting work.
Daily Fantasy Football Roster Optimization
Daily Fantasy Football has become an increasingly popular activity in the last several years. The problem translates into a Stochastic Knapsack Problem: attempting to pack as much value as possible into a constrained environment. The added difficulty is that the value of each “item” is uncertain before the sack is packed. Using machine learning and linear programming, we attempt to create lineups that outperform random and real-world competitors.
Continuous Vital Sign Monitor
Heart disease is the leading cause of death in the United States. Early detection of disease and monitoring of disease progression could prevent or delay many of these deaths. Analysis of the heartbeat rhythm has been shown to be predictive or indicative of certain cardiovascular diseases. The ability to constantly monitor a person’s heartbeat and alert the user and their doctor of a possible problem could lead to earlier disease detection. Simultaneously monitoring other vital signs (e.g. temperature and blood pressure) can provide more information about the state of health of the individual.
Balance Board for Quiet Stance Testing
Tracking and analyzing the center of pressure (COP) of a subject can indicate and track neurodegenerative disorders. Precision balance boards can cost over $10,000. These often measure 6 degrees of freedom, but for COP tracking, 1DOF is enough. This project designed, calibrated and validated a 1DOF balance board for under $500 (without Matlab). More technical information available from the ASB 2016 conference and a forthcoming validation paper.
Instrumented Bicycle
In this project, we attached sensors to a bicycle to track its position, speed and torque in real time. This information can be used to control and electronic assist motor and log the activity of the user.