MATH-DT: Functional Surrogate Modeling for Human Agile Locomotion with Wearable Technology

About This Grant

Warfighters must maintain agility and performance in extreme conditions such as navigating rugged terrain, carrying heavy loads, and enduring prolonged exertion, often while facing unpredictable threats. Wearable technologies like robotic exoskeletons and advanced footwear have the potential to enhance warfighter performance and reduce injury risk. However, current design methods often rely on one-size-fits-all approaches and fail to account for how individuals adapt to these devices in real-world settings. This project addresses that gap by developing Digital Twins of agile locomotion in the form of personalized, data-driven simulations that model the complex and dynamic interaction between human movement, wearable technology, and the environment. By integrating real-time physiological and biomechanical data, these models enable better design, training, and deployment of active wearable technology to improve human agility. In addition to advancing national defense and security, this work has broad societal benefits to public health as the mathematical modeling techniques developed can also be used to improve wearable technology design for other user populations, such as those with motor impairment. The overarching goal of this project is to develop mathematical methods enabling an advanced Digital Twin model of human agile locomotion, aimed at optimizing the design of advanced footwear technology to enhance human agility and mobility. In order to accomplish this, this project will advance the state of the art in statistical surrogate modeling, which currently is focused on vector-valued parameters, to accommodate parameters which are functions. This will require significant mathematical and methodological innovation as the parameter spaces are thus infinite dimensional. The investigators will develop an approach which searches within a manifold of finite but increasing dimension to find candidate functions to test. This new methodology will be developed using data from human locomotion when using wearable technologies in a lab setting. The investigators will first deploy this methodology to develop a novel active model reduction method which searches for parameter settings which are not accommodated by the current reduced model. Next, they will extend sequential design for data-efficient predictive modeling to the acquisition of functions, enabling a model of human adaptation in response to wearable devices. Finally, they will develop an infinite-dimensional extension of the Active Subspace method for dimension reduction to enable interpretable optimization of wearable devices. Taken together, this work will lead to a general-purpose framework for building Digital Twins of systems parameterized by functions, as well as a specific implementation of a Digital Twin for the complex system of a human bearing wearable technology. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.