Bio-Inspired Contact Dynamics Modeling for Agile Locomotion on Realistic Sand

This Dynamics, Controls, and System Diagnostics (DCSD) project will translate the underlying locomotion principles used by bipedal desert rodents into mathematical tools that engineers can use to design more capable intelligent and autonomous machines. The strategies used by these agile animals to move quickly and nimbly across loose sand have been refined over millions of years of evolution, yet remain poorly understood and underutilized in engineering applications. As a result, today’s walking robots must cross sandy terrain slowly and cautiously, limiting their usefulness on beaches, deserts, and in extraterrestrial settings such as the Moon or Mars. Developing intelligent robots that can move rapidly, reliably, and efficiently on sand will enable high-performance operation in these challenging environments, facilitating important functions such as exploration, emergency response, and delivery of essential services. Connecting biology, physics, and engineering to model and analyze the dynamics of small desert animals, this project will strengthen the nation’s ability to operate robots in challenging natural and extraterrestrial environments, in contrast to current designs that sink, slip, or waste energy simply staying upright. The project also trains undergraduate and graduate students across several disciplines and continues an international virtual exchange program that gives future science and engineering instructors classroom teaching experience, providing access to the technical workforce for all Americans. The research plan pursues three connected objectives. The first develops a vision-based, markerless method for estimating the joint motion of small, fast-moving desert rodents as their feet interact with granular substrates whose properties are systematically varied; the method is validated against motion capture measurements on physical models before data are gathered from live animals, yielding a diagnostic platform that supplies empirical data for modeling and control. The second objective confronts the tradeoff between model accuracy and computational speed by first constructing a high-fidelity, multi-scale physical model of foot interaction with natural sand and then systematically reducing its complexity to produce control-oriented models suitable for real-time use; these reduced models are validated across parameter regimes using three-dimensionally printed biomimetic feet. The third objective demonstrates applicability through trajectory optimization for a physical leg mounted on a rail-based sand tank. The leg is deliberately underactuated to emulate the lightweight, agile character of its biological counterpart, requiring control methods that remain robust under limited actuation. Comparing measurements from this platform with reduced-model predictions and with data from live animals validates and refines the models across varied conditions. The central contribution is a set of mathematical representations and control strategies for nonlinear interactions between bodies and deformable granular media, advancing control theory for systems that operate on yielding terrain. 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.