Legged locomotion in soft robotics has largely relied on tendon-driven systems or elastomeric actuators with limited deformation range. However, origami-like soft actuators fabricated following a layer-by-layer approach offer a distinct mechanical profile: large length changes upon pressurization, inherent collapsibility, and origami-like reconfigurability that makes them well-suited as locomotive limbs. These same properties, however, make them significantly harder to model and control. Unlike elastomeric actuators, they undergo large geometric changes during inflation, exhibit highly compliant contact behavior, and lack established modeling frameworks that can capture their interaction with the ground.

Physical intelligence in soft robotics explores how computation, sensing, and control can emerge from the physical dynamics of materials and embedded fluids, rather than relying solely on centralized digital computation. My research investigates fluidic logic as a substrate for mechanical computation, using pressure and airflow as information carriers to construct circuits that implement functions such as switching, gating, and state retention without conventional electronics. This includes the fabrication of sheet-based fluidic diodes, which bias airflow or pressure in a preferred direction and serve as fundamental building blocks for these logic architectures. In parallel, I have also contributed experimental systems used to study how pressure dynamics in soft actuators can be leveraged to understand system behavior, where temporal pressure responses reflect deformation, contact, and internal state evolution.

Electropermanent magnets are a class of permanent magnets whose magnetic state can be switched with a brief current pulse, requiring no continuous power to maintain. Despite this advantage, their integration into soft robotic systems remains limited. My research explores the practical implementation of EPMs across two soft robotic applications: (i) as low-profile, onboard valves that route pressurized fluid across multiple actuator channels without the bulk of traditional valve banks, and (ii) as jamming actuators that stiffen metal sheet assemblies enabling variable-stiffness structures suited for wearable and haptic devices without the pneumatic infrastructure that conventional jamming requires. Together, these two use cases demonstrate EPMs as an electronic-based control primitive for untethered soft robotic systems and devices.