UC Berkeley

For forceful and dexterous manipulation, robots must be able to physically interact with the environment. In the real-world, the environment is dynamic and unstructured. For example, fragile objects and adversarial contact conditions, e.g. a slippery fluid film, may be present. Embodied intelligence and active control strategies are two approaches that tackle these challenges. Soft robots and grippers exemplify embodied intelligence, as the inherent behavior of soft materials allows passive compliance and distributed loading. A sense of touch, also known as tactile sensing, provides feedback for active controllers to provide additional adaptability and articulation. Human hands seamlessly integrate both compliance and tactile sensing, with underactuated joints and soft compliant skin covered with nerve endings. While case studies demonstrate soft robots or tactile sensors in the field, few systems have both.

My dissertation explores the design of embodied dexterity and tactile sensing in robots through fluids. Fluids – liquids, gasses – are ubiquitous. Many soft robots use fluids for actuation, e.g. pneumatic or hydraulic. Fluids for robotic sensing and contact design are less prevalent. In this dissertation, I first present how we can employ airflow for tactile sensing. This design uses pneumatic acoustic resonance to sensitize soft, skin-like surfaces and grippers. I then show how we can mitigate the slippery effects of lubrication for frictional grasping by adding soft, patterned finger pads to existing grippers.