@phdthesis{escobedo26thesis,
  title        = {Whole-Body Robot Skins for Near-Body Perception and Safe Interaction},
  author       = {Escobedo, Caleb Sebastian},
  year         = {2026},
  month        = may,
  school       = {University of Colorado Boulder},
  address      = {Boulder, CO, USA},
  type         = {Doctor of Philosophy Dissertation},
  abstract     = {Human skin is a distributed sensory organ that lets us perceive proximity, contact, and the structure of the world across the entire body, an ability so fundamental to acting safely among people that robots cannot truly share our spaces without it. Yet robots still rely on sparse, task-specific sensors that leave most of their body blind, and the whole-body skins that could close this gap remain hand-built, one-off systems confined to a handful of labs. This dissertation argues that the central challenge in physical human-robot interaction is making whole-body sensing something that can be fabricated, grounded on the body, and deployed repeatably across any robot. I address this challenge at its foundation, in fabrication. I present a method for producing whole-body robotic skins through multi-material additive manufacturing (3D printing), integrating diverse sensing modalities into rigid and compliant structures that conform directly to any robot surface. Because sensor locations are set in the digital model and preserved by the print, a robot knows where its sensing data originates on its own body without per-sensor recalibration. Using this method, I fabricate skins spanning capacitive, time-of-flight, hybrid, and magnetic sensing and deploy them across a humanoid, a quadruped, and a robot arm. Building on this foundation, I show how the distributed sensing on these skins is calibrated and integrated into perception of the near-body space, and how object-aware controllers use that perception to anticipate and avoid contact during physical interaction. Together, these contributions turn whole-body sensing from a one-off engineering effort that only a few labs can take on into a capability any roboticist can reproduce and extend. What has held whole-body sensing back is what each system demands to build: specialized hardware knowledge, custom wiring, hand-placed sensors, and months of integration for a result that rarely transfers to another robot. Removing that burden reframes whole-body perception as shared infrastructure for robotics, a foundation that researchers in manipulation, learning, humanoids, and human-robot interaction can build on without reinventing the hardware, and a concrete step toward robots that leave their isolated cells to work safely alongside people in the everyday physical world.}
}