MorphIt: Flexible Spherical Approximation of Robot Morphology for Representation-driven Adaptation

--What if a robot could rethink its own morphological representation to better meet the demands of diverse tasks? Most robotic systems today treat their physical form as a fixed constraint rather than an adaptive resource, forcing the same rigid geometric representation to serve applications with vastly different computational and precision requirements. Our experiments show enhanced robot capabilities in collision detection accuracy, contact-rich interaction simulation, and navigation through confined spaces. By dynamically adapting geometric representations to task requirements, robots can now exploit their physical embodiment as an active resource rather than an inflexible parameter, opening new frontiers for manipulation in environments where physical form must continuously balance precision with computational tractability . The morphology of a robot, defined as its physical form and structure, serves as the fundamental basis for all physical interactions with the world. This morphology must be represented computationally to enable effective navigation, manipulation, and physical reasoning capabilities. During runtime operation, physical representation becomes particularly critical as robots must constantly compute distances between their bodies and surrounding objects to detect potential collisions or reason about contacts. These distance queries typically consume as much as 71% of the total computation time in motion planning pipelines, creating a significant performance bottleneck [1-3]. When robots with complex geometric models interact with numerous objects simultaneously, these calculations quickly become intractable for real-time applications. All authors are with the Human Interaction and Robotics [HIRO] Group at the University of Colorado Boulder.