Versatile 6-DoF robotic cutting for multi-material objects

Establish methods that enable versatile, high-degree-of-freedom robotic cutting for multi-material objects composed of distinct deformable materials, specifically achieving full six-degree-of-freedom control of a cutting tool through complex three-dimensional geometries while accommodating the heterogeneous mechanical properties of the materials involved.

Background

The paper surveys prior work showing that most robotic cutting approaches assume low-dimensional action spaces (1D or 2D) and single-material objects, often relying on open-loop or planar cuts. Recent learning-based systems, such as RoboNinja, handle multi-material cutting but remain constrained to 2D workspaces, which is insufficient for tasks like chicken shoulder deboning requiring full 6-DoF control.

The authors emphasize that navigating intricate geometries and dynamically adapting a cutting tool’s full six-degree-of-freedom pose is required for realistic multi-material tasks. Despite advances, achieving versatile, high-DoF cutting that generalizes across multi-material objects is not yet solved, motivating the explicit open problem.