Vision-Based Cutting Control of Deformable Objects

Cutting control of deformable objects is of great importance in varied application fields such as surgical robot and food processing industry. However, complex physical properties of deformable objects and time-variant topology during cutting make automatic cutting operation control a big challenge. This paper proposed a vision-based cutting control method that predicts the object's deformation and plans cutting path online. Material parameters of the deformation model are treated as unknown parameters to be estimated with visual measurements of feature points on the object's surface. By visually supervising three non-collinear points on the knife, the control of knife's motion is transformed to a multi-points' visual tracking control problem to achieve desired cutting effect in terms of the desired cutting depth and the minimized deformation of the object. The feasibility of the proposed method is validated by experiment results.