Adaptive Shape Servoing of Elastic Rods using Parameterized Regression Features and Auto-Tuning Motion Controls

The robotic manipulation of deformable linear objects has shown great potential in a wide range of real-world applications. However, it presents many challenges due to the objects' non-linear properties and high-dimensional geometric configuration. In this paper, we propose an efficient shape servoing framework to manipulate elastic objects through real-time visual feedback automatically. The proposed parameterized regression features are used to construct a compact (low-dimensional) feature vector (Bézier and NURBS) that quantifies the object's shape, thus enabling the establishment of an explicit shape servo-loop. To automatically manipulate the object into a desired configuration, our adaptive controller can iteratively estimate the sensorimotor model that relates the robot's motion and shape changes. This valuable capability enables the effective deformation of objects with unknown mechanical models. An auto-tuning algorithm is introduced to adjust the controller's gain and, thus, modulate the shaping motions based on optimal performance criteria. To validate the proposed framework, a detailed experimental study with vision-guided robot manipulators is presented.