Real-time Object Reconstruction

Soft robots are experiencing a steep rise in popularity thanks to their ability to solve challenges such as compliant grasping and dexterous movement tasks with which rigid robots typically struggle. The precise control of soft and continuum robots requires knowledge of their shape, which has, in contrast to classical rigid robots, infinite degrees of freedom. Similarly, robotic manipulation of deformable objects is also a relevant open challenge, arising partly due to the objects’ variable shapes, sizes, and complex material properties such as elasticity and plasticity.
To exploit the full potential of the great variety of soft robots and deformable objects, we investigated vision-based, real-time data-driven approaches for deformable shape reconstruction and tracking.