Vision-based Control of Soft Continuum Robots
Soft continuum robots mark a revolutionary shift in robotics, thanks to their flexible mechanics and hyper-redundant degrees of freedom. They present great potential for applications including construction, industrial inspection, medical robotics, and assistive devices. However, accurately controlling these robots is challenging due to the difficulties in obtaining high fidelity models and integrating proprioceptive sensors. This dissertation presents novel model-free control algorithms that do not require proprioceptive sensor feedback from the robot. Instead, the developed methods utilize purely visual information, observed on the robot’s body from an external camera, to control the robot. Moreover, these methods allow us to leverage the kinematic redundancy of continuum robots by enabling their whole-body control, unlike existing methods in literature that focus on regulating only the end effector pose of robots. This is achieved by observing shape features along the robot’s body instead of exclusively observing features located at the robot’s end effector. Leveraging the kinematic redundancy of continuum robots enables their safe operation in cluttered and constrained environments. Additionally, a novel model-based control algorithm is presented to improve the transient control response. Detailed experimental studies are performed to characterize the control convergence, transient response, robustness, repeatability, and applicability of the proposed control algorithms.
Advisor: Professor Berk Calli
Committee: Professor Loris Fichera, Professor Cagdas Onal, Presentation Zhu Mao