Search for: All records

Soft robots, due to their flexibility, adaptability, and gentle handling over rigid robots, have shown better potential in numerous applications requiring operating in constrained spaces. Most of the soft robotic prototypes are of a linear form that can be modeled as a curve in space and are found in manipulators and limbs of locomoting robots. Planar soft robots have been proposed recently that are modeled as a surface and deform in 3D. Research on planar soft robots has been less extensive due to the challenges associated with modeling surface deformations efficiently. We present a curve-parametric approach for the deformation modeling of planar soft robot modules. Along with the Bezier patch method to approximate the surface at 30 Hz. Experimental evaluations on a prototype were developed and tested to validate that the proposed model can reasonably approximate the planar robot boundaries, and the surface derived from it. 

Soft robotics has witnessed increased attention from the robotic community due to their desirable features in compliant manipulation in unstructured spaces and human-friendly applications. Their light-weight designs and low-stiffness are ideally suited for environments with fragile and sensitive objects without causing damage. Deformation sensing of soft robots so far has relied on highly nonlinear bending sensors and vision-based methods that are not suitable for obtaining precise and reliable state feedback. In this work, for the first time, we explore the use of a state-of-the-art high fidelity deformation sensor that is based on optical frequency domain reflcctometry in soft bending actuators. These sensors are capable of providing spatial coordinate feedback along the length of the sensor at every 0.8 mm at up to 250 Hz. This work systematically analyzes the sensor feedback for soft bending actuator deformation and then introduces a reduced order kinematic model, together with cubic spline interpolation, which could be used to reconstruct the continuous deformation of the soft bending actuators. The kinematic model is then extended to derive an efficient dynamic model which runs at 1.5 kHz and validated against the experimental data.