Additive Manufacturing of Soft Robots based on Graded Porous Structures
Nick Willemstein is a PhD student in the department Biomechatronics and Rehabilitation Technology. (Co)Promotors are prof.dr.ir. H. van der Kooij and dr. A. Sadeghi from the faculty Science & Technology.
Soft robotics is a relatively new approach to designing robots that incorporate soft and smart materials. This approach was inspired by the (soft) octopus tentacle and (soft) elephant trunk, which can perform complex tasks and seamlessly incorporate actuation, sensing, control, and more in a single structure. Researchers have been exploring unconventional materials such as rubber, textile, and foam for these integrated structures.
This dissertation explored the Additive Manufacturing (AM) of porous/foam-like structures to integrate sensing and actuation for soft robots and sensors. These foam-like structures have the advantage of being lightweight, allowing for fluid transport, and stiffness regulation. These features make porous structures a promising platform for soft robotics, while AM provides (among other) multi-material capabilities and geometric freedom.
To have a soft base material, we developed our own dual pellet extruder to print thermoplastic elastomers (TPEs). This pellet extruder could 3D print soft airtight membranes (0.2-1.2 mm) from TPEs down to Shore Hardness 00-30, which could be inflated up to a stretch of 1320%, and utilized for a bending actuator and a membrane-based sucker and gripper.
We then developed the InFoam method to directly fabricate porous structures using TPEs. This method used the liquid rope coiling effect, which is the coiling seen when dropping a viscous liquid such as honey from a height, to create user-defined porosity gradients. Specifically, we established an empirical relationship between the process parameters (nozzle height and extrusion speed) and the resulting coiling pattern. We observed that the porosity magnitude could program the mechanical and actuation properties, which included changing the modulus by more than two orders of magnitude. We combined this large modulus change with normal 3D printing to create bending, contracting, and twisting actuators.
Next, we used an electrically conductive TPE to add piezoresistive sensing to the foam-like/porous structures. These 3D printed piezoresistive sensing foam-like structures were used for multiple applications: a sensorized insole for measuring ground reaction forces, a wide range soft tactile sensor by using a layered porosity gradient, and sensorized actuators (incl. a 3DoF bending segment). The resistance changes of these sensors exhibited nonlinearities and hysteresis, which made estimating the stress/force or strain challenging. To compensate for these nonlinearities, we explored the system identification of Hammerstein-Wiener models, which gave good results (on average RMSE <7%). In addition, through characterization, we observed that the porosity had a power-law relationship with multiple mechanical, actuation, and sensing properties.
Overall, this thesis contributes to AM of soft robotics based on foam-like/porous structures by (i) the design/usage of pellet extruders, (ii) exploring nonlinear system identification for foam-like structures, (iii), developing the InFoam method, and (iv) investigating the empirical relation between porosity and sensing/actuation behavior.
