SpongeBot
Soft Robotic Gripper for Caregiving Tasks
2024
Building on my previous work with soft grippers and robotic arms, SpongeBot combines elements from my research and expands into new areas through collaboration with students from the ILIAD Lab and REAL Lab at Stanford. This project was created for the Stanford Robotics Center launch in November 2024.
Advances in commercial 3D printing—specifically from Carbon3D—now enable the streamlined production of rubber-like materials. In this project, we focused on human-robot interaction in caregiving scenarios where physical contact is both inevitable and necessary for proper task execution—for example, bathing, where force must be applied to the skin.
The gripper fingers’ bending mechanism was designed using finite element analysis (FEA), followed by rapid, low-cost soft 3D printing with Formlabs and high-end Carbon3D printing for durability. Initial single-direction bending designs were replaced with a three-linear actuator configuration to enhance dexterity and enable nearly 360° motion per finger. I iteratively optimized bellows length and diameter, and explored the addition of a “knuckle” between actuator sets for improved bending control. Ultimately, a slightly elongated single phalanx was selected to increase reach without compromising grip strength.
I designed the gripper’s key components—the end effector and the base with air channels for individual finger control—to ensure a firm, secure grasp while maintaining user comfort. During bathing tasks, shear forces can dislodge a towel from the grip. To prevent this, the end effector was given a textured surface and a thumb shape, enhancing natural contact between the gripper and the user. By analyzing pressure fluctuations in the gripper, we provide haptic feedback that informs the arm’s control policy, ensuring gentle, continuous contact during a bathing demonstration.
My work is based on the theory that people feel more comfortable interacting with materials that have elasticity similar to human skin, suggesting that soft robots are better suited for physical contact. To test this hypothesis, I conducted a bathing demonstration with the gripperand carried out an extensive user study. During the Stanford Robotics Center launch in November 2024, over 100 participants aged 18 to 89 took part in the demonstration and completed pre- and post-experiment surveys assessing their comfort and trust in robots, both generally and for caregiving tasks. The results showed a significant increase in trust, with trust levels rising by one point on a 1–7 Likert scale from before to after the demonstration. Participants also reported that the interaction felt natural and comfortable, contrary to their initial expectations.
