A team of roboticists from Carnegie Mellon University have published a paper detailing a soft robot design for closed-loop underwater locomotion, taking cues from the brittle star — and it's called, in a nod to SpongeBob Squarepants fans the world around, PATRICK.
"Soft robots are capable of inherently safer and more stable interactions with their environment since they can mechanically deform in response to unanticipated interactions. However, their complex mechanics can make operation difficult, particularly with tasks such as locomotion, and robust systems are needed for evaluating and testing new planning and control algorithms," the team explains. "In this work, we present the first mobile and untethered underwater crawling soft robot.
"PATRICK is a robotic test-bed inspired by brittle stars that demonstrates closed-loop locomotion planning. PATRICK contains five soft legs actuated by a total of 20 shape-memory-alloy (SMA) wires, providing a rich variety of possible motions. This test-bed is the first instance of real-time position tracking for an untethered soft crawling robot. Experiments demonstrate that a motion planner can command the robot to locomote to a goal state, given a simple set of motion primitives. This work demonstrates progress toward full autonomy of soft, mobile robotic systems."
"To our knowledge," lead author Zach Patterson writes, "it is the first untethered underwater soft crawling robot, as well as the first untethered soft robot that uses feedback control to accomplish high level goal following." In a video demonstration, PATRICK can be seen approaching a target using this feedback control: The robot itself decides on the best direction to take, and figures out which of its actuators are required.
Patterson and colleagues have positioned PATRICK, which is driven by a Nordic Semiconductor nRF52832 and a Laird BL625 module, as a test-bed design for a range of additional research projects, including the adaption and development of control and path-planning techniques for soft robots, machine learning applications in robotics, and for study of the physics and complexities of walking under water.
The paper is available as a pre-print, ahead of its presentation at the International Conference on Intelligent Robots and Systems (IROS 2020) this October, on arXiv.org.