Spatial Position–Force Perception for a Soft Parallel Joint via Pressure-Deformation Self-Sensing

摘要

Robotic perception, the ability to detect internal position and orientation and external stimuli, enables robots to interact safely with humans and manipulate safely in unstructured environments. Soft robots hold an inherited leading edge over rigid-body robots in terms of interaction compliance and safety, but they still require dedicated sensors when accuracy or force interaction is concerned. However, perceiving high-dimensional multimodal information for the soft robot is still a challenge. Previous works focused more on single-type perception for one-degree-of-freedom (DoF) soft robots with complicated sensor fabrications. In this work, inspired by the human muscle perceptive system, we proposed a similar synthetic sensing module with embedded pressure and laser range sensors. Then, a three-DoF [one-DoF axial motion ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\pm 25$</tex-math></inline-formula> mm) and two-DoF omnidirectional bending motion ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\le30^{\circ }$</tex-math></inline-formula> )] soft parallel robot based on the self-sensing origami actuator is developed. The actuator- and joint-level perceptive models are derived with embedded sensors to realize 3-D position and force perception simultaneously. It is validated by the dedicated position and orientation, and axial force and bending torque perception experiments (phase lag <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\approx$</tex-math></inline-formula> 450 ms) in an origami soft parallel joint (OSPJ) with four actuators. The method may pave the way for an inexpensive internal sensing method, not affecting the soft actuator's performance, and provides possibilities for the soft robots to achieve multimodal estimation for closed-loop control and interaction with the physical environment.