New intrinsically stretchable electronic skin (“e-skin”) based perception system capable of providing accurate, high-resolution, real-time and full-geometry characterization of a soft object’s shape, position and environmental stimuli. This technology allows determination of a soft object’s movement, action, location and touch area, enabling the precise control, operability, and practical implementation of soft robots and other soft systems.
This technology allows the determination of a soft object’s movement, action, and location, enabling the precise control, operability, and practical implementation of soft robots and other soft systems.
Soft devices, and in particular soft robots, offer ground-breaking approaches to address technological challenges for which conventional rigid systems are simply unfit. These include surgical robotics, compliant prostheses, industrial processes and wearable technologies. Soft systems offer intrinsically safer and more life-like interactions, however their conformable structure also makes their characterization difficult and limits the determination of an object’s movement, action, location and the sense of environment (so called 3-dimensional perception) which is essential for controllability and operability required for the practical implementation of such innovations. To overcome this, University of Edinburgh researchers have developed a new 3D-perception system that enables real-time, high-accuracy three-dimensional, full-geometry shape reconstruction and tactile sensing suitable for use with soft robots and other soft systems.
The Edinburgh technology is underpinned by a new capacitive ‘e-skin’ design in combination with an innovative deep learning methodology. Our solution provides real-time (30 fps), high accuracy (mm-scale error), full-geometry shape reconstruction of dense point clouds of 3D-geometries under complex multimodal deformations (i.e. bending, twisting, elongation and their combinations). The e-skin can also provide touch recognition from the e-skin signals. This represents a major advancement over existing approaches for soft robot perception that are limited to sparse geometrical inference under one or two prescribed types of deformations or only tactile sensing.
The Edinburgh e-skin technology is intrinsically stretchable, scalable and can be deployed either as an integral component of a product or as an external layer applied to the surface of the object. With a rapid training pipeline which is agnostic of the soft object’s shape, size and geometry, the technology can be readily deployed with any soft device where 3D-perception is desirable.
Yang et al, Smart capacitive e-skin takes soft robots beyond proprioception, pre-print available