MRS Meetings and Events

 

EL04.02.09 2023 MRS Fall Meeting

A Perovskite Nanowire Array Based Spherical Bionic Eye with Filter-Free Color Vsion, In-Device Neuromorphic Signal Preprocessing and Adaptive Optics

When and Where

Dec 6, 2023
10:35am - 10:50am

EL04-virtual

Presenter

Co-Author(s)

Zhenghao Long1,Zhiyong Fan1

The Hong Kong University of Science and Technology1

Abstract

Zhenghao Long1,Zhiyong Fan1

The Hong Kong University of Science and Technology1
Bio-inspired vision systems that utilize curved detector geometry have the potential to create miniaturized cameras with a wide field of view and low aberration. Nonetheless, integrating color vision, in-device signal preprocessing, and adaptive optics into spherical artificial eyes has always been a challenge. Herein, we developed a bionic eye that integrates all of the above functions. The device contains a hemispherical artificial retina and an optical sub-system.<br/>The artificial retina is based on a hemispherical perovskite nanowire array, which possesses in-device color distinguishing and neuromorphic preprocessing abilities. The device structure incorporates a SnO<sub>2</sub>/NiO double-shell nanotube filled with ionic liquid in the core, which is located on top of a CsPbI<sub>3</sub>/NiO core-shell nanowire. It has been observed that under shorter (blue) or longer (green and red) wavelength illuminations, the positive surrounding gate effect of NiO due to photo hole injection can be partially or fully balanced by electrolyte, respectively. Thus, the device can yield either positive and negative photocurrent under shorter or longer wavelength illumination, respectively. Additionally, the color selectivity of such a structure can be adjusted by applying a small external bias, which allows for the acquisition of more detailed color information in multiple measurements. We demonstrated the reconstruction and classification of color patterns based on the artificial retina. Meanwhile, photo-generated carriers can accumulate in the SnO<sub>2</sub>/NiO structure, resulting in a self-powered synaptic photo response that we used to demonstrate the neuromorphic preprocessing of noise filtering. The in-device signal preprocessing, along with the self-powered feature, can significantly reduce energy consumption.<br/>The optical subsystem includes a fixed lens, an artificial crystalline lens, and an artificial iris, all of which mimic the optics of a human eyeball. The artificial crystalline lens is a liquid crystal Pancharatnam-Berry lens. Compared to conventional glass lenses, it exhibits reduced geometrical aberration and a much smaller thickness. The artificial crystalline lens can be switched between light refraction mode and transparent mode, allowing for the tuning of the focal length of the optical subsystem. Similarly, the artificial iris is based on liquid crystal technology, and it has five rings of liquid crystal. The transparency of each ring can be individually controlled from 0% to 95%, which enables the overall aperture size to be adjusted from 3.14 mm<sup>2</sup> to 78.50 mm<sup>2</sup>. The electronic optics, which has reduced thickness and geometrical aberration, can assist with miniaturization and expand the depth of field and dynamic range of the system.<br/>Overall, we have designed and developed a distinctive hemispherical bionic retina and spherical eye device that possesses the missing capabilities of color vision, optical adaptivity, and energy efficiency in previous research. This device presents a new paradigm by integrating structural and functional traits that are similar to biological eyes, and it highlights the potential for further exploration and optimization of artificial visual systems with biomimetic functionalities that can enhance the efficiency of machine vision and robotics.

Symposium Organizers

Simone Fabiano, Linkoping University
Paschalis Gkoupidenis, Max Planck Institute
Zeinab Jahed, University of California, San Diego
Francesca Santoro, Forschungszentrum Jülich/RWTH Aachen University

Symposium Support

Bronze
Kepler Computing

Publishing Alliance

MRS publishes with Springer Nature