Researchers Develop 2-Terminal Retinomorphic Sensor for Reconfigurable Optoelectronic Computing
Updated
Updated · Nature.com · Jun 5
Researchers Develop 2-Terminal Retinomorphic Sensor for Reconfigurable Optoelectronic Computing
3 articles · Updated · Nature.com · Jun 5
Summary
A new multi-responsive retinomorphic sensor uses a two-terminal Sb2Te3/MoS2 heterostructure to switch in situ and reversibly between photodiode and opto-synaptic modes under low-bias voltage control.
Focused light pulses also let the device mimic leaky integrate-and-fire neuronal behavior, giving one sensor multiple optoelectronic responses that existing arrays have struggled to combine at low hardware cost.
By pairing a sensor array with diffractive optical components, the researchers demonstrated multi-mode computing for image processing, video processing and transfer learning through optical spike encoding.
The architecture is designed to boost spatiotemporal dimensionality in front-end visual computing, with potential uses in autonomous driving, satellite remote sensing and robotics.
In June 2026, a major breakthrough in bio-inspired computing was achieved with the introduction of the Multi-Responsive Retinomorphic Sensor (MRRS). Built on a two-terminal Sb₂Te₃/MoS₂ heterostructure, the MRRS stands out for its reconfigurable operational modes, allowing it to switch between photodiode, opto-synaptic, and opto-neuronal functions. This design enables advanced in-sensor optoelectronic computing, where data is processed directly at the point of acquisition, closely mimicking the efficiency of biological visual systems. As a result, the MRRS offers remarkable versatility, high speed, and energy efficiency for vision-based tasks.