A Single-Pixel Event Photoactive Device for Real-Time, In-Sensor Spatiotemporal Optical Information Processing

Area of Science:

• Photonics and Optoelectronics
• Sensor Technology
• Materials Science

Background:

• Growing demand for energy-efficient optical sensing necessitates advancements in spatiotemporal information processing.
• Traditional multi-pixel arrays and event cameras struggle with real-time, high-dimensional spatiotemporal data, causing latency and high energy consumption.
• Need for integrated, in-sensor processing to overcome limitations of external data handling.

Purpose of the Study:

• To introduce a carrier-selective, single-pixel, position-sensitive photoactive device with integrated spatiotemporal event sensing and short-term memory.
• To demonstrate in-sensor parallel optical information processing capabilities.
• To enable ultrafast pattern recognition and optical flow detection at the single-pixel level.

Main Methods:

• Development of a novel single-pixel, position-sensitive planar photoactive device.
• Integration of carrier-selective properties for enhanced functionality.
• Implementation of short-term memory for in-sensor data processing.
• Testing with varying light illumination (continuous and pulsed) to assess pattern recognition and trajectory detection.

Main Results:

• The single-pixel device achieved ultrafast (≈0.4 µs) recognition of input patterns with low energy consumption (25 fJ).
• Demonstrated efficient, simultaneous processing of multibit (>4 bit) data.
• Enabled in-sensor optical flow detection and trajectory tracking by adjusting operating speed.
• Proof-of-concept device successfully integrated spatiotemporal sensing and memory.

Conclusions:

• This single-pixel event photodetector represents a significant advancement for compact, energy-efficient, and ultrafast sensing.
• The device facilitates in-sensor parallel optical information processing, overcoming limitations of traditional systems.
• Paves the way for next-generation photonic applications requiring real-time, high-dimensional data handling.