The research makes significant advancements in AI and visual computing and also introduces practical applications, such as enhanced smartphone camera capabilities.Â
Researchers are advancing optical computing, which utilizes photons instead of traditional electronic components, to develop faster and more energy-efficient computing systems. These systems could potentially improve the processing of visual information through simultaneous, parallel processing. Optical computing has historically struggled with achieving nonlinear responses, essential for applications such as artificial intelligence.
Current research focuses on developing nonlinear materials and devices that require a significant amount of light to function effectively. Challenges have included the need for high-powered lasers, slow processing due to light absorption, and the use of materials that are not light-efficient. A study by the California NanoSystems Institute (CNSI) at UCLA introduces a device that addresses these issues. The device features a small array of transparent pixels capable of producing a fast, broadband, nonlinear response using low-power ambient light. The research highlights a practical application of this technology—integrating the device with a smartphone camera to reduce image glare.
The team emphasized the importance of developing efficient, fast, and low-loss nonlinearities to meet visual computing demands. The potential applications of this technology are broad, ranging from enhanced sensing in autonomous vehicles to advanced image encryption techniques. One significant advantage of the new device is its ability to process images directly without converting them to digital signals, thereby speeding up the processing time and reducing the data sent to the cloud. This could enable the production of higher-resolution images and more precise information capture about object arrangements and the surrounding light spectra.
The device itself is compact, consisting of a 1 cm square transparent plane made from a thin film of 2D semiconductor material coupled with liquid crystal and an array of electrodes. This setup allows each of the 10,000 pixels to selectively darken when exposed to ambient light, achieving a dramatic change in transparency with minimal photon input. This opens new pathways for high-resolution imaging and sensing technologies, making them more accessible and efficient. The researchers are excited about the future applications and the potential of this technology to go beyond fundamental research to real-world applications. The collaboration highlights a significant step forward in optical computing, promising more advancements in the field.References: Dehui Zhang et al, Broadband nonlinear modulation of incoherent light using a transparent optoelectronic neuron array, Nature Communications (2024). DOI: 10.1038/s41467-024-46387-5
