Low-power consumption, heavy-metal-free wide-spectrum image sensors for mass-market computer vision applications

Descripción del proyecto

Tecnología de sensores de imagen para aplicaciones de visión artificial

La industria de la imagen está experimentando cambios significativos impulsados por tendencias como los coches autónomos, el metaverso y la robótica de servicios, todo ello potenciado por la visión artificial. Sin embargo, la tecnología de la imagen actual está más adaptada a los humanos que a las máquinas. Qurv es una empresa dedicada a reinventar los sensores de imagen para garantizar su rendimiento fiable y robusto en condiciones de baja visibilidad. Su objetivo es abordar problemas como el consumo de energía y el uso de materiales tóxicos, aprovechando su experiencia de primer nivel. El equipo del proyecto QSTACK, financiado con fondos europeos, tiene como objetivo desarrollar y probar una tecnología de sensores de imagen rentable, diseñada específicamente para aplicaciones de visión artificial sensibles al consumo energético. En este proyecto se pretende situar a la industria europea de semiconductores en una posición estratégica para capitalizar las aplicaciones de visión artificial, con posibles repercusiones en la electrónica de consumo, la sanidad y la movilidad.

Objetivo

The imaging industry is experiencing a paradigm shift due to megatrends such as autonomous cars, the metaverse and service robotics, all applications powered by computer vision (CV). This subset of artificial intelligence (Al) is expected to contribute more than €2 trillion to the global economy by 2030.

Present imaging technologies were designed to fulfil the needs of humans for high-quality photography but not for machines. Autonomous and intelligent systems will require re-inventing image sensors to deliver reliability and resilience under low-visibility conditions while increasing the amount of information they capture and analyse. Qurv, a company operating in this space that brings together unique world expertise, has identified power consumption and toxic materials as the common drawbacks of current solutions, hindering mass-market uptake.

To address these challenges, the project QSTACK will develop and validate a wide-spectrum image sensor technology and business case that delivers unprecedented cost performance to mass-market, power-sensitive computer vision applications. This ambitious overarching aim will be achieved through the following objectives:

i. Achieve the integration of high-performance wide-spectrum CMOS / TMDC / heavy-metal-free stack SWIR sensitive image sensors as demonstrated by imaging and electrical characterisation analysis.
ii. Achieve a reduction of the total image sensor’s power consumption, enabled by a >1,000x decrease in the pixel engine power consumption.
iii. Increase the investment readiness of QSTACK, as demonstrated by the number of early users interested in pilots and evaluation kits, together with the creation of investment material.

QSTACK envisions putting the European semiconductor industry in an advantageous position to seize the opportunity in mass-market computer vision applications that will have an enormous impact on consumer electronics, personalised health, and safe mobility.