European researchers developed energy-efficient machine vision inspired by human eyesight and the brain
ESPOO, Finland, Dec. 11, 2025 /PRNewswire/ -- Drawing inspiration from human eyesight, a European research project led by VTT has developed machine vision mimicking the cooperation of the eye and nervous system, implemented as edge-computing circuits. Edge computing means processing data where it is generated and where the results of computing are needed. This enables, for instance, intelligent robots and drones that can operate independently in a rescue mission after an earthquake without constant network connections or bulky batteries.
Launched in 2021, the MISEL project (Multispectral Intelligent Vision System with Embedded Low-Power Neural Computing) is now nearing completion. Coordinated by VTT Technical Research Centre of Finland, the project has combined neuromorphic computing-which mimics the way the brain processes information-with semiconductor technologies.
"Our goal is to build truly smart devices that can make observations and decisions on their own, without sending data to supercomputers or the cloud. Neuromorphic computing can be hundreds or even thousands of times more energy-efficient than conventional digital processing," explains Jacek Flak, Research Team Leader at VTT, who coordinates the project.
The MISEL project received nearly EUR 5 million in funding from the EU's Horizon 2020 programme. Alongside VTT, partners include the universities of Łódź, Lund, Santiago de Compostela, and Wuppertal, the Fraunhofer Institute, France's LNE national research institute, and the companies Kovilta Oy (Finland) and AMO GmbH (Germany). The consortium unites expertise across disciplines, including materials science, electronics, and algorithm design.
Smarter mobile devices without constant connectivity
Artificial intelligence typically requires either good network connectivity to cloud services or substantial local computing power. However, the MISEL project brings intelligence to the edge-directly into the devices themselves. This approach allows battery-powered systems to process and interpret sensory data quickly and with low power consumption, while also reducing cybersecurity privacy risks.
"Imagine a drone searching for survivors after an earthquake through smoke, dust, and debris. It needs to interpret its surroundings and make decisions instantly. There may be no network connectivity, and battery life is limited," says Flak.
Human eyesight and the brain as the blueprint for machine vision
Nature has been a source of inspiration for the project. The model for machine vision has been the collaboration between the human retina, visual cortex, and prefrontal lobe for perception and interpretation. The fruit fly also serves as an ideal model as it consumes minuscule amounts of energy to fly, navigate, avoid danger, and find food.
A key result of the project is a system on chip developed by Kovilta, a company specialised in advanced integrated circuit technology. The new circuit includes both imaging and a significant portion of image processing on the same silicon chip. It combines high dynamic range image sensing (over 120 dB), high frame rate (over 1000 frames per second), and massively parallel image processing, enabling versatile motion analysis and pattern recognition.
"Unlike a conventional video camera that actually captures static frames, this sensor detects motion and changes in time and space-just like a biological eye," Flak explains. "The result is a highly efficient, compressed data stream without sacrificing accuracy."
The project also explored so-called quantum dot image sensors, a cutting-edge camera technology that extends vision beyond the spectrum visible to humans into the infrared range. This allows devices to detect patterns and movement even in low-light or foggy conditions where human vision struggles.
Designing the entire signal chain as one system
Within MISEL, sensors, memory units, algorithms, and electronic components have been co-designed as a unified system. Optimizing the entire signal chain is key to achieving a compact and efficient system. The project has also developed specialized processor cores, or edge-AI accelerators, that boost AI performance while keeping power consumption minimal.
"Designing hardware and software in close integration ensures that all components work seamlessly together while maintaining exceptional energy efficiency," says Flak.
Together with Lund University, VTT developed non-volatile memories based on ferroelectric materials, which can be integrated directly onto a chip. These memories have proven functional, and development continues. Kovilta plans to apply the accelerator architectures developed in MISEL to fields such as autonomous robotics and vehicle technology.
"A superior ability to observe the surroundings and accurately interpret observations is a must for robots and vehicles to operate independently and safely among people. For instance, future robots assisting humans need to make rapid decisions to avoid colliding with people moving around them. Simultaneously, the equipment must be small, consume little energy, and be built with moderate costs suitable for a mass product," says Mika Laiho, Chief Technology Officer at Kovilta.
Applications from industry to security
The solutions developed in the project can be utilised in a wide range of applications. Smart cameras could autonomously monitor industrial processes, warehouses, or border areas, while mobile robots could make decisions safely even in the presence of humans. The next step is to leverage the project's results in new projects and VTT's pilot production lines.
"In the future, these results can be used in autonomous devices that see, think, and act as independently and energy-efficiently as a fruit fly," Flak concludes.
Further information
VTT Technical Research Centre of Finland
Jacek Flak, Research Team Leader, MISEL Project Coordinator
jacek.flak@vtt.fi, tel. +358 405366934
CONTACT:
Further information on VTT:
Paula Bergqvist, Communications Manager
+358 20 722 5161, paula.bergqvist@vtt.fi
www.vttresearch.com
This information was brought to you by Cision http://news.cision.com
View original content:https://www.prnewswire.com/news-releases/european-researchers-developed-energy-efficient-machine-vision-inspired-by-human-eyesight-and-the-brain-302638806.html
SOURCE VTT Info