Scientists are developing a 'bionic eye’ that could be used by manufacturers to improve monitoring of industrial assembly lines.
At the back of the human eye is a specialised layer of cells called the retina, which captures light information. This information is then converted into electrical signals and sent via the optic nerve to the brain, where a visual image is produced.
A team of engineers from Imperial College London are replicating this part of the eye, creating an artificial retina that captures light to build an image of its surrounding environment. The researchers behind the EU-funded ‘SeeBetter’ project suggest replicating the retina on a single, specialised silicon chip will enable more accurate artificial visual sensing in industrial processes.
The team, working in conjunction with the Institute of Neuroinformatics in Zurich, and international research centre IMEC, aim to combine the artificial retina with a simple software infrastructure, enabling information to be processed in a similar way to the brain. When applied in industry, this could allow a robot to process and react to information, which could for example, enable manufacturers to recognise faults much quicker in the products they are making.
Ultimately, the team suggest it might be possible to adapt the artificial retina technology into a prosthetic to enable the blind to see. However, this could take another five to ten years of development.
Dr Konstantin Nikolic , one of the lead researchers from Imperial’s Department of Electrical and Electronic Engineering, said: “The ultimate goal for this technology is in healthcare to restore sight to people who are blind, but this is still a long way off. In the short term we see our technology being extremely useful in improving machine vision in manufacturing. In order for a conventional camera to capture and identify a faulty product on a manufacturing production line the conveyor belt must be moving fairly slowly. When you use a faster image capturing and processing system, such as our artificial retina, it could recognise a faulty product and react faster, saving money in the manufacturing process.”
The sensor then works in a similar way to the neurons they are imitating, sending either an ‘on’ or ‘off’ signal to the processing chip, or ‘brain’. It is currently capable of identifying and tracking objects, as well as determining their speed. It can measure minute changes in light intensity, right down to the individual pixel.
The final ‘picture’ is composed of only those moving elements essential for computer processing, making the process quicker and more efficient. The information is then used to produce a video stream that can be transmitted to a screen for display.
Dr Nikolic continues: “Imagine a surveillance camera monitoring a static scene. Just think how many redundant pixels are sent from frame to frame while nothing changes. This concept of collecting only changing data enables a greater range of light to be captured in a shorter exposure time, allowing the sensors to capture fast movements.”
The technology combines an off-the-shelf vision sensor with programming and software developed by the team. This could make it cost effective to manufacture and ultimately more affordable for industry. Dr Nikolic suggests that if industry were to identify specific potential uses, and the relevant testing and assessment was carried out, the technology could be in use within a year’s time.
The next step will see Dr Nikolic and his team adapting the technology for other uses. For example, they plan to develop a visual to auditory ‘sensory substitution system’, in the form of a simple, smart phone app, which could convert visual information into sound to enable the user to see with their ears.
In the long term, the team predict that the technology could be adapted into a prosthetic to restore vision. The technology could interface with the human neural system, conveying visual information directly into the wearer’s brain.