The tiny visual systems of flying insects have inspired researchers to develop optoelectronic graded neurons for perceiving dynamic motion, enriching the functions of vision sensors for an agile response.
With excellent energy efficiency, biological visual systems can accurately detect motion in a complex environment. In particular, flying insects can see objects moving quickly and have high flicker function frequencies (FFF). This natural inspiration helps advance machine vision technologies while using very little technology. Traditionally, “spatial” and “temporal” stream processing architectures are used in complicated artificial neural networks in machine vision systems for action recognition.
The study team demonstrated that optoelectronic graded neurons might combine spatial and temporal information at sensory terminals and conduct high information transmission rates (>1000 bit/s). The research’s results are significant since they enable impossible features with traditional image sensors.
This study significantly broadens our comprehension of bioinspired computing. The study’s findings have potential uses for autonomous vehicles, which must be able to detect fast movements in road traffic. The technology might also be applied to various surveillance systems.
Hardware with physically different image sensors and processing components makes up most machine vision systems. But most sensors can only produce “spatial” frames since they don’t combine “temporal” data. Acute motion recognition necessitates transmitting and fusing “spatial” and “temporal” stream information in the processing units. As a result, this bio-inspired in-sensor motion perception contributes to the advancement of motion processing, a computational challenge that places heavy demands on computational capacity.
Related Content: Moiré Superlattices For Optoelectronics