Stanford Engineers Look to Mantis Shrimp Eyes as Muse for Optical Sensor

March 05, 2021 by Jake Hertz

Polarimetric imaging could be a boon to machine vision. Now, Stanford researchers have developed a new type of light sensor inspired by an unlikely source.

Engineers have long looked to nature for inspiration when designing new technologies. Be it the Wright brothers studying birds to develop flight or modern engineers taking inspiration from octopus suckers to design wearable fabric sensors, evolution has provided insights into electronic, optical, and mechanical design. 

Despite this long history of biomimicry, it seems there is still much researchers can learn from nature. Following this trend, engineers at Stanford recently published a paper in which they describe a new form of optical sensor—this time gaining inspiration from an unlikely source: the mantis shrimp. 


Mantis shrimp are able to see visible, ultraviolet, and polarized light. 

Mantis shrimp are able to see visible, ultraviolet, and polarized light. 


Polarization Imagery and Machine Vision 

Light is generally described by three key characteristics: intensity, wavelength, and polarization. Applications like machine vision detect all three, giving software algorithms as much and as detailed information as possible.

Polarization in particular can provide valuable information that conventional imaging can’t, allowing for more sensitive imaging as opposed to non-polarization-based methods. 


Polarization of light through crossed polarizers

Polarization of light through crossed polarizers. Image used courtesy of FSU

While silicon-based techniques have been used to successfully detect both intensity and wavelength, it turns out that silicon cannot detect the polarization of light. Instead, designers must use polarization filters in front of their image sensors. Three conventional techniques for polarization imaging include division of time, division of amplitude, and division of focal plane imaging. 


Conventional Polarimetry Techniques 

The division of time technique works by rotating a polarization filter on top of an image sensor, which captures data sequentially in time. Division of amplitude consists of a prism that will split lights into different paths, each of which has its own sensor. Finally, division of the focal plane works with a micro-polarizer placed on the focal plane to establish different states of polarization. 


Comparison of polarization filter technologies

Comparison of polarization filter technologies. Image used courtesy of Photonics Media

All three techniques, however, are expensive and time-intensive, which may hinder machine vision applications that require fast, low-cost, and highly-robust solutions.


Inspiration from a Crustacean

Understanding the need for better polarimetric imaging in machine vision applications, Stanford researchers turned to nature for a solution. As it turns out, the mantis shrimp can see both hyperspectral and polarized light, a feat much less easily achieved with modern technology. 

In their paper published in Science Magazine, the researchers describe a new sensor that mimics the mantis shrimp’s eye to achieve the simultaneous registration of four spectral channels and three polarization channels. According to the researchers, the design itself consists of stacked “polarization-sensitive organic photovoltaics (P-OPVs) and polymer retarders.” 


The structure of the sensor

The structure of the sensor, the folded retarder elements, and the spectral retardance. Image used courtesy of Altaqui et al. 

The P-OPVs’ response to the light varies based on the light’s polarization, while the retarder disperses the light based on its polarization state in a deterministic way. Utilizing this kind of architecture, the sensor could detect both types of light in a device that is small enough to fit into smartphones. 


Improving Machine Vision 

While just a proof of concept, the sensor could have significant implications. The ability to detect the polarization of light in a form factor small enough to fit in smartphones could fundamentally change the limitations of current machine vision applications.



Other Bio-inspired Innovations

Engineers continually look to nature to solve hardware-level obstacles. Here are just a few recent findings.