Robohub.org
 

A new artificial compound eye


by
23 July 2015



share this:
Fruit Fly - drosophila melanogaster. Source: Francis Prior/flickr

Fruit Fly – drosophila melanogaster. Source: Francis Prior/flickr

Taking cues from the natural world is growing ever more popular in robotics. This week, in a paper published by the Royal Society, a team from NCCR Robotics and LIS, EPFL present a revolutionary new artificial compound eye, created with the fruit fly in mind.

Evolution has cleverly equipped the arthropods’ compound eye with key features that enable these insects to react instantly to danger, even in cluttered environments.Their eyes are able to create low-resolution images with wide fields of view, high sensitivity to motion and a near infinite depth of field, all at low energy cost. The insect’s eye is made up of many thousands of ommatidia and each is an individual sight receptor. Every ommatidia is positioned slightly differently in relation to its neighbours, allowing the brain to collect and compare information from all and piece together a complete image of the surrounding area. These quick functioning, simple eyes are what allow insects to react so quickly when you try to swat them.

Individual eyes as embedded onto a PCB.

Each small eye can be used individually or combined with others. Source: Pericet-Camara et al., 2015.

With all these advantages in mind, the team, led by Prof. Dario Floreano, has created an artificial version of an ommatidium that functions through three hexagonal photodetectors arranged in a triangular shape, underneath a single lens. These photodetectors work together and combine perceived changes in structured light (optic flow), to present a 3D image that shows what is moving in the scene, and in which direction the movement is happening. This means that even when the eye is moving and the environment remains stationary, relative movement is picked up. With three sensors each receiving and combining optic flow in different directions, the eye is able to function as a stand alone sensor so that the multiple, perfectly aligned single eyes used in previous solutions are no longer required.

The eye has been tested in a wide range of lighting conditions and has been shown to function in environments ranging from a poorly lit room to bright sunshine. It even works three times faster than the insect eyes that it is modelled on, with 300 frames per second recorded during testing. This allows an image of a given environment to build up as the robot is moving, enabling it to spot obstacles and react quickly. The ability to rapidly extract data offline and provide an image suitable for navigation, without the need for high energy expenditure, is what allows this eye to be truly versatile.

Flexible array with eyes embedded

Eyes can be arranged on a flexible array Source: Pericet-Camara et al., 2015).

This new sensor is 1925 x 475 x 860 μm in size and weighs only 2mg, making it ideal for a micro air vehicle (MAV), or other flying robots where both speed and payload are of high importance. A further advantage of this method is that multiple sensors may be used together, each pointing in different directions, to provide a miniature omnidirectional camera for motion detection. Since the sensors work independently, they can be mounted in any configuration, even on flexible substrates. The team has demonstrated the operation of their compositional eye in the shape of a bendable strip (which Prof. Floreano calls Vision Tape) because it can be attached to any planar or curved object.

Future work could see the sensor embedded onto the surface of soft and flexible robots or in smart clothing to allow blind people to “see” obstacles; it could be used in medical endoscopes, enabling more precise operations. The fact that it can also be arranged in layers, allowing for adaptation to arbitrary shapes, makes the possibilities for its use endless.

Reference

R. Pericet-Camara, M.K. Dobrzynski, R. Juston, S. Viollet, R. Leitel, H.A. Mallot, D. Floreano, “An artificial elementary eye with optic flow detection and compositional properties,” J. R. Soc. Interface, vol. 12: 20150414, 2015. http://dx.doi.org/10.1098/rsif.2015.0414

If you liked this article, you may also be interested in:

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.



tags: , ,


NCCR Robotics





Related posts :



Robots to the rescue: miniature robots offer new hope for search and rescue operations

  09 Sep 2025
Small two-wheeled robots, equipped with high-tech sensors, will help to find survivors faster in the aftermath of disasters.

#IJCAI2025 distinguished paper: Combining MORL with restraining bolts to learn normative behaviour

and   04 Sep 2025
The authors introduce a framework for guiding reinforcement learning agents to comply with social, legal, and ethical norms.

Researchers are teaching robots to walk on Mars from the sand of New Mexico

  02 Sep 2025
Researchers are closer to equipping a dog-like robot to conduct science on the surface of Mars

Engineering fantasy into reality

  26 Aug 2025
PhD student Erik Ballesteros is building “Doc Ock” arms for future astronauts.

RoboCup@Work League: Interview with Christoph Steup

and   22 Aug 2025
Find out more about the RoboCup League focussed on industrial production systems.

Interview with Haimin Hu: Game-theoretic integration of safety, interaction and learning for human-centered autonomy

and   21 Aug 2025
Hear from Haimin in the latest in our series featuring the 2025 AAAI / ACM SIGAI Doctoral Consortium participants.

AIhub coffee corner: Agentic AI

  15 Aug 2025
The AIhub coffee corner captures the musings of AI experts over a short conversation.

Interview with Kate Candon: Leveraging explicit and implicit feedback in human-robot interactions

and   25 Jul 2025
Hear from PhD student Kate about her work on human-robot interactions.



 

Robohub is supported by:




Would you like to learn how to tell impactful stories about your robot or AI system?


scicomm
training the next generation of science communicators in robotics & AI


 












©2025.05 - Association for the Understanding of Artificial Intelligence