Do robots require 3D vision? What are the most essential features, for robot vision? Both robotic experts and users struggle to find answers to these questions. A section of people believes that 3D is essential seeing that human beings have stereo vision. How good is 3D vision compared to 2D?
Research conducted not long ago sought to understand a praying mantis’ vision. Researchers have since used their findings to figure out the necessary components for robot vision.
A Praying Mantis’ 3D Vision
According to Newcastle University researchers, the praying mantis in 3D. However, their 3D vision differs from that of humans. Usually, stereo vision is acquired by capturing two images from the same location albeit from varying spots.
The robotic vision computer or the human brain then compares the two images. The existing differences are then utilized to obtain depth information by determining the existing distance between the images.
Praying mantises come with two eyes just like human beings. The insects can accurately recognize depths when they need to capture their prey in a manner similar to that of other predators and human beings. You could be convinced that these insect’s visual system is no different than that of other predators which is not true.
Praying mantises utilize an accurate and efficient 3D vision system. Still, this system is only effective in distinct situations; when both of their eyes are focused on the same target and when their target is moving. Researchers believe that in the absence of pf movement, these insects lack the ability to see in 3D and will probably utilize 2D then.
When the praying mantises’ eyes are focused towards a different moving image, chances are that the insect will recognize 3D moving components incorrectly. This can be likened to the optical illusion in human beings where the visual system displays things which are not there in the actual sense.
Researchers established that a praying mantis could recognize hidden objects with ease and were not deterred by a fluctuation of light. According to the researchers, the insect’s performance in such situations was better compared to humans.
Universal Robots technicians have been on the front line in a bid to explain how to integrate the new findings with robotics. Since the praying mantis has an efficient vision, such stereo vision can be ideal for moderate computation robot assembly.
Lessons to Master by Observing the Eyes of the Praying Mantis
There are various lessons that the praying mantis teaches us which include:
·Pair the Vision System Type to the Exact Needs of the Task
The praying mantis comes with a unique vision system which functions appropriately when paired with the right application; that is seizing moving prey whenever it gets within its vicinity. Still, the insect’s three dimension vision can perform dismally in various other situations.
For instance, it may not recognize the location of a stationary pen neither can it pick it. In this situation, humans perform way better but are unable to recognize hidden moving things.
When it comes to robotics, investing the expensive and complex vision systems in the hope that it will be more effective in numerous situations is not advisable. It’s important to understand that no vision system can work efficiently for all situations.
Just as the praying mantis vision system functions more efficiently compared to human vision in certain situations albeit dismally in others, the same happens in regard to robot vision.
Comparing Three Dimension Vision with Two Dimension Vision
Some experts opine that three dimension vision is more efficient compared to two dimension vision in robotics. However, a section of robotics researchers and manufacturers say that 3D vision is way expensive which explains why many users shun it. The aforementioned statements are both incorrect. This is because 3D is not ideal for all situations. 2D vision could also be more efficient than 3D in many other situations.
Both 2D and 3D come with a variety of challenges. Robot assembly vision is more functional if it can detect the targeted object, which is why specific features such as an easy to program matching template is essential. It enables users to program their preferred components with ease.