Since Apple and Google AR tracking technologies are already in the hands of millions of developers and users, the market has generated a lot of attention and coverage, and you may think that the immersive augmented reality experience we have been imagining is coming soon. Although we are closer to this future than ever before, in reality, immersive augmented reality still needs years of R&D and design work before it becomes mainstream. Here we will take a look at some of the key challenges that current augmented reality technology is facing.
Immersive field of view
After watching the cool ARKit demo video, it's easy to imagine how magical it would be to have a full-screen view of the entire field of view. However, the reality is that even the most advanced portable AR head development kit is still very limited in its field of view (comparable to today's VR head display, and some people think that the current VR head is not enough field of view is not enough ).
In many ways, HoloLens is the best AR headpiece that current developers can buy, but its field of view is only about 34 degrees, which is far less than Google Cardboard (about 60 degrees). The video in the article compares the full field of view with a field of view of about 34 degrees and the results show that you can only see a small portion of the augmented reality at any moment.
This is very important, because to achieve a reasonable sense of immersion, the enhanced world needs to integrate seamlessly with the real world. If you can't immediately see most of the augmented reality, you will find that you need to "scan" the environment unnaturally to find out the actual location of the AR object (just like using a telescope), not to say that your brain is intuitive Map the AR world and see it as part of the real world.
It's not that the AR head of the 34-degree field is useless. It's just not enough to be physically there, so it can't immerse your natural perception deeply. It also means it's not suitable for this kind of intuitive human-computer interaction. It's not Ideal for consumer and entertainment use.
Someone may say, "What about Meta2 AR with a 90-degree field of view?"
Yes, Meta 2 is the current AR head with the largest field of view, which is close to today's VR headline. However, this device is very bulky and there is no obvious solution to miniaturize its optical system without sacrificing most of the field of view.
Meta 2 Optical Lens is actually very simple. The big "hat" part of the head display contains a monitor similar to a smartphone. Some of the large plastic shades are silvered and reflect the contents of the display back into the user's eyes. Reducing the head display means reducing the display and the hood, which obviously reduces the field of view. Meta 2 may be a very good device for developers, and they are willing to endure clunky head-ties for developing future devices, but for consumers, Meta must use different optical solutions to achieve this.
In this regard, ODG is developing a similar but smaller optical system and can achieve a field of view up to 50 degrees, which is $1800 R-9 AR glasses. However, they can only barely come close to what consumers can accept. On the other side, Lumus uses a different optical solution (waveguide) to successfully achieve a 55-degree field of view in a 2mm-thick optical element.
The field of view of about 50 degrees is not bad, but it is far behind the current high-end VR headlight of about 110 degrees, and consumers are still demanding a wider field of view. For a truly immersive field of view, it is difficult for us to judge a specific number, and Oculus used to think that we need to have at least a 90-degree field of vision to experience a true presence (at least in this area, the VR industry Some people agree with it.
2. Real-time object classification
Apple's ARKit technology and Google's ARCore technology allow you to implement some very dazzling and novel AR-like experiences on smartphones, but in most cases, these systems are limited to "understanding" planes such as floors and walls. This is why 99% of AR apps and demos on iOS can only happen on the floor or on the table.
Why are floors and walls? Because they are easy to classify. The plane of the floor or wall is the same as the plane of the other floor and the other wall, so the system has the confidence to assume that this plane can be extended to all sides until it intersects with another plane.
Note that I use the word "understand" here instead of "perceived" or "detected". This is because although the system may be able to "see" the shape of objects other than floors and walls, they are currently not understood.
Let us use the cup as an example. When you look at a mug, you see more than just a shape. You have a good understanding of the mug. To understand how much? Here let us take a look:
You know that the cup is very different from the plane it is on.
You know that the cup contains a certain amount of space for liquids and other objects.
You know we can use cups to drink water.
You know that cups are light and easy to overturn, resulting in spillage of liquids or objects inside the cup.
......
I can continue to say down... I want to say that the computer does not know anything about it. It can only "see" a shape, not a cup. The computer could not get a complete view of the interior of the cup and could not map out the complete shape. The computer could not even assume that there was a certain amount of space inside the cup. The computer also does not know that the cup is an object that is independent of its plane. But you know it all because it is a cup.
However, making computer vision understand "cups" and not just see a shape is a very important issue. So over the years, we saw in the AR demos people attach benchmark marks to objects to achieve more detailed tracking and interaction.
Why is it so difficult? The first challenge is classification. Cups have thousands of shapes, sizes, colors and textures. Some cups have special attributes and special uses (such as beakers), which means different cups are used for different scenes and backgrounds.
You can imagine the challenge of programming such an algorithm that helps the computer understand all of the above concepts; you can also imagine the challenge of writing a code that explains the difference between a cup and a bowl to a computer.
Just a simple cup has such a huge challenge, not to mention the thousands or hundreds of thousands of common items in the world.
Current smartphone-based ARs occur in your environment, but it is difficult for you to interact with them. This is why all AR experiences you see on your smartphone today are fixed to the floor and walls. This kind of system is unlikely to make a convincing interaction with the world around us because although the system can "see" floors and walls, it cannot "understand" them.
For the science fiction AR we aspire to (that is, the AR glasses can show me the temperature of the coffee in the cup; or to show the remaining time of the microwave above), we need the system to "understand" more about the world around us. .
Then how do we achieve it? The answer seems to be so-called "deep learning." Handwriting classification algorithms for each type of object, even ordinary classification algorithms, are all very complex tasks. But we can train a computer's neural network, design this neural network to be able to automatically adjust its programming over time, and reliably detect the surrounding common items. We have reported that researchers are already working on some projects and look very promising. In this video below, the system can detect the difference between any human, umbrella, traffic light, and car with a little more reliability.
The next step is to significantly expand the possible classification libraries and then combine image-based detection with real-time environment mapping data collected from the AR tracking system. Once we can let AR systems begin to "understand" the world around us, we can begin to address the adaptive design challenges of the AR experience.
3. Adaptive AR design
For example, Web developers have spent years developing reliable, practical design rules that make the site adaptable to different shapes of screens. But compared to adaptive AR design, the former seems to be a simple task because the latter needs to support any environment that covers all three dimensions.
This is not a simple question. Even with VR game designs that have been put into practical development for years, developers are still trying to solve a more basic version of the puzzle: designing for different playing space sizes. In general, the shape of the VR play area is square or rectangular and there is nothing but the player. This seems to be a simple matter with an AR experience accompanied by a series of concurrent objects.
Imagine that even people who live in the same apartment unit have different furnishings and items. To understand how to create a compelling entertainment experience, ar game design will take many years to develop. From flat to ceiling, to furniture, to millions of homes, this entertainment experience needs to adapt to a seemingly infinite environmental variable (not to mention more extensive outdoor space).
You may think that it is not difficult to develop a simple AR shooting game. For example, in a one-bedroom design, the enemy will emerge from that particular room. But don't forget that if you do not map the environment in advance, the AR system does not even know that there is another room in the house.
Let's assume that the developer has solved the object classification problem. Such a system can understand the objects around you at the human level. How should the developer create a game that uses these objects?
For example, a simple farming game in which players need to grow and water AR crops at home, this involves using real cups to water AR water. But what if there are no cups around you and if such games will become useless? The answer is no, developers are smart people. For example, players can use closed fists as substitutes for cups, when tilting their fists, AR water will fall out.
So we can now plant crops. U.S. developers believe that everyone should have enough space to grow 10 rows of corn; but in China, the small apartments we live in are not suitable for growing 10 rows of corn, because most people do not have extra bedrooms for sowing.
I can continue to talk about it. What I want to say is: If we are no longer limited to experiencing immersive AR on blank floors and walls, we need to design adaptive ar games and applications that involve using the physical space and objects around us. Therefore, we need to manage billions of variables through some very clever designs.
Although this may be the furthest of the three major challenges, we can now design theoretically before the arrival of future devices that truly enable these experiences.
In the past year, we have heard many people think that AR and VR are comparable in terms of technological maturity, but in reality, AR still lags far behind current VR. AR is very exciting, but from hardware to perception to design, we still have a lot to learn. For AR, it is now an exciting time. This area is still quite open, and the market has emerged a solid foothold, and the time to enter the AR market is ripe.
N-BK7 Precision Windows are ideally suited for industrial and low power laser applications. The high tolerance design yields minimal beam distortion and scatter. Broadband coating options extend the range of these precision windows through the visible and near-infrared spectra.
Our plane plates are also available with the following coatings:
VIS range (ARB2: R < 0.5 % per surface, AOI = 0°, λ = 450 - 700 nm).
NIR range (ARB2: R < 0.5 % per surface, AOI = 0°, λ = 725 - 1050 nm).
Nd:YAG fundamental (ARHS: R < 0.2 % per surface, AOI = 0°, center wavelength 1064 nm)
Reflective aluminum coating (RAL: for λ = 400 nm - 12 μm, damage threshold > 200 mJ / cm2)
Front surface mirror silver coating (RAGV: for λ = 500 nm - 12 μm, damage threshold > 200 mJ / cm2)
For customized optics please contact our sales team for your individual quotation.
Bk7 Wafers,K9 Wafers,Thin Transparent Glass,4 Inch Bk7 Wafers
Zoolied Inc. , https://www.zoolied.com