WaveOptics’ views on AR headsets’ future – An interview by Yole Développement

The dream of Augmented Reality (AR) is continuing, even though we are yet to see people wandering around cities with pairs of glasses on their faces. This does not mean that the industry is not working towards that goal, as we see lots of movement to improve the technology and to structure the supply chain, as explained in Yole Développement’s report “Displays and optics for AR & VR“. The latest of example for this is the Facebook and Plessey deal announced late March 2020.

However, the path to the consumer is still long and strewn with pitfalls.

We have been hearing for so long about how smartphones are going to be supplanted by these AR headsets, turning this event into the next big consumer electronics revolution. Companies have invested lots of money to develop the required technology-enabling building blocks in-house. We think about the display engines and hear a lot about microLEDs. But we cannot separate these engines from the optics that go along with them, as there is a need to build up a system that exhibits good performance, cost-competitiveness, all in an acceptable form factor, and all to provide a compelling use case to be able to convince the consumer. Will this ever happen?

Dr. Zine Bouhamri, Display Technology and Market Analyst for Yole Développement, has had the privilege to address some of these topics in a talk with Dr. Phil Greenhalgh, Chief Technical Officer at WaveOptics. His company is one of the major companies involved in surface-relief gratings based optics for AR headsets. Read on to find out what is happening in the industry.

Yole Développement: Could you please introduce WaveOptics and its activities?

Phil Greenhalgh: WaveOptics is based in the UK, with subsidiary offices in China, Taiwan and Austria. Presently, we have around 105 employees who design and manage the manufacture of a range of optical waveguide and light engine solutions for AR glasses. Our waveguides are designed with both performance and cost in mind. Our view is that only waveguides that use surface relief gratings (SRG) can achieve the very tough price-performance balance for anything approaching consumer volumes. This opinion is clearly shared by the only two public and mainstream AR participants: Microsoft and Magic Leap also chose SRG waveguides for their products.

So, in summary, we design and manufacture optical waveguides and light engines optimised for AR glasses and other types of head mounted displays.

YD: What are the typical products and services you provide?

PG: In the past 12 months, it has become apparent that our customers rarely want a standard off-the shelf waveguide for their product. Even something as basic as the outline of the waveguide defines the whole industrial design and that is not good for product differentiation. So one size most certainly doesn’t fit all.

And the range of waveguide specification requests is considerable: fields of view (FOVs) from 15° to 60°, monochrome, full colour and monocular and binocular configurations have all been requested at some point in 2019.

So, we moved towards a platform approach for 2020 where we have standard configurations to cover the range of FOVs that can be adapted quickly for specific designs. Adaptation is possible in a matter of weeks because our proprietary simulation tools are very accurate now and yield physical waveguides that are close to the required design after a single design iteration.

There is less requirement for custom light engines as they are hidden, costly to tool and our standard designs perform very well and compliment the waveguides.

YD: What is so unique about how you approach this technology?

PG: WaveOptics approach waveguide technology as an envelope of trade-offs. FOV, eyebox size, efficiency, colour uniformity and can all be varied usually at the expense of one of the others. We involve our customers early on with the trade-offs and show how our SRG waveguides have the widest design space of all that can provide them with the best AR product. The previously mentioned trade-offs will always exist in waveguide technology but what we can and are doing is to make the enveloping limits as large as possible. We are doing this by constant research into new materials, manufacturing methods and new design of the gratings themselves.

YD: How do you compare with the SRG competition such as Microsoft, Magic Leap or Dispelix?

PG: All I can do is to repeat what our customers say and that is we compare very well.

YD: How about competition with Holographic Optical Element (HOE) waveguides like Apple’s? And Reflective Optical Element (ROE) waveguides like those from Lumus, LetinAR and Optinvent?

PG: It is public record that Apple bought Akonia and most of the industry speculated it was to gain access to their volume holograms for AR waveguides or maybe films for a reflective near-eye combiner. Like everyone else I don’t know the status of this work, but I have used volume gratings before, and they are very difficult to get right. If they don’t behave then they are very difficult to diagnose why as there is nothing that can be easily seen or measured that would explain why. At least with SRG gratings they can be imaged and measured to compare with the design intent.

Reflective waveguides are subject to their own envelope of limitations. They have generally better efficiency and colour uniformity compared to diffractive waveguides, but some other key parameters fall short. Two significant limitations with reflective guides are manufacturability and that pupil expansion can only be easily and efficiently provided in one dimension.

The process to make a reflective waveguide is effectively serial – there is no clear way to parallelise the assembly unlike nano-imprinted SRG waveguides. Having built many reflective waveguides in the past, I can say it’s very easy to make a bad one but very, very hard to make a good one. The world leader in reflective waveguides is Lumus and their waveguides are very good. So, I can imagine it is very difficult to set up a line that can make say even 10k/month. That is easy for nano-imprinted SRG waveguides.

Two-dimensional pupil expansion is critical to simultaneously achieve a large eyebox and small light engine. Pupil expansion is where the output structure of the waveguide copies the light engine pupil over a rectangular area called the eyebox. Only single dimension pupil expansion has been commercially shown in reflective waveguides so far. Horizontally the eyebox is OK as it benefits from pupil expansion but vertically it is limited by the size of the light engine optics. To achieve a vertical eyebox dimension of say 12mm would require an optic about that diameter as well. That just demands a big light engine that is difficult to integrate into a glasses form factor. Contrast that with a light engine from a SRG waveguide that has pupil expansion both horizontally and vertically and can have a light engine with lenses as small as 3-4mm in diameter. The output grating replicates the small pupils over a larger area to create a big eye box.

WaveOptics’ proprietary IP uses a method of achieving two-dimensional pupil expansion in a single SRG. This means smaller form factors and eased manufacturing tolerancing compared to other types of output grating that use two SRGs such as HoloLens and Magic Leap.

So overall, in my opinion the design space offered by SRG waveguides make them a better choice for most larger-volume applications. There is no question that in some circumstances reflective waveguide or reflective free-space type combiners are more suited.

YD: All in all, the market has been mostly driven by professional use cases with very low volumes. With the dream of this AR consumer electronics revolution that would replace smartphones, there is still a need for a compelling use case for the consumer. We believe this can only happen with the big consumer device OEMs like Apple, Samsung and Huawei jumping into the game. What do you ­­make of this?

PG: Entirely agree. The best AR hardware in the world would be unappealing without some great content and applications that are useful/fun for the consumer and had a clear business case for professional use. But certainly all the big companies now getting involved know their markets very well and have an implicit understanding of their customers. Until recently there has been no interest from the larger companies as the technology would not support any decent consumer-oriented industrial design. That is changing and improvements in the optics are a pathfinder for this to happen.

YD: Speaking of which, it seems like the supply chain for higher-volume AR headset components is setting itself up. For example there has been this press release of your partnership with Schott, EVG and Inkron. How was it built, and how strategic is this for your future development?

PG: There is no question building AR optics is hard and irrespective of how theoretically clever our structures are, poor materials would block them from working. So that is why we deliberately set out to partner with the world’s best providers of glass, nanoimprinting machinery and index matched resin respectively. We can’t build waveguides without these sorts of companies and similarly they can’t create adequate materials without feedback from waveguide developers.

YD: Speaking of the supply chain, optics are one part, but we tend to explain that optics and display engines cannot be considered separately. It looks like everyone is waiting for the microLED opportunity to happen, but in the meantime incumbent technologies will have to make do. What is your opinion on that?

PG: My opinion on microLEDS is the same as most people, that they offer great potential. It is an obvious thing to create a small projector if the light source and image former are one and the same device. Green-only 1080p panels are arriving but still do not have a very efficient means of coupling the very broad angular emission characteristics into a waveguide. Full colour panels are a way off. In the interim until they are ready? All I can say is we are working on novel Liquid Crystal on Silicon (LCoS) and Digital Micromirror Device (DMD) configurations plus scanning laser systems much like everyone else.

YD: What kind of business model are you targeting? Design and licensing? Optical foundry?

PG: Now, we are only supplying fully manufactured waveguides. In the future it is quite likely we will offer a design and license model to suitably resourced customers.

YD: What is your current status in terms of optical performance and manufacturing? What are your current focus areas for development, and your roadmap for the next few years?

PG: Today we feel we are the leading open market provider of AR waveguides and light engines all things considered. This is based on who we are working with on solutions but of course I can’t disclose that.  But as a company, we are not complacent and see several strong competitors whose products have performance attributes that exceed ours in some areas. Overall, I believe our balance of parameters and operational readiness make us today’s best option, but we need to keep pushing ahead to maintain our position.  With waveguides this means higher efficiency and better colour uniformity. It is the area of light engines where our biggest step up in effort will be – the push to get lower volume and higher efficiency light engines is very important to complement the waveguides.

YD: From design to manufacturing, what is typically the timeline to get a product? What are the main milestones?

PG: The graphic below shows the formula we use to create a totally new design based on one of our three platforms. From the Design Workshop through to physical waveguide prototypes takes around 16 weeks. The Design Workshop is typically a very intense two-to-three day event where we gather as much data from the customer as possible. Usually most engineering disciplines and industrial designers attend. Sometimes the data is intangible, such as what parameters matter the most and what parameters can be flexed a bit to get a better blend of performance. Our simulation tools give some pretty intuitive graphical representations of image quality as well as hard numerical performance data so we can give good visualizations as well.

All the prototypes we created in 2019 only required one loop around so were ready in one go for production from a design and performance perspective. It is product qualification in the customer structure and mechanical housings that take maybe another 6-9 months. So overall, 12 months from the Design Workshop to volume ramp is quite achievable.

YD: Do you feel the impact of this virus situation?

PG: Yes, we are feeling the social impact of COVID-19. Like many places we have largely closed our office and only essential lab work is going on. Socially, it is strange being apart from co-workers and getting familiar with about five customer video conferencing systems has been a task for us all.

But fortunately, the technical and business impact has not been anywhere near as great as it might be. Coincidentally with the office closure we received a couple of big design wins so for the next few months most of my R&D staff are on customer jobs simulating and working from home.

But we are all looking forward to normality resuming for sure.

Interviewee

Phil Greenhalgh graduated with a PhD from University of Kent in the UK and then joined the Electronic Engineering Department as a Lecturer. After 12 years in academic research and teaching, he left to join an IT start-up that had a number of contracts to do very large scale desk-top PC and software roll-outs across the UK banking and travel retailers. Appointments and ownership of various companies followed gradually evolving into a UK technology consultancy called 1066 labs. 1066 Labs accumulated significant awareness of system level augmented reality devices and optics by working with some very high profile customers and in 2016 was acquired by DAQRI. With the 2 years at DAQRI, Phil was head of R&D and latterly SVP Hardware before joining WaveOptics as CTO in September 2018.

Interviewer

As a Technology & Market Analyst, Displays, Zine Bouhamri, PhD is a member of the Photonics, Sensing & Display division at Yole Développement (Yole).
Zine manages the day to day production of technology & market reports, as well as custom consulting projects. He is also deeply involved in the business development of the Displays unit activities at Yole.
Previously, Zine was in charge of numerous R&D programs at Aledia. During more than three years, he developed strong technical expertise as well as a detailed understanding of the display industry.
Zine is author and co-author of several papers and patents.
Zine Bouhamri holds an Electronics Engineering Degree from the National Polytechnic Institute of Grenoble (France), one from the Politecnico di Torino (Italy), and a Ph.D. in RF & Optoelectronics from Grenoble University (France).

Related reports

Displays and Optics for AR & VR 2020
Optics are getting ready – now MicroLED displays are the next roadblock for the implementation of augmented reality.