We’ve talked about how semiconductor memory connects our homes – running through our appliances, supporting our gaming systems and enabling our smart fitness equipment. But how does it connect our homes to other homes, and cities to other cities?

Whether we realize it or not, the beating heart of modern life is powered by semiconductor memory. In cities around the world, memory-enabled technology like 5G connectivity, IoT capabilities, augmented reality (AR), virtual reality (VR), artificial intelligence (AI) and automotive tech keep our lives running at lightning speeds.

The world is positioned to expand how AI products think and learn, to build networks of smart cities and hospitals through IoT’s enhanced data depth, to help autonomous vehicles process information and make split-second decisions as they learn new roads and situations, to discover new worlds through augmented reality, and to lift communities out of the dark by bringing Big Data to light.

Image Download

Beating Heart of Modern Life

Today, the world runs on 5G. The 5th generation mobile network, 5G was designed to connect anyone – anywhere, any time – at ultra-fast speeds. Our smartphones, computing devices, smartwatches, and smart medicine machines all rely on 5G to receive and transmit information around the world at record speeds. Industries from transportation to healthcare to information processing benefit from the high performance and efficiency of the 5G network, and all of the technologies it enables.

And behind all of these smart devices and machines is the need for memory. Generally speaking, more information transmitted means more information stored – and the more information stored, the greater the need for sophisticated memory systems.

Source : Pixinoo/Shutterstock.com

Image Download

For Heightened Security

One of these memory-enabled technologies is facial recognition CCTV – which relies on memory to process and store the data from billions of unique faces. In China, two AI companies called Megvii and Sensetime¹ are providing face recognition AI technology to governments and police forces.

Just recently, their technology was instrumental in catching a criminal in Nanchang, Jiangxi Province, China. The suspect was identified by concert security footage that was a match with a face in the country’s surveillance database. In Malaysia, the police force has even begun to use this type of facial recognition technology in body cams to capture images of suspected criminals and quickly check them against the police database. According to E2 Technologies², these tools have helped reduce the crime rate in regions where they are used.

Source : Bear Robotics

Image Download

Autonomous Goods and Services

AI technologies have more light-hearted uses as well – especially in the worlds of hospitality and dining, where robots are now helping with everyday functions in hotels, restaurants and bars. At Amichi’s³, a pizza restaurant in Mountain View, California, a self-driving robot named “Penny” is serving orders.

And they’re not alone – more and more American restaurants are using memory-embedded robots from simple food preparation to serving and washing dishes. Robots not only streamline the work of restaurant employees, but also reduce overall operating costs. According to The Spoon, 2020 will be the year for ‘food robots’ to enter the market in earnest.

Similary, as autonomous vehicles for the delivery are quickly approaching on the horizon, they require functions such as multi-camera vision processing and self-driving in order to monitor surroundings around the vehicle. Thus, ultra-huge computing processing has become necessary, so has semiconductor memories.

In fact, over the last few years, and across many states in the US, autonomous vehicles for the delivery of food and goods have been approved for real-world testing from some of the top automakers and technology companies including General Motors, Ford, Toyota, Google’s Waymo, and Uber, among many others. These companies have been partnering with grocery stores, restuarants and small businesses to help ease the problems of first/last mile delivery.

Source : Nuro’s blog site. Nuro R2 self-driving delivery vehicle in Houston, Texas

Image Download

Technology startup Nuro⁴ partnered with grocery giant, Kroger, to launch pilot cities in Scottsdale, AZ and Houston, TX, where they have been testing autonomous vehicle delivery for residents with its small toaster-shaped vehicles, called R2’s. During the COVID-19 pandemic, the company has seen increased use as the company can safely deliver groceries to citizens and limit person-to-person contact.

Creating Memory for the Future

As the world grows increasingly reliant on high-speed connection, demand for DRAM and NAND Flash memory solutions is higher than ever. And memory providers like SK hynix are innovating nonstop to keep up.

These technologies – and the memory systems that support them – are leading the way towards the smart cities of tomorrow, and SK hynix is at the forefront of the movement towards the future of information technology.

¹Blake Schmidt, “Hong Kong Police Already Have AI Tech That Can Recognize Faces”, Bloomberg. (2019)
²Amanda Lentino, “This Chinese facial recognition start-up can identify a person in seconds”, CNBC. (2019)
³Ahn Jung-rak, “Cooking is basic, serving and washing dishes… Robot wind blowing in an American restaurant”, Hankyung. (2019)
⁴Dave Ferguson, “Introducing R2, Nuro’s Next Generation Self-Driving Vehicle”, Medium. (2020)

Puzzle:

The post [Invisible Memory: Part.3] Semiconductor Memory in Our Smart Cities first appeared on SK hynix Newsroom.

To the Virtual World by Just Wearing Glasses… “Smart Glasses” with a VR function, When Would They Be Commercialized?

A daily life with virtual reality, everyone has imagined this at least once. But this is a technology of the future that is not yet feasible in reality. Nowadays, it is easy to play simple virtual games or to experience VR video content through a VR Head Mount Display (HMD) connected to a PC, a smartphone or a console. However, the “smart glasses,” which enable VR at any time or anywhere, can yet only be seen in films.

Nonetheless, it is not right to consider smart glasses as Science Fiction (SF) which will be only possible in the remote future. As IT technologies are being rapidly developed with the arrival of the 4th Industrial Revolution, the technologies required for VR are now being released one by one. As global IT companies are actively entering the VR industry these days, it is expected that smart glasses in movies will be realized in reality in the near future.

Image Download

The volume of the related market is expected to grow rapidly as well. Digital Capital, a British market research institution specialized in VR and augmented reality (AR), predicted that the volume of the global VR market will increase quickly from USD 3.9 billion in 2016 to USD 105 billion in 2022. Another global market research institution, IDC, also expected that the volume of shipments of the related hardware (HMD) will increase from 9.6 million units in 2017 to 59.2 million units in 2021.

Today, Facebook is the most active company in developing smart glasses. Since the acquisition of an HMD-developing startup named Oculus in 2014, Facebook has been expanding its VR-related product lineups, jumping up to be a market leader in the VR HMD-based hardware platform development. Recently, it started to develop a next-generation AR smart glasses code-named Orion, aiming to commercialize them in 2025. Mark Zuckerberg, CEO of Facebook predicted the future of smart glasses at the annual developer conference in 2017, saying that smart glasses will replace smartphones in five to seven years.

(Credit: Vive Newsroom)

HTC and Sony are also considered as the market leaders as well. HTC has been showing notable movements recently in the high-end VR HMD market, while Sony has been presenting VR HMD based on its own gaming platform PlayStation. Especially, without considering cost-effectiveness, Vive Pro launched by HTC in 2018 is one of the best VR HMDs with the highest specification ever, as it boasts a resolution of 2880 x 1600 pixels, a field of view (FOV) of 110 degrees, and a refresh rate of 90Hz. In South Korea, Samsung Electronics is joining the related market by a smartphone-based VR HMD called “Gear VR” through collaboration with Oculus.

Along with the VR HMD, technological developments regarding Eye Glassed-type Display (EGD) based on the AR and mixed reality (MR) are being made continuously. Some of the most recognizable companies are Magic Leap and Microsoft.

Magic Leap succeeded in developing a “Photonic Lightfield” technology, which controls the direction of light reaching the retina, and unveiled its MR EGD named “Magic Leap One” in 2018, which makes virtual objects look more realistic. Microsoft also raised expectations of the market by unveiling its next-generation AR EGD, HoloLens 2, at CES this year.

Nonetheless, Magic Leap One: Creator Edition is aimed mostly at developers, as you can see from its name. Microsoft is also currently focusing on the AR EGD market for enterprise rather than for consumers, suggesting that commercialization of AR glasses would require more time.

Current Status of “HMD,” a Prototype of Smart Glasses

In general, for implementation of VR, convex lenses are placed in front of a user’s eyes, so that focal points can be adjusted to give a three-dimensional effect and a sense of distance to images delivered by an exterior display. In other words, this method utilizes a type of optical illusion, making the brain think the virtual images as real. With currently available technologies, the user’s vision should be fully covered, as a sense of immersion can be raised when the visual information in the reality is blocked from the user’s view. Furthermore, there should be some space for the display device, as well as enough space between lenses and the display to adjust the focal distance. This is the reason why all the commercialized VR HMDs so far are adopting a design element of a goggle shape.

This goggle-shaped VR HMD is far too heavy and uncomfortable to wear for long in daily life. To evolve from this to the shape of glasses, the first step should be the weight reduction of peripheral devices, including sensors and batteries. Along with this, development of other related technologies is required as well, such as optical technology and display technology.

Image Download

Nevertheless, current optical and display technologies have clear limitations. To allow a user to be immersed in VR without any blocking of exterior visual information, the FOV of HMD should be able to cover 220 degrees, the maximum FOV of human eyes. However, even high-end products of the major HMD manufacturers in the current market only cover around 110 degrees, which is less than the average FOV of human eyes, i.e., 120 degrees.

The current level of image resolution is not good enough for allowing users to sense the released images as reality. To maximize the sense of immersion, a high-resolution over at least 4K (3840 x 2160 pixels) is required. Especially, in the case of a VR HMD which shows images with a wider FOV than previous displays, the Screen Door Effect, where fine black lines between the pixels make you feel like you’re looking through a mesh screen, should be considered as well. This means that, in 360-degree VR, the resolution sensed by users can be much lower than that of the original videos. To solve this problem, it is expected that high-resolution videos over at least 8K (7680 x 4320 pixels) are required.

Image Download

Securing a stable frequency of display (refresh rate) is crucial to minimize 3D motion sickness that can occur when using VR HMD for a long time. While the industry defines the refresh rate for the optimal VR technology as 120Hz, current products based on Organic Light Emitting Diodes (OLED) have refresh rates of around 90Hz.

“Development of Semiconductor Technology,” a Prerequisite for Commercialization of Smart Glasses

The development of semiconductor technology is one of the most important factors for the commercialization of smart glasses. Even with the great image resolution and display technology, the commercialization of smart glasses is still far-off without semiconductor performance supporting and implementing them. For this reason, in the semiconductor industry, diverse efforts are now being made to secure technologies to lead the VR sector.

(Credit: Qualcomm Website)

In the Application Processor (AP) sector, Qualcomm stands unchallenged. Qualcomm’s Snapdragon 835 was adopted by both of two high-end Standalone VR HMD products in the current market, Facebook’s Oculus Quest and HTC’s Vive Focus. Meanwhile, Intel is actively focusing on VR systems for professionals, based on PC-powered VR HMD, with its high-performance CPU “Core X Series.”

AMD is targeting the global VR system market in the GPU field, with its LiquidVR, a solution exclusively for VR, and AMD Radeon VR Ready Premium / Creator, its GPU certification programs. It also acquired a wireless VR chip manufacturing startup named Nitero in 2017 to secure technologies.

Technological development of memory semiconductors is also a prerequisite of the realization of smart glasses. It is because memory semiconductors are used during the entire process of providing a realistic VR environment for users; from producing and saving VR images and securing storage space within a device, to releasing images on a display. Especially, the development of high-performance graphic DRAM is essential for receiving large amounts of data of 8K images in real time, to show this to users in a stable way.

In general, one piece of Full HD (2K) image accounts for around 9.5Mbyte. It increases up to 16 times for 8K (UHD). In addition, VR images consume around four times as much data as ordinary UHD images, as VR images should be displayed in an enlarged form, in a short distance from eyes. In other words, a huge amount of data which is at least 64 times larger than ordinary Full HD images should be processed in real-time for VR images.

Image Download

To meet these conditions, how much more should the graphic DRAM’s performance for VR supporting system be developed? According to SK hynix’s own estimation, 8K images can be processed stably without any delay through a graphic DRAM with a data transfer rate of over 20Gbps for videos in high resolution over 8K with more than 90 frames per second. It means that a high-speed graphic DRAM like GDDR6, the latest one released, is required. For memory semiconductor adopted within the HMD device itself, a design with efficiency and low power consumption is necessary as well, so that data can be processed rapidly in a subminiature chip equipped inside the glass frame.

Seoyeong Jeong, Technical Leader of Graphic/Auto team at DRAM Product Planning department of SK hynix said, “As VR technologies are likely to continue to be sophisticated and used in a wider variety of applications, it is expected that the volume of data to be processed will increase exponentially. To respond to this trend of the future market, SK hynix is making continuous investment in the high-bandwidth, low-power memory semiconductor.”

The post Virtual Reality in Glasses first appeared on SK hynix Newsroom.