At the start of 2017, we gave our predictions for the year ahead and with 2018 rapidly approaching we’re now made our predictions for 2018. Here, Simon Forrest, Director of Segment Marketing at Imagination, bravely reviews some of our predictions from last year and provides his own thoughts for the upcoming 12 months.
We predicted that 2017 would be the year of hardware ‘insecurity’. While there has been a long list of well-publicised data breaches, we were more fearful of an exploit of popular IoT devices or a compromise of a smart ecosystem that would result in serious damage or service outage. This would have negatively impacted public opinion of smart technology and potentially placed market adoption back several years.
Throughout 2017 we’ve not seen anything significant that would shift public perceptions; meantime experts in penetration testing continue to illustrate potential exploits and vulnerabilities in toys and consumer appliances.
Overall the multi-domain security story is beginning to take hold. Indeed, Imagination built up some great traction with operators such as Deutsche Telecom, Telecom Italia, and notably BT, with these vendors understanding the need for hardware-backed security to be built into the silicon chips they now rely upon.
Artificial Intelligence (AI)
We predicted that artificial intelligence will become useful in 2017. In my view, despite all the positive progress we’re seeing through ecosystems such as Google Home and Amazon Alexa, when it comes to truly effective use of AI it’s still very early days. One change I think will happen is that increasingly more of this ‘AI’ processing will happen on the device. We certainly advocate this as it offers benefits in terms of security, latency and bandwidth while making issues around data protection easier to manage. Indeed, changes in data protection laws across the EU next year will force all companies to review their data harvesting and management policies, and this certainly will have a knock-on effect on how systems employing AI are designed.
Smartphone devices offering integrated facial recognition powered by a hardware network networks accelerator, coupled with our launch of our own PowerVR 2NX, yields credence to our prediction.
However, I believe that the applications that truly harness AI are still yet to emerge. Even some of the leading tech giants don’t yet know how best to exploit the technology: face recognition to unlock smartphones is one thing; animated “emojis” are totally underplaying the power of the technology and are largely just showcases of the capability that is available. But those very same companies are playing a smart game, opening up their AI ecosystems to developers in order to broaden the applications and seed wider adoption. In this instance, my prediction is 2018 will indeed show the appearance of some useful applications of AI.
In all honesty, the move to 360° video hasn’t been as rapid as most investing in the technology would have liked. First introduced in 2015, we need to see VR head-mounted displays become successful in order to fully enjoy the capability of 360° video. The smartphone or TV equivalent – rendering on a flat screen and demanding user interaction – feels gimmicky in comparison. Surround audio formats also have to mature in order to improve immersion. Meantime YouTube is working on more mundane sounding, but arguably more practical, 180° video, reasoning that consumers don’t really need a full surround view (as 75% of users only look at the quadrant in front of them). The half-hemisphere alternative could be more successful, notably in reducing production costs and sharpening the image, as the full 4K resolution would be available across a smaller field of view. 180° video also provides a safe haven for the TV production crew… they’re no longer hiding in shot!
We expect VR devices will eventually “cut the cord” but this hasn’t happened to a large extent in 2017. We believe VR experiences will become untethered, using embedded processing to drive the costs down, and capture the consumer’s hearts and minds.
From a technology perspective, we distinguish between standalone devices, where the processing is done in the headset and tethered devices where the processing is achieved on an external device such as PC, even where this employs a wireless link to the headset. Analyst predictions are improving for standalone VR devices, with volume forecasts increasing on each review of the market conditions.
Untethered VR headsets don’t yet have the high-end immersive capability today but are on a pathway to success. The addition of surround audio will enhance the feeling of presence; haptic feedback will improve the immersion, and indeed our powerful embedded GPU and NNA technologies will help create the illusion of a fully interactive environment.
Overall, VR devices will become more cost-effective and capability will steadily improve. But more importantly, the learning and development achieved in this phase will inform technical choices for augmented and mixed reality devices to follow in the next few years.
Voice recognition and interaction
It seems like virtual assistants are virtually everywhere in 2017. Whether it’s Alexa, Cortana, Google Assistant, to name the well-known ones, or Duer, GiGa or Sherpa as examples in other markets, these Virtual Personal Assistants (VPAs) have infiltrated homes, cars and more inside various products. Notably, in 2017 Amazon and Google launched a new generation of products to further push technological advances across the ecosystem.
According to industry estimates, over 12 million were units were sold worldwide in 2017. Clearly, users are beginning to understand the benefits of VPAs in their day-to-day lives, but they are also concerned about the privacy issues. Moreover, a recent study suggested that 30 per cent of people abandoned using their VPA regularly, as they offered little benefit over using a smartphone for the same interaction. But this is just the first wave of technology; the addition of screens enable VPAs to become more adaptable, enabling conversations to be more fluid as they can display a selection of choices rather than reading each one out aloud.
For 2018, I expect to see AI and NNA capabilities to improve these devices further. For example, being able to hold up an item in view of a camera-enabled VPA and simply say “order me two of these” would be a major step forward.
In 2017 we did a lot of work in promoting virtualization to operators, resulting in our invitation to present at the BT Innovation Week back in June 2017. We translated the technical elements of virtualization into tangible business benefits for operators and OEMs. They’re now beginning to request hardware-enforced virtualization in home gateways and IoT products. In all honesty, we’re probably still two years away from seeing routers/gateways harnessing the technology, but we’re making good progress.
Meanwhile, hardware virtualization is beginning to make inroads in the automotive market in 2018 as car manufacturers begin to specify IP, such as our PowerVR Series8XT, that features hardware virtualization support, enabling them to offer services securely in the infotainment system. Car manufacturers want to be able to maintain control over the car environment rather than relinquish control to the mobile OS providers, and hardware virtualization will be an essential technology that underpins this objective.