Artificial Intelligence / Machine Learning
Although true intelligence seems a little way off yet, and whilst in Silicon Valley the term AI has become a bit of a joke because of the hype around it, Machine Learning is finding real-world niches. We believe that 2020 will see it integrated into some innovative and real products. For example, computer-assisted gun/weapon detection in public buildings and airports from live camera feeds; heavy machinery cut-outs on abnormal inputs to minimise potential injuries.
Smarter car integration
We can look forward to cars either coming with data plans or piggybacking off mobile Wi-Fi during 2020, learning habits/routes, such as a daily commute. This technology could be used to help redirect traffic or suggesting alternative less busy routes. We’d expect it to be able to access and record using a car’s internal cameras (reverse parking, lane departure, collision warning). Telematics will increase in use for road, vehicle and insurance purposes.
EV chargers: rolled out for new builds, greater security focus
EV chargers will start to become a standard part of any house that has off-road parking. All new builds need a charger fitting. Therefore, adoption will increase steadily. Communication standards for energy infrastructure are still under development, however, OCPP (Open Charge Point Protocol) seems to be leading the way and will do so throughout 2020. Security threats against a connected EV infrastructure can be devastating - expect to see developments to protect smart chargers from attack.
Greater hardware acceleration for FPGAs
Next year should see a big push into machine learning-driven by tools from Xilinx and Intel, allowing software engineers to take advantage of hardware acceleration. The area of machine learning requires a lot of computational power for real-time applications. FPGAs are well positioned but due to their high development costs, it has been prohibitive for engineers to leverage the capabilities of these devices. With the recent deployment of high-level C based tools aimed at software engineers, it is now possible for machine learning applications to take advantage of large gains in performance and power efficiency offered from hardware acceleration in FPGAs.
The Rise of RISC-V
We expect 2020 to see a big increase in silicon vendors offering real RISC-V architecture chips. An alternative to ARM, they are completely open-source and so no royalties are paid to the copyright owners. There has been a flurry of discussion and real software written for RISC-V - including Linux kernel support – however, little real hardware. There are a few niche AI designs in the wild, but so far nothing from the big silicon manufacturers. We anticipate a significant change to this during 2020, with RISC-V chips being announced by some of the large silicon heavy-hitters.
Big Data v Big Brother
Controversial discussions surrounding facial recognition, the use of anonymised data, locked-down internet in some parts of the world and the “Big Brother” state will feature more prominently over the coming year. There will have to be a fine line between gaining access to everyday data to help the betterment of society and careful access to this data so that it isn’t misused by the industry or anyone with malicious intent.
Evolving virtual assistants
There has been growth in virtual assistants with built-in screens such as Echo Show or Google Home Hub. Adoption of these devices will increase during 2020 as they can serve a dual purpose, such as acting as a mirror when not actively in use, or are able to blend more into the background.
Against stiff competition, manufacturers will move away from using “neutral” virtual assistants next year and start giving them personalities. Amazon is starting this with celebrity voices such as Samuel L Jackson, which come with optional explicit modes.
More sophisticated mobility products
Existing mobility scooters are typically very basic – just a speed controller and horn.With an ageing and increasingly tech-literate population, we expect this to change and for more features to be integrated during 2020 - greater range, better battery management, Bluetooth or other wireless connectivity for location tracking and app link-ups, security features. Heated seats anyone?
Smarter shopping
Supermarket scenario: type in the product you are searching for and follow your trolley as it takes you to the isle where your item is located... could even go one further and scan it for you and weigh it as you pop the item in so all you have to do at the end is pay at the machine with a ‘trolley code’ -a new level of the ‘scan and go’.
Increase in automatic medical diagnoses
If given access to large datasets in 2020, AI and Machine Learning could help spot developing conditions and perhaps more usefully draw correlations to help aid early diagnosis of terminal conditions.
Smarter security
The general opinion for the use of smart home products is for convenience or security. Smart locks are undergoing major developments and combine security with convenience. Intercoms with cameras and voice communication have been around for years, but not over the internet.
We anticipate smarter locks during 2020, which will grant the user the ability to remotely check who is at their door, unlock their door to trusted individuals such as for deliveries - and even communicate with said person. On the security side, multiple technologies are being tested, with the most promising being biometric recognition.
Also, expect to see; higher quality, lower latency, IP cameras available with better operation in low light conditions, tied to both home and wider business applications with IoT functionality for crime prevention; super low power GPS trackers leveraging NBIoT for automobiles and other “easy” target items.
Intelligent drones and curbing rogue pilots
As the market opportunities continue to grow, so do the predicted sales for the drone industry. As well as campaigns raising awareness of new drone registration laws, expect to see advances in sensor technology for data acquisition, imaging and real-time data processing across drone technology during 2020. Drone microcontrollers implemented with machine learning models are being researched for a multitude of application including agriculture field analysis, infrastructure inspections, terrain, mapping, automated deliveries, identifying individuals in disaster search and rescue scenarios to name a few.
In 2020 we should see campaigns raising awareness of the law to establish its importance and help curb rogue pilots. Until recently, drones were largely unregulated, but a new category for the registration of remote-controlled “aircrafts” will help to deter disruptive pilots. Online tests make registration easily accessible to all, there is no limit on the number of attempts and a multiple-choice test is straightforward.
