AI on the agenda
Artificial intelligence (AI) is no longer on the horizon — it’s here, and we need to be ready to make use of it.
That was the key takeaway of the 2018 Hong Kong Electronics Fair (Autumn Edition), organised by the Hong Kong Trade Development Council (HKTDC) and held from 13–16 October in conjunction with electronicAsia. But rather than being an ominous warning about robots taking our jobs, the overall tone of the show was one of optimism, with the future being a world where AI is simply a tool — albeit a very useful one — that is set to make life easier both at work and at play.
The fair opened on 13 October with the Symposium on Innovation & Technology, exploring the theme ‘AI Empowerment — Grow Without Limits’. Opening remarks from the Hon Nicholas W Yang, Secretary for Innovation and Technology from the Hong Kong Government, revealed that AI is developing rapidly around the globe, particularly in the manufacturing industry in order to improve efficiency, and the government has set AI as a key focus area.
First studied as a serious field of research in the 1950s, AI these days refers to a collective umbrella of connective technology that analyses behaviour and predicts and develops patterns based on this behaviour. Some examples include computer vision, machine learning, speech recognition and translation, and robotic process automation, to name a few.
For example, Hong Kong-based company SenseTime specialises in computer vision — a technology which can be deployed in a wide variety of areas. In the area of autonomous driving, computer vision enables cars to ‘see’ other cars as well as pedestrians, at any time of day and in any kind of weather condition. It is also useful for assisted driving, alerting the driver if it senses the car drifting lanes, or even if the driver is using their phone or smoking. Insurance companies are especially fond of the technology, as it encourages responsible driving, as are logistics companies who want to be able to monitor their drivers.
SenseTime also works in the area of smart retail, offering data analytics on customer flow (particularly useful at stores such as IKEA) and facial recognition using in-store cameras. The technology has even led to the rise of unmanned convenience stores in China, where a camera acknowledges the customer as they pick up items from the shelves, scans them at the checkout and deducts money from their eWallet.
We’re also moving closer to the world of service robots as depicted in The Jetsons, with the fair including a demonstration of a robot named ‘Yoyo’ from CANBOT in Beijing. Featuring facial and voice recognition, plus obstacle avoidance and motion, Yoyo can be used as a novelty guide in hotels, shopping malls, banks, hospitals and even schools. It is also a modular robot, meaning its parts can be easily replaced so they are more suitable for specific tasks.
Less humanoid, but similarly useful, is the PudoBot delivery robot from PuduTech, which can be used as a waiter in a restaurant as well as in casinos, hotels and hospitals. Deployed from a restaurant kitchen, for example, a member of staff simply needs to place the food on the robot’s antislip map and input the number of the customer’s table. PudoBot then maps out the most efficient route and begins its journey, detecting and avoiding other robots, humans and objects as it slides along. Once it has arrived, the customer can take their food and press a button to send the robot back where it came from.
So what do businesses themselves want out of AI? This question was answered by Vincent Wong, Associate Director, Consulting, from Deloitte China. His company conducted a survey of 250 senior executives which saw 87% of respondents say AI is important to products and services offerings and 92% say AI is important to internal processes.
Wong suggested that AI can be deployed for processes included back-office accounting, human resources, supply chain management, IT maintenance, data analytics and even customer service. Yet when asked what benefits they sought from AI, the majority of senior executives were focused on enhancing products and services, rather than reducing headcount.
“AI is a tool,” Wong said. “AI is your friend. AI is coming to enhance the products and services you are offering … Your people will be part of the journey.”
Of course it cannot be denied that automation will replace some jobs, as noted by Dr MeiKei Ieong, Chief Technology Officer of the Hong Kong Applied Science and Technology Research Institute (ASTRI). He referred to two reports: one from the University of Oxford, which claimed that up to 47% of current jobs were at risk from automation, and one by the OECD, which concluded the number was only 9%. Why the discrepancy? Dr Ieong listed three possible reasons: adopting new technology is normally a slow process; workers can adjust to changing technology by retraining to do other tasks; and technology change typically generates more jobs.
Training and retraining are therefore key to keeping employment steady. And this is already happening, with Dr Ieong pointing out that cities such as Paris and San Francisco provide free coding universities and even free housing for students, and work with companies to supply tech workers.
“AI will take over a lot of tasks, but will free humans to adopt more value-added, personalised and analytical roles,” Dr Ieong concluded. SenseTime Vice President Jessie Lin added that AI simply would not work without human intervention: the future is therefore human + AI.
Two days later, AI-enhanced reliability was discussed at the event as part of the Hong Kong Electronic Forum. This covered AI less as a way to improve a company’s products and more as a way to ensure such products are fit for purpose in the first place.
Professor Michael Pecht is Director of the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland, which specialises in assessing the health and predicting the failure of electronics. He claimed that AI, particularly through machine learning, can be used to categorise data, discover relationships, identify patterns and determine anomalies.
Specifically, CALCE’s machine learning systems use Mahalanobis-based detection methodology, which looks at how different parameters should work together and recognises when they don’t. This enables them to predict when subassemblies, including motherboards, power supplies and disk drives, should be replaced. Like a doctor, he said, the system becomes more experienced over time, taking a prognostics-based approach that ultimately helps with preventing catastrophic failure, forecasting maintenance and planning logistics.
“AI is providing the tools that are enabling us to put a doctor on the shoulders of today’s electronics,” Prof Pecht said.
This idea was expanded on by Norbert Meuser, Managing Director of Viscom. He focused particularly on machine vision in surface-mount technology (SMT), noting that humans are typically good at identifying objects and classifying them — but less reliable when it comes to mass inspection of electronics in the factory. That’s where machine vision comes in.
Unlike machine learning, machine vision is based on very specific programming, with deterministic algorithms ensuring the same input always results in the system providing the same output. This is far less likely in humans, who could give two different interpretations of the same information.
Viscom’s machine vision technology has culminated in a process called automated optical inspection (AOI), in which a machine is programmed to look for specific defects in PCBs. Once the system has identified a defect, this is confirmed via human verification (Meuser suggested that we may one day reach autonomous verification — with no human operator at all — but this day is not yet upon us). And if defects are consistently identified in product after product, that could be an indication that there’s an issue with the production equipment itself — perhaps it just needs to be cleaned?
Product reliability can even be monitored on an ongoing basis, according to Patrick Kabasci from the KEX Knowledge Exchange and INC Invention Center, whose field of choice is Industry 4.0. Kabasci noted that Industry 4.0 is often associated with automation but is also about using data — particularly real-time data — for predictions and decision-making.
According to Kabasci, data integration can be an ongoing process over the entire product life cycle — from predicting its original lifetime, to tracking and tracing it in the field, and finally seeking to improve it based on the real-time data collected. This could prevent companies from being subjected to mass recalls, with tiny changes in product parameters (eg, a raised temperature) enabling manufacturers to pick up on potential failures sooner and faster than previously. This will become easier in the future, Kabasci said, with the emergence of 5G and the increasing integration of small, cheap sensors into products.
It’s clear that AI has come a long way since the ’50s, and there’s no going back. So if your company has been reluctant to enter this brave new world, you might want to consider taking a look at what’s out there. Shows such as the Hong Kong Electronics Fair and electronicAsia are a great source of new and innovative products, as well as informative seminars featuring industry experts. And who knows — you might just be surprised to learn how AI can enhance your business for the better.
Engineers have successfully measured the fidelity of two-qubit logic operations in silicon, with...
In-mould electronics will enable high-volume production of structural electronics where the...
Italian scientists have achieved a billion-fold increase in the electrical conductivity of...