Hearing lots of voices?

AI takes connected devices to the next level

In our smart world, a huge number of devices are part of the internet of things (IoT), or becoming so, many of them integrated with our homes, cities, manufacturing or transport systems and infrastructures. Added to this, a growing number of connected consumer devices, appliances and systems are able to carry out many human daily tasks in the home or workplace, whether for healthcare or entertainment. Research by Gartner forecasts the number of connected things will reach 20,8 billion by 2020, of which 13,5 billion will be from the consumer sector.

Personal robots help at home
Personal assistant robots play music, plan your day and control your smart home (Photo: Ubtech Lynx robot Brandon Widder/Digital Trends)

Artificial intelligence (AI) was one of the main themes at the 50th Consumer Electronics Show (CES) in Las Vegas this January, incorporated in devices ranging from hair brushes to cars, fridges and robots.

Two types of AI were prevalent, the first using voice as a way of operating smart devices, appliances and systems. The second theme concerned the way manufacturers are making their products more useful through deep learning. This form of AI enables the quick and effective analysis of huge amounts of data gathered by smart devices and used subsequently to improve their functions.

Standards supporting connected technology

The work of several IEC technical committees (TCs) and their subcommittees (SCs) contributes towards the development of aggregated learning technologies in general, and in particular to the area of voice recognition. IEC TC 100: Audio, video and multimedia systems and equipment, has set up Technical Area (TA) 16: Active Assisted Living (AAL), accessibility and user interfaces, which covers voice recognition. Additionally, aspects of artificial intelligence, speech recognition, machine learning and neural networks are dealt with by ISO/IEC JTC 1, the Joint Technical Committee (JTC) for information technology, established by IEC and the International Organization for Standardization (ISO).

IEC also has a Working Group dedicated to the IoT. Given the importance of this domain, it was decided at the end of 2016 to convert this into ISO/IEC JTC 1/SC 41: Internet of things and related technologies. This TC will focus on the JTC 1 standardization programme for the IoT, comprising sensor networks and wearable technologies, and provide guidance to JTC 1, IEC, ISO and other entities developing IoT-related applications.

Read more about this in the interview with Sangkeun Yoo, Convenor, ISO/IEC JTC 1/WG 10: Internet of things, in this issue.

Robots in all shapes and sizes

Finance, retail, healthcare, agriculture, manufacturing and many more industries will be impacted by robots, which will change our lives. In addition to this, the International Robotics Federation says that by 2018, sales of service robots for personal and domestic use will reach 35 million units.

CES 2017 was awash with such robots - lawnmowers, and personal assistants, both speaker versions that responded to spoken commands and humanoid versions - able to answer questions, do household chores and run other smart systems and devices in homes. Some humanoid robots already carry out meet and greet jobs in airports, providing flight or other useful information, while others will teach students skills and how to solve problems. Other developments are certain to follow.

The great potential of voice control

During CES, 170 000 people attended and watched as 20 000 products were launched over four days.

Shawne DuBravac, Chief Economist for the US Consumer Technology Association, said of the boom in AI technology, “In the last three years, the word-error rate in speech recognition has made great strides, dropping from 25 to 5 or 6%, which is similar to human speech”. DuBravac estimated that around five million voice-activated assistants have been sold to date and predicted another five million are likely to sell in 2017.

Activate the alarm

Many products at CES were operated by voice. Users ordered the different smart appliances and systems to dim, increase, switch on or off lighting, carry out various cleaning duties, arm or disarm a home security system, run the washing machine or select a specific TV programme and much more.

Making big data useful

AI information gathered by household appliances can add value to them in the form of extra features. For example, fridges no longer just preserve food. New models can manage their own temperature, scan food item bar codes and reorder groceries directly from the shop when needed. The personal assistants can connect to other home systems, find recipes or play requested music.

Smart washing machine apps allow users to start their machines from anywhere with an Internet connection, monitor washing at all stages, track energy usage and, once finished, send a notification via the app.

IEC produces International Standards for many components of this technology. Some examples include sensors, which are vital to connected devices (IEC TC 47), automatic identification and data capture techniques for bar code technology (ISO/IEC JTC 1/SC 31), audio, video and multimedia systems and equipment as well as touch screens in an increasing number of our home appliances (IEC TC 100). These Standards help to ensure reliability, quality and interoperability of these systems and devices, as well as to establish a common vocabulary and test methods.

Start the car please

As part of the internet of things, research by Gartner estimates that a quarter of a billion connected cars will be on the road by 2020, enabling various levels of automated driving and opening up the scope for new in-vehicle services. For example, such cars could connect to an intelligent street lamp and be guided to a nearby free parking space. 

As auto manufacturers race towards achieving fully-autonomous driving by 2020, cars are becoming more intelligent and complex. However, virtual assistants will help make them easier to interact with.

Car makers showcased various manifestations of AI technology that will revolutionize how and what we drive in the future. We will gradually drive less and talk more to all the connected technology and infrastructures in and outside the vehicle. We will tell our cars to open the connected garage doors, start up, adjust the temperature, and where to drive. Some cars will be able to connect to smart home systems and appliances while on the road.

Deep learning will enable driver profiles to be established, based on monitoring driver habits such as the radio stations they listen to and how they drive, with the data collected from sensor-embedded chairs. These cars will recognize drivers before or after they get in, using facial recognition capabilities, and automatically load the specific profile when the driver “logs in” or starts the car.

In a fully-connected world, cars will monitor vast amounts of technical information in real time and be able to signal in advance when a part is going to break down, then connect to the nearest garage and find out if the replacement piece is available.

Facing the future

Facial recognition technology is already used in security systems in government, commercial or industrial contexts. Consumer technology will adopt it, for example to access smart home systems or connected cars. Registered users will be alerted if an unrecognized person enters the home, while cars could be stopped in their tracks if a registered driver profile isn’t activated.

Several subcommittees of IEC/ISO JTC 1 produce International Standards for biometric technologies pertaining to human beings, to support interoperability and data interchange among applications and systems. They also cover biometric data protection techniques, biometric security testing and evaluation methodologies.

Making the IoT secure

Finally, anything which connects to the IoT could be subject to cyberattacks. As AI technologies become more widespread, gathering huge amounts of personal information, ensuring the effective provision of data protection and privacy will be paramount.

The IEC undertakes a number of activities towards this. ISO/IEC JTC 1/SC 27 has produced the ISO/IEC 27000 family of International Standards on security techniques for IT Standards, which covers many aspects, including detecting and preventing cyberattacks. Other examples include protecting data in medical equipment, industrial automation and energy generation.

The IEC Advisory Committee on Security (ACSEC) is mandated to deal with information security and data privacy matters which are not specific to a single IEC TC; coordinating activities related to information security and data privacy and providing guidance to TC/SCs about this, both generally and for specific sectors.

In addition to the TCs which publish International Standards on this topic, Working Group 3: Cybersecurity Task Force, of IECEE, the IEC System of Conformity Assessment Schemes for Electrotechnical Equipment and Components, was established in 2014 to check cyber security for industrial automation against the IEC 62443 series of International Standards.