AI And The IoT: Leveraging Digital Disruption

By Sani Theo

356

This article discusses the combined applications of artificial intelligence (AI) and the Internet of Things (IoT) technologies, which are set to transform the modern industry and lead to a bright and promising future

Imagine a piece of wood doing amazing things using artificial intelligence (AI) and the Internet of Things (IoT)! This is exactly what mui smart wooden plank (Fig. 1) does. It is an interactive wood panel for smart home applications, created by a Japanese company and showcased recently at Consumer Electronics Show (CES) 2019, Las Vegas.

mui is an Internet-connected plank of wood with built-in microphone, touch-sensitive interface and Google Assistant. It can display information including weather, and also control Internet-connected devices and smart assistants. Its design interface includes elements of AI and the IoT. From metal (conductor) to semiconductor, and now to wood! Who knows, in the future, with the power of technologies such as AI and the IoT, your chairs and tables could become smart enough to start tracking your daily routines and activities.

Currently, AI and the IoT are the buzzwords in the industry. These technologies fascinate many researches and industries across the globe. A combination of these two technologies is revolutionary, and is set to transform the landscape of modern industry. The growth of both technologies will impact the industry, including its structures, transactions and practices.

These technologies will impact future businesses and help reduce global poverty, solve water shortage problems and improve agriculture productions. These will greatly benefit emerging countries like India and Africa. A smart connectivity ecosystem coupled with AI will be a disruptive force that will change the world in a big way. Almost everything will be smarter, including self-driving cars, smart homes and smart cities.

This article is not about the merits and demerits of AI and the IoT; rather, it is about the combined applications of these technologies, which are set to have a bright and promising future in transforming the modern industry. AI and the IoT work hand-in-hand in areas such as Big Data, edge computing, smart homes, automotive, healthcare, agriculture, manufacturing, wearables and military, among others.

Convergence of AI and the IoT

Before delving into AI’s and the IoT’s specific applications, let us first see the main differences between them and then their point of intersection.

mui smart wooden plank (Credit: www.kickstarter.com)
Fig. 1: mui smart wooden plank (Credit: www.kickstarter.com)

Broadly speaking, AI and the IoT are two different types of technologies. AI emphasises on making of intelligent machines work and react like human beings, including speech recognition, planning, problem solving and learning. It involves programming intelligence or making computers behave like humans. AI-powered devices try to imitate the natural intelligence of humans, mimicking the cognitive functions that humans use to perform tasks.

On the other hand, the IoT is the interworking of physical devices like sensors and actuators, which are able to communicate among themselves or with the external environment including cars, home appliances and human beings. The IoT involves extending Internet connectivity beyond standard devices, such as computers and smartphones, to any range of non-Internet-enabled physical devices and everyday objects. As per Wikipedia, the number of IoT devices increased 31 per cent year-over-year to 8.4 billion in 2017, and it is estimated that there will be 30 billion such devices by 2020.

The IoT can be found in wearable devices, smart homes, various modes of transportation, agriculture and manufacturing. With the explosion of IoT devices, unprecedented amounts of data generated by the IoT daily across the globe will have a huge impact across the entire Big Data universe.

Since data volume and complexity are growing everyday, the IoT Big Data has thrown new security challenges. One cannot easily distinguish between useful, harmful and redundant data. Collecting data is an easy process as compared to analysing it. To use data in a meaningful way, it must first be checked and processed with advanced tools (analytics and algorithms). This is where the role of AI comes in.

AI often revolves around the use of simple to complex algorithms. Heaps of data can be managed and used efficiently and effectively using AI. Many vendors of IoT platform software are now offering integrated AI capabilities. IoT devices with AI could be a good combination to serve humankind in a better way.

AIOps—a new term coined by Gartner—is the outcome of disruptive technologies. AIOps platforms utilise Big Data, machine learning and other advanced analytics technologies to support and enhance IT operation processes, including monitoring, analysis, automation and service management. These platforms enable concurrent use of multiple data sources, data-collection methods, and analytical (real-time and deep) and presentation technologies.

Big Data

Big Data is the name given to huge volumes of data that can be analysed to get useful patterns and make business predictions. The IoT Big Data is found in many applications including predictive models, analysis, smart devices and machine learning.

AI is required to make sense of Big Data that is generated from various sensors. Hence, it becomes a prerequisite for an IoT system to work intelligently for analysing, collecting, manipulating and generating insights, from the enormous amount of data at high speed.

Edge computing

Edge computing is important in machine learning. It makes decisions by applying intelligence based on the deployed machine learning models. For example, IoT devices are deployed for monitoring gas pipelines. Here, costly human inspection is replaced by technology that can rapidly detect leakages or other anomalies, and automatically alert the concerned person(s) for appropriate action. In such cases, distributed computing at the edge can dramatically cut response times.

Edge computing architecture can be visualised using a three-tier architecture (Fig. 2). The first layer has local devices and applications, the second is the edge layer and the third is public cloud. Devices are sensors and actuators that are responsible for collecting data and controlling devices. These are connected to the cloud through the local edge computing layer.

Edge computing architecture (Credit: thenewstack.io)
Fig. 2: Edge computing architecture (Credit: thenewstack.io)

Many companies are investing in the convergence of the IoT and AI. For example, Microsoft is working on its vision for Intelligent Edge. Its Azure IoT Edge platform enables low-power devices to run containers and perform AI locally while retaining a connection to the cloud for management and modelling. Azure IoT Edge is an open source project available on Github.

Amazon Web Services (AWS) launched its edge computing platform, Greengrass, to incorporate machine learning. It is built to deliver capabilities similar to Azure IoT Edge.

Smart homes

Smart devices like phones, computers, home appliances and the like are embedded with sensors, software and actuators. Connectivity and communication between devices are possible using the IoT. Through the combination of AI and the IoT, one can instruct smart assistants like Alexa to control devices.

Another example is a smart thermostat connected to the Internet. Some smart thermostats include intelligence that can learn when the house is likely to be occupied. This allows automatic pre-heating, so that room temperature is comfortable when a resident arrives. If lifestyle of the resident change, the thermostats can gradually adjust the schedule, maintaining energy savings. A combination of AI and the IoT can teach decision-making to machines. These machines can act as smart assistants, like Alexa, to control smart home devices.

Smart security cameras

These are self-contained, standalone vision systems with a built-in image sensor. These utilise the Internet and intelligent programs that analyse images from cameras to recognise humans, vehicles or objects. One can do real-time monitoring remotely from a phone or computer. Latest automated surveillance with new machine learning technique could give CCTV cameras the ability to spot troubling behaviour without human supervision.

Automotive

AI-assisted automotive technology has been employed in driverless cars. Recently, IBM team has developed an IoT for automotive—a program to eliminate driver errors through connectivity. Since many accidents are caused by human errors, researchers are trying ways to minimise human errors through AI algorithms. In the future, a car is likely to have well-packaged complex computer programs. Automated vehicles use AI, sensors and global positioning system coordinates to drive themselves without a human operator.

Healthcare

AI and the IoT find applications in healthcare facilities and maintaining patient health records, medical equipment and assets in hospitals. These technologies are also being used in the pharmaceutical industry.

Agriculture

AI and smart algorithms can analyse massive amounts of data on weather, environment and historical information to make increasingly accurate predictions on what, where, when and how to plant crops for optimal performance and yields. The IoT for Agriculture, or IoT4Ag, is a new area of technology development that holds tremendous potential for improving global food production. There is great potential of AI combined with IoT4Ag.

Manufacturing

Industry 4.0, or the fourth industrial revolution, includes integration of the IoT and AI, and a move towards semi-autonomous decisions. AI-powered robots are used in many manufacturing setups.

The IoT and machine learning in predictive maintenance can be found in machines for manufacturing. As per a report from Forbes, 44 per cent of respondents from automotive and manufacturing sectors classified AI as highly important to the manufacturing function in the next five years, while 49 per cent said it was absolutely critical for success.

Wearables

Modern wearable technologies have enhanced sensing and AI-based cognitive abilities, and can create augmented reality.

Wearable sensors enable soldiers to be tracked easily, especially in high-alert operations to monitor their safety.

Military

AI, Big Data and machine learning are used by the military to safeguard people against rogue entities. These technologies are used to sift through and analyse drone footages and other activities.

AI and machine learning could also be used in cybersecurity to minimise data theft and other malicious activities. AI and the IoT hold the promise for defence, including actionable information and building a complete picture on critical decisions to be made.

Challenges ahead

While AI and the IoT combination has great potential to transform the industry, there are numerous challenges coming to light for implementation and faster adoption, including hardware, lack of standardisation and regulations for ethical issues, and interoperability.

For example, an AI-chatbot, which pops up on websites to interact with customers, can either follow a scripted text or use machine learning. If it does not work as per scripted text and has collected sensitive personal information of the customer, or if a surveillance camera exposed personal photos to the outside world, there is no clear provision for accountability. Also, if a self-driving car got involved in an accident, accountability for damage to property or death of a person is unclear.

The IoT sensors are embedded in different devices, including industrial and commercial. It is difficult to synchronise the data flow between all this hardware without the help of a professional team. It is difficult to calibrate sensors and keep IoT hardware updated on a regular basis. Both software and hardware are equally important in AI and the IoT. Although computing capabilities have advanced significantly in recent years, bottlenecks have shifted to computing hardware such as processors, memories and sensors.

As more and more data is generated and complexity increases, new computing power is required in hardware systems. Newer computer chips and platforms require high-bandwidth memory systems to address the bandwidth needs of AI and related systems. Therefore a number of engineering challenges arise, such as design complexity, high number of inputs and outputs, cost, systems integration and so on.

Latest deep neural network designs have become more accurate, and so their sizes, parameters and number of operations have also increased dramatically. But performance of typical hardware platforms cannot keep pace with new deep neural network designs. Therefore innovations in hardware are required to meet the increasing computational demands in deep neural networks.

The way forward

AI and the IoT are creating unique outcomes and opportunities in Industry 4.0. These technologies are disruptive forces gradually paving their way from health and finance to manufacturing and commerce industry. Use and development of AI and the IoT are not driven by pure academics or research; it is done to address the real needs of the world today.

There is a lot of progress and development being made by various industries to improve these technologies, including memories, processors, sensors, power efficiency devices, machine learning and deep neural networks. However, due to lack of standardisation and regulations, there are certain hurdles that are impacting governments, enterprises, institutions and individuals in multiple ways. To get maximum benefits from these technologies, a set of governance frameworks is needed for a holistic approach.


 

EFY Network

LEAVE A REPLY

Please enter your comment!
Please enter your name here