How AIoT Will Help Us Manage The World Better

As we witness the explosive growth of the Internet of Things (IoT), most people are probably thinking, “Great, we will get so much data!” But as massive datasets pile up across industries, the conversation likely shifts to, “Great, how do we make sense of this much data?”

The IoT is an emerging technology that connects virtual and physical objects—anything from cell phones, tablets, wearable devices, digital assistants, smart appliances and more. Across the globe, billions of devices can communicate with each other and share information, and real-time data can be collected from these devices simultaneously and continuously. By 2025, there will be 41.6 billion IoT-connected devices globally, producing 79.4 zettabytes of data annually, according to IDC. With the IoT market projected to be worth $2.7 trillion by 2024, this explosive growth is set to continue. 

However, connecting all the devices is only the first step. The devices must be organized and managed, and the data harnessed for actionable business decisions. How do we make sense of the massive datasets, or Big Data, so the IoT realizes its full potential?  

Artificial intelligence (AI) refers to the intelligence machines use to solve complex problems. First entering public awareness with IBM's Deep Blue, which beat chess grandmaster Gary Kasparov, and Watson, which crushed everybody on Jeopardy!, AI use has become more prevalent. That is especially so with the introduction of virtual assistants, such as Siri and Alexa, and facial recognition software. 

The Artificial Intelligence of Things (AIoT) combines AI and IoT infrastructure to improve the IoT. AI’s tremendous data analysis capability can help extract meaningful information and insights from the data IoT devices collect. AI can also help IoT devices interact with humans and other objects and make autonomous decisions. IoT devices are like the terminals of a nervous system, and AI is the brain that controls the nervous system, processes the information from the terminals and makes decisions. 

AI helps identify abnormal patterns in massive datasets, sending alerts when things deviate from the observed norms. Therefore, it can identify anomalous patterns earlier, faster and with greater accuracy. Businesses can employ AI to identify and mitigate potential risks, helping enterprises avoid costly and unplanned downtime, increase operational efficiency, and improve products and services.

Applications of AIoT 

The four major AIoT application types are wearables, such as smart watches, augmented/virtual reality (AR/VR) and wireless earbuds; applications for the smart home, such as smart appliances and home security; smart grids, streetlights, public transportation and other facets of the smart city; and smart industry applications, such as autonomous manufacturing robots, automated supply chain management and predictive maintenance. IoT devices contain sensors that can collect many different kinds of data, such as temperature, pressure, humidity, air quality, vibration and sound. The almost endless combination of the different segments and many measurement parameters in AIoT will enable numerous applications.

Smart Healthcare 

People are living longer. The proactive management of chronic diseases, such as Type 2 diabetes and hypertension, as well as minimizing costly hospitalizations will be critical. Wearable devices, which have worked well for sports and fitness, can be developed into medical-grade instruments to monitor patients remotely. For example, wearable glucose monitors that continuously collect patient data for analysis enable doctors to monitor patients’ vital signs and blood sugar levels remotely and in real-time. AI-assisted diagnostics will also lessen doctors' workloads and shorten the time it takes to obtain a second opinion. Lastly, AR/VR and cross reality technologies can be used to train medical students on virtual cadavers and help patients rehabilitate. 

The IoT sensors installed around a building can monitor personnel movement and adjust temperatures and lighting to maximize energy efficiency. AIoT can also control building access through facial recognition technology. The combination of connected cameras and AI can analyze faces in real-time against a database to determine who should be granted access to a building. Any unidentified personnel or abnormal activities are recorded, and alerts are sent to the central hub for decision-making. Even a building’s restroom convenience and cleanliness can be improved with AIoT. For example, sensors installed in a bathroom can monitor stall, faucet, soap and paper towel usage. Staff can be alerted when supplies are running low or a particular area requires emergency cleaning. 

AIoT will accelerate the implementation of task-based robots in manufacturing and enable a broader utilization of digital twins in product design, development and production. In addition, AIoT can help a business monitor its fleet of trucks, ships or oil tankers and identify unsafe driver behavior or vehicles needing maintenance. Streamlining pickup and delivery routes to reduce fuel costs is yet another task AIoT can take on. Lastly, AIoT can help maintain a cold-chain for critical supplies, such as extremely temperature-sensitive COVID-19 vaccines, during transport. 

AIoT can provide value to the retail sector on multiple fronts. Using AIoT, stores can more closely monitor supply and demand and restock accordingly with minimal human involvement, generating savings. 

By pairing a camera system that tracks shoppers with AI analytics, a store can gather information on customers' gender and their product preferences to tailor product recommendations. Smart cameras can help streamline the checkout process and minimize the wait time. AIoT-controlled autonomous robots can be used to deliver goods. 

In a smart city, AIoT can help monitor traffic with drones that fly over the city and transmit traffic data. Then, AI can analyze the collected data and reduce congestion by adjusting the speed limits and timing of traffic lights. This system can also detect and respond to accidents quickly and efficiently. Moreover, cities with flooding problems can install sensors in the storm drains and collect data for analysis to declutter the drains and prevent floods. In addition, smart grids can manage energy imbalance in the grid proactively to prevent the spread of blackouts.  Of course, the AIoT applications in a smart city garnering the most attention are autonomous vehicles (AVs). Many sensors can be installed on electric vehicles for fleet management to optimize usage, minimize accidents and increase fuel efficiency.

Climate change is perhaps the most significant challenge humankind faces. In response, AIoT can be used to make agriculture and renewable energy smarter. Farmers can use drones to survey their crops and identify drought or overwatering. Then, AIoT can help adjust the water systems accordingly. AIoT can also be used to monitor the weather and identify plant disease outbreaks. Lastly, automated harvesting can help farmers examine the yield of different crops in different fields. 

On the other hand, AIoT can be used to manage renewable energy facilities. For example, by monitoring the wind speed, temperature and humidity at wind turbines or solar farms, a business can mitigate the risk of catastrophic damage or personnel injury. 

Criteria for Successful AIoT Implementation 

In addition to the Big Data collected by the IoT devices and AI, 5G is crucial to the success of AIoT. A functional 5G network is needed to transfer the massive amounts of data collected from the IoT devices. For example, during a remote surgery, videos need to be streamed in real-time to ensure the patient's safety and the success of the operation. 

Data standardization, however unglamorous, is also essential. AI algorithms are only as good as the data obtained. A lack of data standards will lead to garbage-in-garbage-out scenarios. 

In addition, there is a dimension of human ethics. Software is only as good as its programmers. There have been reports of sexist and racist bias in algorithms, such as an image recognition software that is less accurate in identifying non-Caucasian faces. Such biases may even be life-threatening when the software fails to diagnose patients from a specific population accurately. Data scientists need to take care to leave their prejudices at the door when they come to work.

Moreover, the massive datasets, which are equally attractive to data scientists and hackers, need to be safeguarded. International standardization of cybersecurity will be a constructive step forward. Furthermore, the cost of hardware, mainly high-performance CPU chips, will be a significant bottleneck to the development of AIoT. Lastly, balancing the energy consumption and analytics performance at the level of IoT devices must be addressed. Edge computing is a potential solution.

Intelligent things will get smarter. Someday, AI will be used to help humans manage all this AIoT.