Artificial Intelligence in IoT

Artificial Intelligence in IoT. Potential Use cases

Body Trackers

Complexity: 2 (most tasks already have general solutions due to high demand in the area. However, the challenge is in the individuality of each case).

Data needed: Generally, the data on human activity is gathered over some period. Usually, it comes from smartphone accelerometers, fitness trackers, or other body sensors. The amount and variability of data should be significant enough to build a generalized model for everyone.

Influence on business: In areas that mainly rely on employees’ work, either in sports or effort-consuming areas, being up to date with their health is an impressive advantage. Understanding every person’s needs will help choose an optimal schedule or working parameters, making the work more comfortable and practical.

Benefits:

• It will be possible to choose individual approaches for every person depending on their most effective schedules or lifestyles.
• Often, health patterns can go unnoticed both from an outside perspective and by a person themselves. However, tracking their body parameters would allow smart devices to pick up patterns that would suggest some changes in health. Such changes may be detected with several days of constant tracking, making one-time checks at a doctor’s office impossible. Early detection can be beneficial for most disease diagnosis or determining the daily schedule.
• Tracking body parameters during training or exercising enables picking up personalized details of what activity works best. Applications can range from optimal fitness schedules to highly effective athlete training.

Price: AI-powered body tracking is one of the cost-effective IoT applications. This is usually because only one or several sensors are needed, such as gyroscopes or PPG sensors; their data is used to calculate other parameters. These sensors are also the most widely known technology, making them cheaper.

Examples: Often, people and companies use body trackers as training assistants. Cases vary from everyday wear to usage in professional sport. Some companies are researching implementing them into non-athletic areas. Possible goals include optimizing the working conditions or monitoring the employees’ comfort and wellness, both in-office and work-from-home modes.

Smart Homes

Complexity: 3 (generally relies on recognizing patterns or predicting situations. Security applications are usually the most sophisticated).

Data needed: There are many types of devices for smart homes, and each of them requires historical data about their functions.

For example, the thermostat will need some information about optimal and non-optimal temperature and air conditions. Usually, it comes from sensors installed at home. At the start, it can use data collected from similar situations, adjusting them with implicit instructions. Then, devices record how the owner prefers them to function and learn how to be optimal on the fly.

Influence on business: This field is application-based. But if the company is connected with home technology, installing AI-powered devices can help personalize the services and reduce the number of false alarms in security applications. 

Benefits:

• Some traditional home solutions are built for the “average person,” not considering personal preferences and patterns. This generalization usually satisfies no one, so smart devices allow us to better adjust things to our preferences.
• Home is the area where people want to customize their surroundings the most. As some may like it warmer in the evening, but cooler the following day, smart devices can make this switch during the night.
• In a situation where there are many devices across the area, it becomes annoying to learn how to control them individually. Reliable voice-control systems or home automation apps like Apple HomeKit allow people to interact with the system intuitively, without diving into technical details.
• Security standards improve, as it is possible to set commands to respond only to a particular voice or feature. This fact is also used in security itself, resulting in such applications as smart locks.
• Usage of Artificial Intelligence in IoT also improves surveillance and alerting of potential threats. AI-improved smoke or carbon monoxide detectors can alert the owner of the existing danger faster and in more convenient ways.

Price: Just like for self-driving cars, the cost is one of the slowing factors in the spread of AI in smart homes. Smart home technology itself still has an impressive price tag, and adding AI raises it only slightly. However, it is extremely promising and widely developing right now, so the costs steadily fall yearly.

Examples: Some applications of AI in smart homes are already in the market. The most prominent examples are digital assistants: Google Nest and others. Even robotic vacuum cleaners are one of the oldest such cases. Smart thermostats, charging stations, voice-controlled devices, and smoke detectors are also present in homes worldwide.

Self-Driving Cars Technology

Complexity: 4 (most examples right now are automated to some defined extent, as there are levels of car autonomy. Complexity depends on how much you want to automate. The less human interaction required, the more complex it gets).

Data needed: Different kinds of data on the car’s surroundings in real-time: other vehicles, road elements, pedestrians, road signs, etc. Cameras, LiDARs, and sensors usually gather such data.

Influence on business: This is highly prospective for public taxi services, deliveries, and other services where reduced driving work means saved costs. Autonomous vehicles can potentially change the world around us by eliminating some industries and giving birth to others.

Benefits:

• Road situation recognition steps are straightforward to implement from the technical side. However, considerable attention to reliability is a must.
• Probably one of the most developed IoT fields so far. There are many semi-ready solutions or at least hints on how to implement many tasks.
• This is a well-known and popular use case. Companies that use self-driving technology are prominent members of the world’s innovation leaderboard.
• It can vastly reduce the driver’s fatigue, minimizing the number of potential accidents. In cars with the highest autonomy levels, exhaustion and concentration matter even less as vehicles do not require much human intervention.

Price: The development cost itself is the same as for an average AI implementation, but the hardware is much more expensive here. Also, the exploitation proves to be energy-consuming, as many real-time computations and wireless connections are required. Self-driving cars need different kinds of reliable sensors and cameras. The price is one of the main bottlenecks of the fast spread of it right now.

Examples: Obviously, Tesla’s cars with auto-pilot features do not need an introduction. However, even Tesla’s Full Self-Driving Capability feature is not on the highest level of automation possible, as it still requires the driver’s constant attention. But almost every big automotive company is investing in this kind of technology, from Audi to GM. Self-driving truck development is developing quickly as well. Some companies are already close to the goal of full self-driving.

Retail Analytics

Complexity: 4 (technologies for the area are rapidly developing, and the main challenge is optimal interaction with customers).

Data needed: Data on each item in the store and clients’ purchase history are the basis of the analytics. Tracking the customer movement around the store space with cameras is popular as well.

Influence on business:  Predicting the customers’ needs becomes possible with each piece of data gathered.  Embedded sensors and cameras can adjust the way the shop operates and where the consultants are needed. Such a modern and individual approach can help the company stand out and make the customer stay longer. Everyone uses predictive systems, but IoT helps with better precision and delivers suggestions in the most optimal time.

Benefits:

• Keeping track of all the products in the store helps to identify their demand patterns. Also, AI can identify the external factor influence, such as the store configuration or time of the day.
• Tracking the customers’ movement and all their actions in the store helps understand the choosing process. For example, finding out that many people choose between particular products can help the store find suitable locations.
• Artificial Intelligence for retail IoT is often similar to online recommendation systems . A store can interact with customers on an entirely new level, and closer understand their needs. Traditionally Internet-based solutions find their way into the physical store.

Price: Cost depends on the type of information to gather. Often, keeping up with processes requires cameras, which can be the most demanding hardware here. Also, most of the solutions are generalizable, slightly increasing their availability.

Examples: The most known examples include AI shop assistants and even fully automated stores, like the ones set up by Amazon. Other implementations include AI-enhanced choosing techniques, sometimes using Virtual Reality to enrich the shopping experience for customers.

Security Devices

Complexity: 4 (the complexity lies in the needed reliability level. Unlike other uses, AI here needs to be completely fault-proof).

Data needed: Depends on what security measure the solution implements. Usually, the first thing to have are examples of different access situations. For example, for face recognition, this will be facial data. For behavior analysis – typical behavior patterns. If you want to build an identification system, gathering enough relevant data on the person in question is essential.

Influence on business: AI in security relies on detecting patterns that traditional systems cannot pick up. Therefore, falsifying the access parameters will be much more difficult. This technology enables better safety in a location or at a workplace without implementing some extreme security measures.

Benefits:

• Security is often in access control and ensuring the safety of every action and person. Smart built-in sensors in higher-risk areas can alert about dangerous situations so that people can leave dangerous places in time.
• Analyzing behavior in a picture. AI can detect some abnormal situations and react to them or issue a warning to a supervisor. This means AI can easily prevent or deal with threats or human mistakes.
• Access systems using AI can work in subtle ways to identify a person and estimate several parameters at once. It allows the installation of such systems without huge infrastructure spendings.
• Many human mistakes happen when attention naturally decreases at some point. Even the most dedicated guard cannot keep an eye on 100 cameras all the time, while for AI, this will not be a challenge.
• Cyberattacks become more and more sophisticated every day, and many of them already use AI. Often the best way to repel such attacks is to use AI to detect them as well.
• Another usage of AI in IoT security is threat assessment. Such a system will identify all possible threats, assess their priority, and instantly notify humans about the most important ones.

Price: Many developed security systems offer their AI services to companies. Often, such solutions are optimal, as a lot of development effort is needed to reach the required level of reliability. They usually cost only a little more than traditional security systems, except the equipment that makes the AI work, like cameras or sensors. 

Examples: The most well-known implementations of AI in IoT security devices include smart locks or access systems, often with built-in facial or behavior recognition. The market of AI in Cybersecurity is also closely connected to IoT. Examples of such usage include fraud detection or network security.

Smart Cities and Traffic Management

Complexity: 5 (an extensive area of usage. Complexity is in various factors connected to the city management. Largely depends on the degree of automation with Artificial Intelligence).

Data needed: For each specific use, data about its historical usage patterns is needed. For example, you will need records of where and when people typically park and their points of interest for smart parking slots allocation.

This kind of data for every use case is usually gathered by sensors installed around the city. Different types of sensors are connected, forming a complete model of what is happening in the city. There is a lot of open-source smart city data gathered around the world.

Influence on business: Using AI in Smart City management significantly decreases infrastructure and maintenance costs. AI helps to find the most effective solutions with a minimal amount of resources. Some of the developments in the area are also usable on a smaller scale.

Benefits:

• Making the environment smart using AI can substantially impact citizens’ lifestyles. It instantly improves the quality of life, increases safety, and decreases crime rates in the neighborhoods. Apart from security improvements itself, waste processing and commute shortening also work for this goal.
• Transportation is one of the main areas for AI implementation in smart cities. AI helps in finding the most needed commuting routes, either optimizing the traffic infrastructure or suggesting new ones. One of such examples is the rise of the Mobility-as-a-service idea, where transportation is built around the personal needs of everyone.
• AI can determine the optimal use of city resources. For example, load on power grids often has distinctive patterns, so it can be optimized.  It is possible to set city resource grids to ensure optimal use by finding and analyzing these dependencies.

Price: The development of AI-powered decisions is possible in some parts of the city, making transformation more gradual. Generally, smart city development is an area that requires a lot of investment. However, it allows to improve so many faults of city design that the benefits of a healthy environment usually outweigh the costs. One of the most effective areas is traffic management. Saving the unnecessary time spent on commuting every day increases the city’s productivity and boosts business.

Examples: Examples of smart cities that already use AI and successfully function are numerous. The most famous ones include the following locations around the world:

• Amsterdam: one of most significant city-wide waste management projects implemented so far;
• India’s Smart City Mission program.
• The smart city of Neom in Saudi Arabia.
• Singapore, Barcelona, Dubai, and other cities already implement AI in urban services to an extent.