Swarm Intelligence of Cameras
Your front door video camera can automatically determine whether the object it has spotted is familiar (cat, neighbor, grandma) or new and unfamiliar. Thanks to AI, the camera continuously learns when to raise the alarm and when not to.
The camera, which is connected to the internet, makes this determination independently without having to access a centralized intelligence and, if required, takes adequate action by sending an alarm signal. The actual learning process takes place between all cameras in the network. The camera makes its own experiences available to the others, which accelerates and decentralizes the learning process.
This is an early example of how AIoT will influence our lives: There will be an increasing number of intelligent and interconnected devices.
What is AIoT?
The Internet of Things (IoT) is seeing a meteoric rise. Even today, there are more connected things than humans, and some predict that by 2025, there will be more than 55 billion IoT devices . IoT devices provide information on their environment and are controlled and monitored remotely. This generates massive data volumes and flows that need to be analyzed and processed so that meaningful actions can result.
Until now, IoT devices were hardly ever intelligent or capable of learning. The sensor data was forwarded and then analyzed and processed by a central unit – with or without AI.
AI allows machines to learn from experience. They can better adapt to new situations and solve some problems autonomously by simulating human thought processes. Bringing together AI and IoT opens up a host of new opportunities.
Combination “at the Edge”
The combination of AI and IoT turns passive sensors into learning machines. Both support one another. AI becomes more valuable and useful through IoT through real-time connectivity, signals and data exchange (“better data”). Meanwhile, IoT becomes more useful through machine learning (ML) by, for example, allowing for more intelligent decisions (“better decisions”).
Current studies see enormous potential in AIoT. The global market for AI in IoT devices is expected to grow to more than USD 26 bn by 2023 .
Up to now, devices would send their signals to central instances, such as “data lakes”, for analysis. There, the information was analyzed with the help of big data mechanisms and AI. However, this is not always possible or even useful, and often, there are latencies involved. If the system is mobile or in a remote location, if the data volume for analysis is particularly large or analysis needs to be fast, it helps to process data as close to the data source as possible (“edge computing”). This leads to faster results and a lower demand for bandwidth. In future, many of the AIoT applications will be used “on the edge”, sometimes without an additional centralized AI instance.
Smart, Smarter …
AIoT will quickly gain a foothold in areas such as industry 4.0, smart cities or home automation. But other industries will also profit from using AIoT:
Smart Retail: Customer behavior is analyzed in-store with various technologies. This data is then amalgamated with existing data and the customer profile to generate customized offerings in real time.
Smart Farming: With the help of local AI, robots identify weeds to be picked in a field. They improve their identification accuracy by exchanging information with other robots. The location data helps improve the efficiency of weeding, for instance, by marking areas that have already been weeded. But there is also an opportunity to systematically identify areas with high weed prevalence and to use pattern recognition to identify critical areas in advance.
An Exciting Future
It will be exciting to see the new applications that result from the convergence of AI and IoT over the next few years. Many of them will be innovative and disruptive and will have a lasting effect on the way we live, work, and play.
 IoT Report: How Internet of Things Technology Is Now Reaching Mainstream Companies and Consumers (Peter Newman, Business Insider 27 July 2018).