A sensor can recognize that something is going on. But what if the system could recognize that its condition is abnormal, and determine what should be done as a result.
This is where the convergence of IoT and AI can be interesting.
IoT involves connecting and sensing the real world, and then using AI to make decisions.
A simplified rendition of this process would look like this:
Devices → Connectivity → Data → Processing → AI → Decision → Action
- IoT Starts With the Physical World
An IoT system involves sensors, machinery, trackers, cameras, gateways and other devices.
Depending on the application, these devices could produce data that informs the operator on:
Temperature
Location
Environment
Conditions
Tracking
Telemetry
The key point at this layer is to get actionable data.
- Connectivity Gets It to the Next Layer
When information is produced, it needs to get to the appropriate processing/storage environment.
Depending on the architecture, this could involve network infrastructure, gateways, edge devices, or the cloud.
It's worth noting that architectural decisions at this point can impact bandwidth, latency, reliability, connection stability, power and more, depending on the hardware involved.
- Processing the Output Is Necessary
Unless the output is something as simple as True/False, it still has to be put into some form of organized data before anything else can happen.
This is where the data gets collected, processed, refined and prepared for analysis. The importance of this layer cannot be understated, as it directly informs what insights the AI can produce.
- AI and ML Can Inform Decisions
When information is available in a proper format, it can be used by AI and machine learning algorithms to find relevant insights, detect anomalies or patterns, or help with operations.
As an example, let's imagine an IoT system that uses AI to keep track of machinery and detect any issues. This involves using the data from the sensors to find patterns that deviate from the normal operation of the equipment.
- From Digital Insights to Real-World Actions
The last layer involves using the information, likely with human supervision.
Depending on the application, an AI layer could be used to flag abnormal data or provide recommendations.
In this example the AI layer makes suggestions, but the decision is still up to the human operator. As another example, an AI layer could make decisions or run algorithms that impact the real world directly, such as adjustments to machinery.
The sensor and AI layers work in concert to gather insights about the physical world and help operators make informed decisions.
Why AIoT Is Interesting and Important
AIoT is more than simply a combination of IoT and AI. The real value comes from the engineering around the system.
A lot of thought has to be put into how the information is gathered, processed and refined, how it gets to the AI layer, what the AI layer actually does with the information, and how the output of the AI gets used to make decisions in the physical world.
It's a complex but rewarding space to work in, and it can have a major impact on industrial applications, manufacturing, operations and more.
Developer Considerations
Developers building AIoT applications would be well served to consider a number of factors beyond simply picking the right AI model and sensors. These could include:
The quality and reliability of the data.
Latency and processing requirements.
The architecture's ability to scale.
Security and system protection.
The value of the output and how it informs operations.
The need for human supervision and manual intervention.
System reliability and potential points of failure.
These are just a few of the many considerations that have to be made when designing an AIoT system.
Final Thought
IoT and AI are two of the most talked-about technologies of recent years. But the real engineering starts when you combine the two.
At their best, AI and IoT systems act as a unit.
The information from the physical world is used to train and inform AI models, while the AI layer can add intelligence, insight or even decision-making to the IoT system. When the two are combined, they create opportunities for new data to be recognized as valuable intelligence for operators and end users.
Top comments (0)