IoT ecosystems offer a compelling proposition: to answer the question of what is happening.

From IoT to Industrial AI: Use Cases of Artificial Intelligence

IoT ecosystems offer a compelling proposition: to answer the question of what is happening.

A sensor can tell us the temperature. A tracker can share its location. A machine can report its vibrations. An IoT-enabled production system can signal the beginning and the end of an operation.

But all of these data points represent only an intermediate stage of analysis.

The more interesting question is: what can we know based on this information?

Artificial intelligence and IoT together can generate a substantial amount of value by answering this more challenging question.

IoT Generates Information About Physical Objects

An IoT infrastructure normally consists of hardware and software systems that acquire information from connected devices, process this information, and store it in a database or data warehouse for subsequent use and analysis.

A simple IoT architecture is represented below:

Physical Asset –> Sensor –> Connectivity –> Platform –> Application

Depending on the use case, the information captured by IoT devices can be significantly different and may include:

Temperature;

Humidity;

Pressure;

Vibrations;

Location;

Movements;

Status of machines or other physical objects;

Events occurring in connected systems;

The data captured by IoT devices often provides valuable insights into the working conditions of connected equipment or processes. However, it may rarely answer the question of what these information indicates.

AI Offers an Additional Layer of Analysis

AI and machine learning algorithms can consume the information generated by IoT systems and discover patterns, relationships, irregularities, or other phenomena of interest.

For example, consider a machine producing vibration data as shown in the figure below:

A simple IoT analytics module can display this information or raise an alert when the measured value crosses a predefined threshold. An AI-based analytics module would analyze the recorded data patterns to identify interesting trends or behaviors.

Note that IoT devices can normally answer the question of what is happening. But they may rarely provide an immediate response to the more interesting question of what this means.

The type of analyzes carried out depends on the data available, the business requirements, and the selected AI models. In general, however, these algorithms can significantly benefit from the additional layer of information provided by IoT devices. It is worth noting, however, that AI is not always a viable option for IoT data.

AIoT Enables a Feedback Loop

AIoT systems, when properly implemented, create a feedback loop in which new information and valuable insights can lead to further improvements in operational performance. The architecture of such a system can be represented as follows:

Connect –> Collect –> Analyze –> Understand –> Act

IoT infrastructure normally dominates the first two stages of this loop and provides the information necessary for subsequent processing, analysis, and decision-making algorithms. Advanced analytics and AI technologies consume this information and discover patterns of interest. However, the most important element of the loop is the feedback and specific actions taken to adjust or optimize operations.

For example, an AIoT system may include connected devices that monitor the equipment under control. These devices provide information on the current state of the equipment, which is processed by subsequent analytics modules. If abnormal patterns are detected, this information may be used to notify an engineer, who will take specific measures to eliminate the detected effects. In other words, AIoT often operates within a closed loop that includes not only information processing and analysis but also specific actions following from these processes.

Practical Applications of AIoT Data Analytics

The described concepts can be applied to different types of connected systems. In particular, the same general architecture can be used to analyze the data generated by:

Manufacturing equipment;

Transport infrastructure;

Inventory systems;

Environmental sensors;

Each arrangement will normally have its own requirements and considerations, but the general idea remains the same: collect data from connected systems, process and analyze it, and use the results of this analysis to optimize operations.

Data Issues Undermine AIoT Success

One of the main challenges that affect the practical implementation of AIoT systems is the quality of data. A robust analytics model is of little use if the underlying data is incomplete or incorrect. Thus, when designing such systems, developers and analysts must pay special attention to the quality of information provided by connected devices. Particular attention should be paid to the following questions:

What data sources are used?

How often is this data collected?

How accurate are the measurements?

Where is this data processed and stored?

How can historical data be used?

Are there mechanisms for validating unexpected results?

How are network issues handled?

These and many other practical considerations are of great importance for the successful implementation of AIoT systems.

Beyond IoT Data Analytics: Capturing Value from Physical Objects

The described approach is only one of many ways in which AI can be used in combination with IoT technologies. The general idea is that connected systems generate valuable information about their operating conditions that can be used to optimize operations. This information can be collected and processed in different ways, but the core principles remain the same. When considering how to use these technologies, organizations may also want to consult Aperture Venture Studio , which offers extensive experience in developing practical applications of artificial intelligence for the physical world.

Finally, note that AIoT systems are normally designed to answer the question of what to do with the information collected from connected devices.