DEV Community

Hopeseeker
Hopeseeker

Posted on

How AIoT Is Transforming Automotive Manufacturing: The Role of Sensors to Smart Operations

Modern auto manufacturing facilities are distributed data centers.

Within a single factory, there may be PLCs, machines, robots, RFID readers, UWB anchors, BLE devices, cameras, MES, ERP systems, warehouse software and more — all generating data, but gathering this data is only part of the challenge

The more interesting task is to derive value from these disparate signals.

This is where AIoT, the combination of Artificial Intelligence and the Internet of Things, can be valuable to the automotive manufacturing industry.

An Overview of AIoT in the Automotive Manufacturing Industry

At the most basic level, an AIoT architecture can be thought of as connecting physical assets and sensors to a software platform capable of deriving value from the data.

A simple representation of this is as follows:

Physical Assets -> Sensors -> Connectivity -> Data -> AI/ML -> Insights -> Action

Where a system might be drawing data from:

Machine state information from PLCs

Asset identification from RFID

Position information from Ultra Wideband, BLE

Vehicle production information from MES

Inventory information from ERP, warehouse systems

Environmental sensors, etc.

However, the challenge is less in the sources of data, and more in creating a reliable mechanism to unify the relevant information with the necessary context.

Understanding the Importance of Context

Taking the example of a vehicle's location, this information can be useful, but adding additional information adds significant context and value - which production stage is the vehicle in? What is the VIN of the vehicle? Has the required material arrived for the vehicle? Was there a recent production event? Is the vehicle waiting due to a process issue?

Likewise, this is one reason why AIoT solutions are so powerful — data can be related to additional information for insights and actionable events.

A similar analysis can be performed on equipment, tooling, materials, and mobile assets.

Bridging the Divide Between OT and IT

Modern manufacturing facilities are home to a variety of operational technology and information technology systems.

Operational technology systems can be PLCs, SCADA systems, machines, controllers, and robotics, while IT systems can include MES, ERP systems, warehouse software, databases, and analytics applications. One way to begin thinking about AIoT solutions is to bridge this divide between OT and IT systems, but doing so involves additional considerations.

Some examples of technologies and standards for connecting these systems can include OPC UA, MQTT, Ethernet-based industrial networks, APIs, Edge gateways, etc.

Additionally, an edge layer may be needed if processing data closer to the source is desired, rather than sending all data to a central cloud or analytics system.

Leveraging AI and Machine Learning

With relevant data being collected, AI and machine learning algorithms can be applied to this data to either detect anomalies or as the basis for higher-level insights and action. For example, machine learning can analyze historical telemetry data from a machine to detect unusual patterns that could indicate a fault. Likewise, production data can be mined to detect patterns that distinguish between different types of production events.

Another benefit of leveraging AIoT is the ability to correlate disparate data sets.

A data point in isolation may not indicate an issue, but in conjunction with other data points, it can be possible to reason about a situation and determine the best course of action.

It must be stressed, however, that not all problems require a machine learning solution. Rules-based approaches, dashboards, alerts, and deterministic reasoning may still be appropriate for a given situation. The question is one of whether an AI solution provides meaningful benefits over such an approach.

The Role of AIoT in Automotive Manufacturing

AIoT systems can be applied to a variety of automotive manufacturing use cases, including:

Vehicle traceability and production visibility — tying together VINs, production events, and location data

Asset tracking using RFID, BLE, UWB, or other location technologies

Production analytics and insights using manufacturing data

EV battery traceability, including across production steps

Intralogistics, including equipment and forklift tracking

Equipment diagnostics and maintenance prediction

A more detailed overview of some of these automotive manufacturing use cases, along with examples of real-world AIoT solutions, can be found in OEMNex AI's manufacturing technology overview.

The role of AIoT technologies can be much broader — in fact, many industrial connectivity, AIoT, RTLS, and manufacturing technology companies will offer a wide range of solutions to address automotive manufacturing needs. Some examples can be seen here on OEMNex AI's industrial technology marketplace.

Beginning the Process: From Data Challenges to AIoT Opportunities

There is one critical pitfall when beginning an industrial AIoT project — starting with the wrong question. In particular, it is tempting to begin with a technology question: how can we add more sensors to our system? How can we implement an AIoT solution?

While these may be important questions, it is critical to instead begin with an operational question: what decision would benefit from having more and better data?

By thinking through what decisions need to be made, the type of data that would be relevant to these decisions can be identified, which in turn allows the creation of a solution tailored to a particular need. This is likely to be much more effective and scalable than attempting to build a generic solution that attempts to account for every possible scenario.

Some examples of this process can include:

Operational Need Decision

Improving production scheduling and resource allocation More accurate information about production events, equipment status, and material location needed

Better maintenance scheduling Predictive maintenance insights require data on equipment state and usage

Faster root-cause analysis of production issues Correlation of production, material, equipment, and vehicle data needed

Ensuring timely delivery of materials to assembly lines Ensuring that materials are on-time, available, accounted for, in the correct location

Predicting equipment failures Ability to identify equipment conditions that could lead to failure

These are just some examples of how to formulate an operational need and desired outcome.

As mentioned earlier, an AIoT architecture is not centered around trying to deploy AI everywhere possible — rather, it is about identifying a valuable insight to be had from the connection between physical and digital worlds, and building a system around it.

Top comments (0)