AI and Vision Team Up to Improve Assembly Line Processes

Industry 4.0 demands technologies that bridge the gap between the IT and the OT systems. A successful solution does more than monitor devices. It synthesizes vast amounts of information and identifies trends and weak points to enable data-driven improvements on the factory floor.

Article Key

Image credit: Drishti Technologies

When a problem hits the assembly line, the operator wants to know what happened as soon as possible. Easier said than done. In some cases, a problem can cause an immediate cease in production, but in other cases the issue isn’t identified until several steps down the line—maybe not even until the finished product goes through quality assurance testing. Isolating the issue and finding the root cause can hold up manufacturing, resulting in lost production and profits.

Artificial intelligence and edge computing are changing production lines for the better. Technologies that monitor both automated and human workers can spot potential problems before they get out of control. Companies such as Drishti Technologies and Falkonry have built manufacturing solutions that use AI-powered analytics to bring greater efficiency to the factory floor.

Drishti technology analyzing production

Image credit: Drishti Technologies

AI for Manual and Machine Activity

Drishti’s video analytics solution combines AI and computer vision to track events on the line. Its “action recognition” technology adds a step to object recognition and digitizes manual activity as well as machine and robotic activity.

The Drishti software can process multiple frames and associate each with a cycle time. That level of detail provides operators with the exact amount of time it takes to complete a specific task, as well as how long it takes to complete each step within that task. It also can pinpoint a bottleneck or note a skipped step.

The software-as-a-service solution uses Drishti cameras with integrated computer vision to stream the video. The company’s proprietary neural networks establish the cycle time data and identify and sequence the actions within those cycles. Operators can view the data on dashboards or reports in real-time. In-depth analysis can identify manufacturing inefficiencies.

Falkonry graphic: Patterns are predicted and explained

Image credit: Falkonry

A Pattern of Discovery

Falkonry’s Time Series AI Suite monitors all assets and associated processes and provides resolutions to potential problems. The suite of applications includes Insight, Clue, and Workbench. Here’s how they function:

  • Insight provides automated and continuous anomaly detection;
  • Clue automates corrective operating actions based on patterns found during data collection; 
  • Workbench reviews causes and sequences of behavior on the line.

The company uses a unique pattern discovery system to identify patterns in operational data that enable problem detection and prediction capabilities. The software also provides the operator with specific data about each pattern. To identify trouble spots, the Time Series software is able to determine which signals are most critical to making the prediction and weighs the relevance of each signal in reaching that prediction.

On a manufacturing line, Time Series AI can correlate patterns to product flaws. When that happens, the software will automatically adjust operations to remedy the situation. It also suggests when and where operators should ramp-up quality assurance testing processes to improve yield. In addition, it can identify and alert operators to problems that arise in known troublesome areas.

The Industrial Internet of Things (IIoT) generates an unimaginable amount of data. Solutions that analyze that data effectively allow manufacturers to improve operations, resulting in better products, safer factories, and increased profits.

Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.