Machine Vision: Vision Systems in Manufacturing

Machine vision refers to the ability of a machine to capture and process images. It is essentially a system of hardware and software that enables a machine to “see” and extract information from its environment.

Machine vision, also known as computer vision, has been around since the 1950’s, but the later development of machine learning has allowed machine vision capabilities to expand past repetitive tasks to more complex applications in manufacturing such as quality assurance, compliance, and inventory management.

Within an Industry 4.0 ecosystem, machine vision systems work continuously and in conjunction with operators, leading to improved efficiency, fewer errors, and better data.

The main components of a vision system are:

• Cameras

• Image Processing Software

• Computing Platform

To understand how these components work together, let’s take a look at an example of how machine vision can be used for product inspection.

Once an object is in view of the camera, software or an operator can capture an image of the product that needs inspection. The digital input is analyzed by the image processing software with a model that has been trained with inspection criteria. If a defect is detected, the product is rejected and an operator is notified to take action. Models can be trained with manually collected images, and continuously improved with machine learning as more products are inspected over time.

Automating Workflows and Tasks

Detect a change (new activity in a region, color change, jig code visibility, etc.), record that event information, and then trigger new steps.

Augmenting Operators with Feedback

Notify operators when they use the wrong tool, pick from the wrong bin, or incorrectly align parts.

Production Tracking and Visibility

Through jig stickers, text reading, and more, effectively and reliably collect and record materials and task completions as products move through the shop floor.

Quality and Documentation

Document snapshots as products move through production and visualize trend data.

Tulip’s approach to vision systems is different for two main reasons:

1. Tulip sets up frontline workers for success with guidance and effortless data collection

2. Tulip allows those closest to operations to set up and modify the system with drag-and-drop configuration and cost-effective off the shelf cameras

Tulip’s no-code vision apps can be integrated with external, cloud-based APIs such as Amazon Lookout for Vision to easily create and continuously improve machine learning models.

Learn more about Tulip’s vision capabilities: