Robot with camera system — A solution in 6 steps

A robot needs a camera system! Or does it? This text will tell you when Robot Vision makes sense and how to choose the right camera.

1) When is a camera system necessary?

Some misconceptions are persistent – such as the belief that a camera is necessary, especially for gripping in pick-and-place applications. It's not true! A camera always means additional complexity and costs. A vision system is necessary when your parts are not oriented– that is, when they have no exact, predictable position for the robot. This can often be solved with vibratory bowls or trays. However, if there is no mechanical solution to orient your parts, a camera-based robot application is right for you.

2) What functions are relevant to the application?

Consider: What is the camera to detect, and how? Only when you know your exact requirements you can choose the right device from the large portfolio of industrial image processing manufacturers. Are you trying to

  • recognise contours or surface conditions?

  • scan bar codes?

  • measure objects?

The process in which the camera is embedded must also be analysed: Is the recording taken from the conveyor belt or by bin-picking, for example? This difference is reflected in the scope of services and therefore in the price of the vision system.

Overview of the most important camera functions from various manufacturers.

3) Camera compatibility with the overall system

The integration effort is often what makes a camera system in a robot expensive. So check the following items before purchase to avoid unpleasant surprises:

  • Camera integration into the robot control system: Are all necessary interfaces and communication links available?

  • Installation and durability: Does the camera come into contact with dust, dirt, moisture, and/or mechanical influences such as vibration or impacts?

  • Cycle and detection times: How quickly does a process repeat itself, and how quickly does the camera have to detect an object?

  • Flexibility requirements: Do different objects need to be detected? And as already described under Point 2, what characteristic must be recorded?

4) What are the lighting conditions?

Changing lighting conditions lead to costly robot camera re-calibration or to incorrect measurement results. So pay attention to the following issues:

  • Are there influences from daylight?

  • Are there any light sources that need to be shielded?

  • Are there reflections?

  • Is lighting or backlighting necessary?

5) Design and setup

Attach the camera firmly and at a safe distance from moving parts. Any shift in the vision system results in costly re-calibration and integration effort.

6) Example: Robot with a camera system for pick & place applications

Bin-picking – object recognition using 3D image processing

Reaching into the box is one of the most demanding movements for a robot, since the unstructured environment makes it difficult to recognise the individual parts. 3D image processing systems are a promising solution for enabling robots to see.

Camera-based automated quality control

A visual quality check can also be carried out with a camera system. In this example, watch faces are being checked automatically.

Vision guided robot for pick & place application

This vision guided gantry robot uses an ifm O2D camera for a simple pick & place scenario. Further components are the igus Robot Control (iRC) and an igus 3-axis gantry robot.