Using a Cobot with Vision to Pick Parts from Partially Filled Trays

Collaborative Robots Pick Parts From Partially Filled Trays

Josh Westmoreland | September 5th, 2017

When thinking about the potential of adding a robot to production, the first thought usually goes to the activity that is to be accomplished. It could be loading a machine, performing a visual inspection, or any host of other potential applications. However, there is another part of the process that must be considered whether you are looking at a traditional 6-axis robot or a collaborative solution: part presentation. The part presentation is extremely critical and could be the difference between a successful project or one that never gets off the ground.

There are many different methods for giving a repeatable pick location. A very popular option is a gravity feed system, however, this can prove challenging as you are limited by the amount of available space. Another option is to have someone load a fixture that the robot will pick from. This option gives a repeatable position, but you lose some of the benefits of a robot by having someone take the time to load the fixture every cycle. In some applications, this can still be viable, as it allows the operator to run multiple machines.

Loading Parts in Trays

The most common part presentation method we encounter is loading the parts into trays. This allows for large numbers of parts to be loaded and presented to the robot at a time. This reduces the amount of interaction that is needed with the robotic system. The challenge here is that often times the parts are not precisely located in the tray or the tray may only be partially loaded.

This is where the Robotiq Wrist Camera comes in. By leveraging the intuitive software, you can easily teach an object for detection. Once taught, you can simply place the tray within the field of view of the camera and let the combination of Robotiq and Universal Robots do the rest.

The Robotiq software will evaluate all of the parts within its field of view, and pick the part with the highest detection percentage. Once the process is complete, it will move back to the “snapshot” position to start the process again until the tray is cleared. This eliminates the need of starting with a fully loaded tray, as the camera will only send positional information of the parts it detects, and it will ignore the empty locations in the tray.

You can learn how easy it is to program the Robotiq Wrist Camera by watching the video on Robotiq’s YouTube Channel or by visiting Robotiq's Wrist Camera page.

Profile picture for user Josh Westmoreland

Josh Westmoreland


As a Collaborative Robot Specialist, Josh is responsible for the growth and development of new and existing accounts as a part of the CrossRobotics team. With his knowledge and experience involving collaborative robots and surrounding technologies, Josh is able to take a consultative approach with customers to help them find the best solution for their application.