Bin-picking using robots and 3D vision

The most competitive businesses are identifying pain points in their processes and using automation to make completing these tasks easier, faster and cheaper. Bin picking is a great example of this.

Taking a product out of a bin and placing it into the next container for further processing or packaging is boring, repetitive and dull. Nevertheless, it requires a high level of accuracy and consistency, making this process an ideal candidate for automation.

Traditionally, visions systems that were designed to automate bin picking have been heavily dependent on CAD software. Unfortunately, this doesn't work for non-uniform products - think fruits or vegetables.

Today, processing plants are opting for 3D vision systems that no longer require CAD software to identify products. These vision systems are much simpler to use and can recognise the position of randomly assorted items in a bin, regardless of shape or order.

Non-uniform objects

Using these systems facilities that can handle complete randomness, in relation to a products shape or size. Manufacturers no longer rely on CAD to recognise the shapes of products - I like to use the example of mango and apples.

There's no CAD data for fruit. However, today's vision systems can cope with this challenge and make intelligent decisions on how to carry out this task, picking the fruit effectively in the most logical order, even if the bin contains exceptionally wonky fruit.

This is carried out through passive imaging. The latest vision systems use two high-speed cameras capturing 3D images at 30 frames per second. This generates a stable 3D image of the workpieces, which is sent to the robot via a robot controller. The system can then process these images and identify the most logical order to pick up the wonkiest of mangos and apples, with sub millimetre accuracy.

The best robot for the job

Bin picking automation requires a robot with six axes of movement, so it's the six-axis vertically articulated robots that are being used for this application. These robots can vary their approach as they reach into the bin, meaning they can reach more workpieces. What helps is that the stereo camera is constantly checking the robot's working envelope, if a workpiece is out of reach, it is skipped.

This is important when you considering the high levels of strength and speed of most six axis robots. Having the robot hit the side of the box could be problematic, driving production to a halt or damaging products.

Bin picking is not only boring, but painfully repetitive for a human worker. By automating this task with the correct vision software, robots can pick-and-place even the wonkiest and non-uniform items.

More information and a short video about Toshiba Machine's 3D vision system, TSVision3D, can be found on the TM Robotics website www.tmrobotics.co.uk/ts-vision3d.