Depalletising heavy products such as large sacks of tea, flour or cement is a challenge for humans because of the dangers involved with heavy lifting. Removing a sack from a towering pallet of sacks is hard physical work.
This is an example of an industrial application that would be solved by robotic lifting. Until recently however, picking a sack from a pallet was a difficult problem to solve because the ideal presentation of sacks is uniform and tidy. The reality though is often very different as product moves during transport and the height, plane and shape of a sack cannot easily be determined using standard 2D vision systems.
Ideal stack of sacks...
|versus the reality|
A major UK based tea manufacturer who processes thousands of tons of tea for the home based consumer market, producing millions of tea bags a year, required a sack depalletising system to deal with 2 metre high pallets of tea sacks. We installed the first system in 2014 and since that date the system has been operating sometimes 24 hours per day, 7 days per week.
The 3D sack picking system is now a critical system in the overall tea bag production and so the customer relies on the robot depalletiser to fulfill this difficult task with robotics and very minimal human intervention.
The system deals with poorly laid out tea sacks that are threatening to slide off the stack. The Scorpion 3D Stinger camera builds a 3D pointcloud that is used to model the top layer of the tea sacks. Using a combination of 2D image data and 3D pointloud data, Scorpion Vision Software works out the height, position and pose of the sack in space.
The robot is confidently supplied with this data and so is able to rotate and adjust the end effector to collect the sack from the stack, no matter how it lies or the angle at which it lies. Should the sack be split, or the angle at which the sack is lying is too extreme, the vision system will stop and await intervention to rectify the problem so the likelihood of split sacks being picked is minimised (and in fact is very unlikely).