The growing demand for fast supply-chain processes has reached yet another niche market: the unloading of boxes from pallets, or a so-called depalletization. Today, the once arduous task can be automated by implementing a smart solution that will increase productivity, increase throughput and ultimately save costs.
What Is AI Depalletization?
AI-powered depalletization is the automated process of unloading pallets laden with boxes using AI. The system is able to recognize each individual box and the robot places them one by one on a conveyor belt or other predefined place. In contrast to delayerization, where the robot picks the whole pallet with uniform, neatly stacked boxes of the same height, and as opposed to the classic depalletization, AI-powered depalletization is based on smart machine learning algorithms.
The solution presents a higher level of unloading pallets and as such offers numerous advantages. These include a smaller placement area required (sufficient for the size of the largest box in contrast to the whole pallet) and also a smaller robot and gripper since there is a lighter payload that needs to be handled. These advantages merge into one great benefit — significant cost savings. Despite the fact that the gripper is smaller, it can pick boxes as heavy as 50 kg, depending on the particular solution.
How It Works
The secret behind a successful depalletization application is a combination of high-end 3-D machine vision with superior robot intelligence enabled by advanced and sophisticated machine learning algorithms. These algorithms were pre-trained on a large database of boxes. The system immediately recognizes these types of boxes and in case it comes across new ones, it is able to retrain itself very quickly. This happens on a continuous basis and ensures outstanding universality enabling recognition of boxes of different shapes, sizes, or materials.
Shiny, reflecting or black surfaces, varying texture, various patterns or pictures that “mislead” the 3-D vision, tapes coming unstuck, or tightly packed boxes so that it is difficult to recognize the gap that separates them (be it as thin as 0,5 millimeters) — these are challenges that might pose significant problems and play a crucial factor that differentiates the best solutions from weaker ones. The most advanced way that takes the lead in segmentation of the individual boxes on the basis of 3-D image and texture analysis is to use a convolutional neural network (CNN).
Sophisticated depalletization solutions work out of the box, without necessitating any training of the system. Their universality also resides in the fact that the boxes do not need to be stacked in ordered patterns but can be placed randomly and the robot is still able to pick them.
A Perfect Vision
As already said, the key to successful depalletization is the combination of AI and the “right” machine vision. But what does the right one mean?
The deployed 3-D vision needs to provide a large scanning volume on the one hand and high resolution and accuracy on the other. Because the pallets are often stacked upon one another in several layers and the 3-D scanner needs to be able to scan the lowest as well as the highest one and because there needs to be enough space left for the robot to manipulate with the boxes, the scanning volume has to be large enough to scan the pallets from a sufficient distance. Therefore the scanner needs to be mounted approximately 3 to 4 meters above the pallet and still be able to provide scans in superior quality.
Benefits in Numbers
Let’s take an advanced system that was trained on more than 5000 types of boxes. In combination with superior 3-D vision, the sophisticated machine learning algorithms can provide 99.7% pick rate accuracy, gripping precision within 3 millimeters, and picking speed of 1000 boxes in our hour. These numbers are of crucial importance in the context of ROI and also in the decision-making process whether to opt for a smart automated depalletization solution.
Higher Productivity, Safer Workplace
Increased ROI, boosted productivity, and greater time and cost savings present only one side of the spectrum of benefits that can be won by implementing automated depalletization solutions. The other end presents significant elimination of risk of injuries and errors. The manual operation involves handling of large and heavy boxes that may be stacked a few meters above the ground. This often leads to serious back injuries and other health detriments. Automation of this process eliminates these risks and allows for a non-stop depalletization, without the robot ever getting tired, in contrast to human workers.
The workers’ potential can instead be used in areas that require creativity, critical thinking, and decision-making — helping companies enhance efficiency and increase potential for future success.
Andrea Pufflerova is a PR specialist, and Michal Maly is director of AI, at Photoneo.
Timely, incisive articles delivered directly to your inbox.