Innovative Random Depalletization Solution | CMES Robotics
                     Case study
Random Depalletization
Case Study - Random depalletization
1. CHALLENGE
·Difficulty handling orders that have surged over 130,000 per day.
·Limitation of manpower due to pandemic
·High staff turnover due to repetitive and laborious labor
·Existing robot automation, which was one of the solutions, had many restrictions such as  various box sizes
2. SOLUTION
·Box depalletizing software (EQUAL™ for logistics) SUMMARY
   - New-Box registration package 3. RESULTS 
   - Calibration package (vision, robot, vision to the robot) ·EQUAL™ simulator reduces process setting time and 
   - Advanced path generation package    increases maintenance convenience
·6-axis robot and Multi-channel vacuum gripper ·Greater throughput (600 boxes per hour)
·3D scanner for box recognition on pallets ·Enhanced speed and accuracy
·Processing unit ·Effect of replacing two workers (depending on 
·Conveyor system    working time setting), ROI 18-24 months
CHALLENGE
1. BACKGROUND
Korea's e-commerce industry has grown by 250% in the last 
eight years, and the total market size is expected to reach 
120 billion dollars as of 2020. The gross revenue of top 
e-commerce groups in 2020 increased by more than 45% on 
average compared to the previous year. In particular, online 
consumption compared to offline has multiplied because of 
the COVID-19 pandemic, so e-commerce orders in the first 
quarter of this year increased by more than 50% compared 
to the same period last year. 
Meanwhile, e-commerce companies are struggling to come 
up with a way to handle orders that have risen sharply to 
over 130,000 per day.
It is because the pandemic not only requires strict prevention management of the working environment but also limits the 
density of the workforce on the shop floor. Also, high staff turnover is occurring due to the repetitive labor of moving heavy 
products in the distribution center. To solve this problem, the need for robot automation emerged to tackle efficient 
workforce assignment and unloading of the warehoused product boxes in areas of high staff turnover, and CMES participat-
ed in this project.

Case Study - Random depalletization
2. CHALLENGE
The first process of the e-commerce distribution 
center starts with the process of putting product 
boxes received from various suppliers into the 
distribution center in the form of pallets.
The operator moves the product boxes on the pallet 
to the conveyor belt. At this time, the product boxes 
are not uniform in size, the position on the pallet is 
not regular, and there are a lot of boxes that are 
heavier than 10 kg.
“Currently, conveyor belts are responsible for 
transporting products from one process to another, 
but the majority of the other work, especially the 
handling of millions of boxes of goods of varying 
weight and shape, is often done by humans.” 
explains Operation Manager of a logistics company, 
one of our clients. Source : Coupang YouTube
As the existing robotic automation could only handle boxes of consistent size and with certain conditions, there was a limit 
to realize full automation in reality.  It stems from the fact that the robot could not take boxes of irregular sizes. Considering 
these limits, CMES offers advanced automation to reach new levels of output and performance.
SOLUTION
1. PACKAGE SPECIFICATIONS
ITEM SPEC
Depalletizing Speed*1 600 boxes/hour (6~8 sec/box, based on condition of environment)
Maximum Box Size*1 600 mm x 500 mm x 400 mm(W,H,D)-can be customized
Minimum  Box Size*1 250 mm x 200 mm x 100 mm (W,H,D)-can be customized
Maximum Box Weight*1 < 20 Kg
Available Workspace 1.1 m x 1.1 m x 2.0 m
( pallet size )
>3 kW, 200-600VAC, 50~60Hz
Power consumption*1 Robotic System : 1.62 kW (average)
Processing Unit : 0.7 kW
IP classification IP54
Air Inlet Dry Compressed Air @ 4~5 Bar
Input / Output Digital In (8 ports), Digital Out (8 ports)
Communication TCP/IP interface to PLC
Warranty 1 year
*1: Can be modified 

Case Study - Random depalletization
2. PACKAGE COMPONENTS
1) Software *Robot specifications
Box depalletizing software ITEM SPEC
( EQUAL™ for logistics )
Controlled axes 6
    - New-Box registration package Maximum payload (kg) 40
    - Calibration package
    - Advanced path generation package Repeatability (mm) +/-0.06
2) Hardware Horizontal reach (mm) 2,552
Compact booth style equipment Weight (kg) 435
① *Robot (6 axis)
② Multi channel vacuum gripper Ambient temperature 5 to 45̊ C
③ Processing unit (GPU for AI)
④ 3D scanner for box detection on the pallet Power type 200~600V, 50/60hz 3-phase
⑤ Vision for box height calculation
⑥ Calibration target Ave. Power consumption (kW) 1.62
⑦ Pallet conveyor system (including PLC)
3. MECHANISM
③ AI processing 3D image capture
① 3D image capture
3D vision
② Send image data
AI processing
Boxes
Conveyor
Send coordinate
6-axis robot
⑤ Depalletize Depalletize
After the 3D scanner scans boxes of various shapes and sizes, AI Processing recognizes the shape of the box and locates it.
The location information found is transmitted to the robot along with the path in the form of coordinates. 
The robot transfers the box to the conveyor belt using the coordinates.

④ Send coordinate
Case Study - Random depalletization
4. FUNCTIONS
1) BASIC 2) ADVANCED
① Accurate detection · Collision-free robot path generation
  · 3D scanning and box location detection · Automatic calibration(Vision, Robot, Vision to 
  · Randomly placed box picking     the robot)
  · AI-based box recognition + Rule-based Algorithm · Box registration software for AI training of a new box
     for accurate detection · 3D simulation interface
  · Box size detection (including box height)
② Comprehensive handling for full automation
  · Retry in case of box drop and non-recognition
  · Self-diagnosis
  · Statistics reporting about depalletizing results
③ Powerful and fast handling
  · Multi-channel vacuum gripper 
     (increase gripping robustness)
5. ADVANTAGES
1) AI-BASED BOX RECOGNITION
Boxes are recognized using an AI core that has learned by big data of 
hundreds of thousands of box images.
Distinguishes the boxes from other objects, and can recognize individual 
boxes regardless of the waybill on the box and various taping conditions.
Provides software that enables evaluation and registration of new boxes.
AI Recognition result
2) 3D POINT CLOUD PROCESSING
Minimizes interference with surrounding objects by extracting only the data 
necessary for box recognition from the 3D scanned data.
In addition, using 3D depth information, it is possible to reduce the probability 
of erroneous detection/non-detection of the box and extract the exact 
location of the box.
3D Point cloud processing and 
picking position calculation
3) ROBUST VACUUM GRIPPER 4) RETRY FOR CONTINUOUS OPERATION
Multi-channel vacuum system and sealing foam can be In the case of occasional erroneous detection and 
used to pick up objects that may be wrinkled such as the non-adsorption, rescan and retry are automatically 
box surface. Through periodic replacement of sealing performed to minimize equipment stop time to maximize 
foam, the gripping performance can be maintained in an productivity. Statistics on the frequency and number of 
optimal state. retries are made to help to maintain equipment.

Case Study - Random depalletization
5) COLLISION-FREE PATH GENERATION 6) EASY & AUTOMATIC CALIBRATION
In order to prevent damage to the box itself or surround- The calibration between the 3D vision sensor and the robot 
ing facilities while the robot picks up the box, and to can be changed continuously after initial setup, and this 
guarantee the life of the robot and the gripper, a robot can degrade the robot's gripping performance. Through a 
path that avoids collision is automatically created. This very easy interface, the user can perform calibration either 
feature is a must-have feature especially for roll pallets. directly or in a periodic automatic mode, so that the 
calibration state is always optimally maintained.
7) 3D SIMULATION INTERFACE
Hardware setup can be reviewed in advance in a 3D simula-
tion environment, and intuitive setup for the workspace of 
robots and 3D sensors is possible.
It is easy to check current equipment status, box detection, 
and robot path generation results, and calibration results in a 
3D simulation environment.
8) COMPATIBILITY WITH MOST INDUSTRIAL ROBOTS
Interface with almost all industrial robots is possible, and robot movement can be optimized to increase productivity.
RESULTS
“The existing robotic automation could only handle boxes of a certain size and 
under specific conditions. So, there was a limit to realize full automation because 
the robot could not handle boxes of irregular sizes. However, the robot can now 
recognize the irregularly shaped boxes and correctly pick them up and accurately 
move them to the conveyor,” automation process managers from a lot of logistics 
centers explained. If the robot moves one box at a time, it can handle 600 boxes 
per hour.
The robotic system such as random mixed depalletization is usually able to run at 
Source : CJ Logistics YouTube
95 percent efficiency, while the average worker is usually about 20 to 25 percent 
efficient during any given shift due to the consideration of break time and working time. If the gripper moves 2 boxes at once, 
it can process more boxes within the same working time.
CMES' Random Mixed Depalletizing Application improves production efficiency by sorting or moving boxes of irregular sizes 
and heavy boxes by processing the planned workload at a fixed time.
If a person or a robot picks up and moves one box on to the conveyor belt at a time and there are 7,500 boxes to move from 
a pallet to the conveyor belt, 9 to 10 people are required, compared to 12.5 robots when random mixed depalletization is 
applied on the line. In terms of the cost-saving aspect, ROI would be between 18 to 24 months. If the robot moves two boxes 
at a time, the productivity effect will be more significant.
Website www.cmes3d.com
USA Office 2205 152nd Ave NE, Redmond, WA 98052 
R&D Center 3-7, Seolleung-ro 132-gil, Gangnam-gu, Seoul, Republic of Korea (06065)
South Korea HQ No.502, 35, Techno 9-ro, Yuseong-gu, Daejeon, Republic of Korea (34027)
Contact South Korea - [email protected]/ +82 1800 8647 / FAX : +82 0505 355 6009
USA - [email protected] / +1 206 200 0444