In the context of the industry 4.0, defined by highly personalized processes, there’s a rising interest of technologies related to the extraction of data of the process in real time to better decide future actions i.e. define more efficient and productive configurations of the industrial plant. From this, Cyber-physical systems such as digital twins are the most suited technology to follow the process performance through data and reliably validate different outcomes from simulations. Furthermore, COBOTs that can adapt to the environment are ideal to deal with such changing demands which have to work alongside industrial robots. Therefore, this study aims to use the digital twin capabilities of data extraction and simulation from the FANUC’s robot simulator software to validate the performance of industrial and collaborative robots in order to offer insights to better apply each type of robot. It was used a robotic cell with the robot PRBT 6 as a reference to make 2 digital twins of the cell with 2 different robot models, one collaborative and one industrial. Two digital twins were made using ROBOGUIDE simulator, version 9.4 for the COBOT and version 9.1 for the industrial robot. It’s presented the method used to build the digital twin of the robotic cell and the results show that the cycle time of the COBOT’s cell is significantly higher than the industrial one but has higher flexibility.

Keywords: digital twin, simulator, industrial robot, COBOT, ROBOGUIDE, lead time

TCP, Tool Center Point; TP, teach pendant; COBOT, collaborative robot; EOAT, end of arm tool.

According to¹ the use of cyber-physical systems has been growing more and more in Industry 4.0, which includes the use of digital twins of industrial processes in order to extract data in real time and perform simulations of different scenarios. From this, these systems gain relevance considering the work of² who cites the “smart factories” derived from Industry 4.0 to deal with dynamic production, characterized by high customization of the product or service, which combine physical and virtual structures and highlights the use of COBOTS as the best solution to deal with the variable process. From this, the digitalization of the process is a requirement of great importance for Industry 4.0 since, according to,³ monitoring plant information is necessary to obtain control data that can be used for better decision-making and generation of insights, such as the work of⁴ and,⁵ who monitor a low-power transformer and a hydraulic plant respectively, using digital twins capable of providing key data in real time from sensors installed in the physical structure. Thus, according to,⁶ digital twins offer companies an opportunity to become more competitive. However, considering consolidated industrial robotics, COBOTS, which are still gaining ground in the industry, need to be applied in the manufacturing process in harmony with industrial robots in order to obtain the best possible combination based on the production characteristics. Thus, the clear difference between industrial and collaborative robots becomes a necessity that this research aims to meet in order to better apply them in the industrial process. Therefore, from the physical robotic cell composed of the PRBT 6 robot that is the object of this research, the objective is to assemble 2 digital twins of the robotic cell using 2 models of robots, the CRX-10iA COBOT and the LR Mate200iD industrial robot model with the robot simulation software ROBOGUIDE version 9.4 for the CRX-10iA and version 9.1 for the LR Mate200iD. Thus, through the simulator tools, it is possible to obtain the performance data of each robot performing the same application.

In order to validate the performance differences between industrial robots and COBOTs, (by FANUC) a physical robotic cell with the PRBT 6 COBOT was taken as a reference to build two DTs using the ROBOGUIDE robot simulator, one with the industrial LR Mate200iD robot model (version 9.1 of the simulator) and the other using the CRX-10iA COBOT (version 9.4 of the simulator). Both virtual robots have to perform the same task, using the same tool (the gripper used by the physical reference robot) while in the same position to validate the impact of using each robot model in the performance of the cell, by validating each robot’s differences and how they affect the cycle time of the cell. The time related data was chosen because of the Lead time importance to increase industrial productivity and profits which is intrinsically related to the performance of the process. It’s important to observe that the physical controller of both FANUC robots is the R-30iB Plus controller. In the virtual cells the following robot models were used: LR Mate200iD (H755-model) and CRX-10iA (H700-model). Which are compatible with the physical controller.

Modeling of the physical workbench

The workbench that composes the reference robotic cell has to be modelled in 3D to be imported in STEP format by the simulator and start building the DT of the whole cell. So, it was used the Inventor professional CAD of AUTODESK to make the workbench using a measure tape to get the measurements needed from the object.

Collaborative gripper

To stablish the DT both physical and virtual robots have to share the same tool to avoid positioning errors while programming the robot offline. The tool used by the physical robot is the SHUNK gripper Co-act EGP-C 40 N-N-ASSISTA adapted, so its 3D CAD model was obtained from the gripper’s producer (SOURCE: SHUNK, 2024) shown on the right of the Figure 1 below and have been modelled on Inventor professional CAD software, AUTODESK, to add the finger extensions used by the physical robot to perform its original task, as shown on the left of the Figure 1 below which are attached to the original fingers. The virtual gripper is also needed to enable the virtual robots to perform the “pick and place” simulations that require a tool to move each part (manipulated element) of the task.

Figure 1 Collaborative gripper.

Structuring the virtual environment: The first thing done after creating the robotic cell was importing the place where the robot model is positioned i.e. the workbench, the coordinates of its position and orientation had to be adjusted to properly stand on the ground. Then the robot model was positioned identically to the physical cell using the measurements of the distance of the manipulator from the workbench edges as coordinates.

Installing the end-effector of the robot: Once properly positioned the tool had to be attached to the robot’s flange and properly configured. The gripper model SCHUNK Co-act EGP-C 40 N-N-ASSISTA was imported to the first tool of the robot (EOAT1) divided in pieces called “body” and “finger” to allow the movement of the fingers when the motion properties of the tool were set.

Defining the “pick and place” task: While making the LR Mate200iD’s DT the task chosen for both robots was defined as a “pick and place” task” repositioning 3 identical cubes from ROBOGUIDE’s library (length of 30mm), initially positioned on the workbench as shown below on the Figure 2. Each block has your base (drop positions during the “pick and place” task). To use as a base for each cube the “plastic pallet” model was imported from ROBOGUIDE’s library and scaled down to 1:10 and positioned in a way that the robot could reach all the bases’ centers to place each block forming the yellow trajectory profile shown in the Figure 2 below.

Figure 2 Initially positioned on the workbench.

Programming of the robot: The first thing done to begin the robot’s program was defining the user frame 1 identical to both DTs, as shown on the Figure 3 below, to have all the program’s position registers referenced to it and not the robot’s base allowing to add precision to the movement of the program’s cloud of points so that the points as a backup for future use. The program was assembled using position registers to guide the robot and have an accurate record of its movements. Thus, each register follows the following naming rule: capture approach points begin with “APROXin”, capture points begin with “PICK”, palletization approach points begin with “APROXout”, palletization points begin with “PLACE”; and all points end with the color of the block they are manipulating, “GREEN” for the green block, “BLUE” for the blue block and “RED” for the red block. The position registers are shown in Figure 4, each one used to form the robot’s path. This way, after setting all the position registers in the first DT with the LR Mate200iD model, by copying the coordinates of each position register from the user frame 1 it was possible to recreate the same position registers in the CRX-10iA DT allowing the COBOT to perform the exact same trajectory as the industrial robot, needed to validate exclusively the differences from each robot’s work capabilities and not the difference of paths

Figure 3 The robot’s program was defining the user frame 1 identical to both DTs.

Figure 4 The position registers.

Simulation of the gripper

In order to add movement to the gripper fingers under the control of the virtual controller forming a DT of the gripper, the movement properties of the moving CAD files of the gripper were set to remain open while the logical state of robot output 1 is HIGH (RO[1]=1) otherwise (RO[1]=0) the fingers would reach a position of 30mm distant of each other, defined this way to simulate the capture of the cube as shown in the Figure 5 below. This way: using the movement properties of the tool links (CAD files) it’s possible to simulate the working of the gripper perfectly identical to the physical counterparty since it’s possible to add the information of speed, set the initial and terminal positions of the movement and the position that the intended captured object will assume on the tool when closed. From that, all the parts (blocks) that the gripper interacts with had to have a position on the gripper defined for the future simulations where each part assumes upon the “pickup” command as shown below on the Figure 5.

Figure 5 Way to simulate the capture of the cube.

Simulation of the robot: Just running the program created won’t track the data needed to validate the performance difference between the robots so a simulation is required to not only validate the performance data but also simulate the robot’s “pick and place” task because the movement of the parts (blocks) depends on the simulation commands of pickup and drop as shown on the Figure 6 below the commands used while teaching the simulation to set the movements of the part 2 (green block) when the simulation is running allowing the robot to capture it, hold it and drop it on its base. All the other two blocks (parts 3 and 4) have the same commands.

Figure 6 Simulation commands of pickup and drop.

While on automatic mode the robot changes the position of the blocks between the following elements: workbench; block bases and gripper. Therefore, by stablishing the positions of the blocks during the task’s transitions it’s possible to properly program the robot positions to capture and place each cube following the real scenario and extract the data needed to validate the robot’s performance.

Structuring the cell’s environment: The COBOT CRX-10iA model was added in the ROBOGUIDE software’s library of the version 9.4. For this study the versions 9.1 and 9.4 of the simulator were considered. The process to build the DT using the CRX-10iA robot similarly to what was done with the LR Mate200iD DT starts importing the same CAD files used before to from the same cell of the LR Mate200iD: workbench; gripper; blocks and pallets from the simulator’s library. But instead of using the industrial robot model the CRX-10iA COBOT model was placed exactly where the LR Mate200iD should be. From that, as both environments share the same universal reference frame and reference frames on the objects it was possible to using the coordinates of position and orientation from the first DT, shown in the Table 1 below, quickly and precisely position all the fixtures (fixed objects), including the COBOT model, achieving two identical robotic cells with different robots.

Simulation of the gripper: With the information previously defined of the movement of the gripper fingers: start and finish positions; I/O control and speed. It was possible to after assembling the gripper using the first tool (EOAT 1) the same way that happened with the first DT.

Programming and simulation of the CRX-10iA robot: Firstly, the user frame 1 of the cell was created in the same position and orientation of the DT using the industrial robot. After that the program of the CRX-10iA DT was built transferring all the points of the LR Mate200iD program by copying the position registers using the coordinates related to the user frame 1 of each point. The exclusive function for ROBOGUIDE cells “POSN” was used to write the coordinates of each point taking the user frame 1 as reference, such tool positions the TCP according to the coordinates imputed by the user. After having all the 13 position registers from the previous DT the program was made following the exact same path and instructions defined before. Therefore, any difference on the performance of both robots comes from the robots’ characteristics only.

Difference of discreet movements

During each simulation in ROBOGUIDE After each movement of the robot its path is registered as a trajectory profile showing the exact path taken by the TCP between each recorded position of the program. Between each point, the path is divided in shorter linear discreet movements that add up to make the robot’s movement profile. The CRX-10iA default trajectory profile has considerably more and shorter discreet movements during aerial motions than the LR Mate200iD trajectory profile as shown in the Figure 7 below: with the purple and green lines respectively the COBOT’s profile and the industrial robot’s profile, the blue lines indicate the speed (50%) of the joint movements i.e. unpredictable moves that follow the quickest path as possible.

Figure 7 LR Mate200iD trajectory profile.

Difference of time between points of the program

ROBOGUIDE has a tool called “MotionPRO” which can validate different performance data from the simulation profile and was used to study the time taken for the motion between points of both robots and the results of the time consumed through all the 22 points in the program are shown in the Graph 1 below.

The Graph 1 below shows the time taken between points of the CRX-10iA and LRMate200iD robots, each time value is the one taken by the robot to get to the point e.g. time of point 1 is for the robot to go from the HOME point (0) to the first, the second point time for the move from the first point to the second. From that, with the data of each movement time, the variance of the values from both robots was calculated finding the following results: variance of movement time for the CRX-10iA robot is 0,3528 seconds and the LR Mate200iD’s variance is 0,007019 seconds. Such difference can be seen on the Graph 1 below with more constant values of LR Mate 200iD’s time between points.

Graph 1 Moving time between points.

Different workspace envelope sizes

The ROBOGUIDE software has the “work envelope” tool that allows to see the space that the selected robot can reach with all the end-effectors registered (0-10). Not necessarily if the point is inside the envelope the robot will reach it due to other issues such as singularity and joint angle limits. Such feature was used while programming the robots to define each point of the program which the robot can reach and perform the “pick and place” task

The definition of the acceptable robot positions proved to be harder to accomplish using the LR Mate200iD robot model due to the smaller reach of the industrial robot compared to the COBOT as the Figure 8 shows below. Using the industrial robot even inside the workspace envelope couldn’t reach the first three positions of the pallets properly, needing to change multiple times the pallets’ positions until the configuration used for both DTs.

Figure 8 LR Mate200iD robot model due to the smaller reach of the industrial robot compared to the COBOT.

Discreet movements differences between the robots

The CRX COBOT according to (DMYTRIYEV et al,)⁷ needs a feedback system that allows the robot to react to the environment in order to guide the system according to the previously established objective. Therefore, the COBOT has considerably more discreet movements to be able to adapt to any change in the environment quickly such as a force applied from the operator, an obstacle or other signals that communicate with the COBOT showing its intelligent capacities, but sacrificing cycle time to improve flexibility by avoiding shortcuts i.e. by using more discreet movements (lines), the path between the points is less linearized and slower compared to one with less discreet movements. Furthermore, the movements of the COBOT are more fluid than the industrial robot avoiding short turns which increases the time spent for each movement, but increases the predictability of the robot’s actions which according to (PALIGA Mateusz)² it increases the confidence of the operator to cooperate with the robotic agent and increase the cell’s productivity.

Difference of work envelope sizes

Also, according to (ZAATARI et al.,)⁸ the manufacturing process needs to quickly react to the varying demands proactively i.e. choose the best motion path for the scenario at hand, if the work envelope of the COBOT is too strict it could limit most of the optimal paths needed for the task, that’s why the larger work envelope of the COBOT could be seen which also applies to this kind of robot’s multiple applications.

Difference of variance of motion time between the robots

The points which the CRX-10iA moves quicker are the ones set to work with the speed of 250mm/s (maximum speed) on linear motions between approximation and contact points, i.e. predictable movements, while the aerial ones set to 50% of max speed on joint motions, unpredictable movements, take more than the double of the time spent during pickup and drop motions. Such difference of speed between aerial and close motions is much higher for the CRX-10iA robot compared to the industrial robot proved by the difference of variance of time spent for each point between them. The CRX-10iA’s considerably slower aerial movement can be explained by the unpredictability of such moves compared to the ones closer to the object, since obstacles such as the operator can interact with the robot during the transition of approximation positions. On the other hand, the industrial robot’s joint movements can work with higher speeds similarly to the linear motions which is proved by the lower variance and faster joint movement compared to CRX-10iA.

Applications of COBOTs and industrial robots   

The industrial robot, by choosing the shortest paths to reduce time consumption, reduces the time consumed for each movement, but without the environment feedback this robot’s moves become a risk to any human close to it, having to remain isolated. This way, the industrial robot is ideal to work continuously on the same task independently to any change of the environment ensuring high productivity for the cost of low flexibility, which can be explained by the smaller work envelope focused to the task which was designed to perform. So according to (SAHAN MOHAMMED et al,),⁹ the industrial robots fit better for large batch production while COBOTS with a slower pace and high flexibility are perfect for small and variable batch productions.^10–14 

The current research presents the complete process of making two working digital twins of robotic cells using the dedicated robot simulator ROBOGUIDE, explaining the method to guarantee the virtual coherence. According to the clear difference in terms of cycle time, range and versatility of the robot types studied, industrial robots are better applied for non-variable scenarios working faster than COBOTs isolated from human activity ensuring productivity over flexibility. While COBOTs due to their high adaptability to the environment can function nearby humans adjusting to multiple scenarios on the run guided by the operator but sacrificing speed and payload for it.

I thank God for my second chance to live and be able to develop my studies.

I thank the Advanced Robotics Institute- IAR- for all the support with all of its modern infrastructure and resources of software and robots.

I thank my research advisors Dr. Rogério Adas Pereira Vitalli and Prof. Julio Cesar de Almeida Freitas for all their knowledge, research topic and orientation.

I thank My family for all of the support that let me focus on this research.

The authors have no conflicts to declare.

1. Alice L, Gianlucca C, Fábio S, et al. Real-time inventory request and monitoring system using Node-RED for remote teaching.  Presented at the XV BSIA: Brazil. 2021.
2. Mateusz P. Human–cobot interaction fluency and cobot operators’ job performance. The mediating role of work engagement: a survey. RAS. 2022;155:104–115.  
3. Hitalo R, Italo S, Vinicius P, et al. Virtual setup - automatic virtualization of industrial plants. Presented at the XV BSIA: Brazil. 2021.
4. Rodrigues LA, Santos MA, Rodrigues G, et al. Low transformer monitoring based on internet of things and digital twins. Presented at the BCA: Brazil. 2022.          
5. Caio N, Ademar N, Juan V, et al. Methodology applied to the development of digital twin of water pumping systems. Presented at the BCA: Brazil. 2022.  
6. Guiffo K, Jürgen R. Value-driven robotic digital twins in cyber-physical applications. TOII. 2020;17(5):1551–3203.
7. Yevheniy D, Marco C, Hermes G, et al. On cobot programming in industrial tasks: a test case. Presented at the HORA: Turkey; Ankhara. 2022.
8. Shirine Z, Mohamed M, Weidong L, et al. Cobot programming for collaborative industrial tasks: An overview. RAS. 2019;116(3):162–180.
9. Mohamed AS, Kathiravan S, Lokesh M, et al. Role of cobots over industrial robots in industry 5.0: a review. ICAECA. 2023.
10. Saixuan C, Omar A, Guanwu J. A General analytical algorithm for collaborative robot (cobot) with 6 degree of Freedom (DOF). ICASI. 2017.
11. Panariello D, Caporaso T, Grazioso S, et al. Using the KUKA LBR iiwa robot as haptic device for virtual reality training of hip replacement surgery. IRC. 2019
12. Joris G, Oliver G, Eric N, et al. Learning local trajectories for high precision robotic tasks: application to KUKA LBR iiwa Cartesian positioning. IECON. 2016.    
13. Stephanie G, Brigitte K. A communicative perspective on human–robot collaboration in industry: mapping communicative modes on collaborative scenarios. IJSR. 2024;16(6):1315–1332.    
14.  Claudio T, Francesco A, Nicola P. COBOT applications—recent advances and challenges. Robotics. 2023;12(3):79–106