Autonomous integration of an external asset to an assembly line

Suppose the most critical asset of an assembly line is incapable to handle a temporary surge of demand or worse it is out of order. One solution to address such a scenario would be to have a fully automatic and versatile Plug and Produce (PnP) system, ready on standby that can be summoned to the assembly line. This solution could serve as a support to keep up with the demand and also be utilized as a substitute for existing assets to reduce potential outages.

Current Setup

At the Institute of Mechatronic Systems (IMS) such solution a is in the implementation phase on the digital learning factory SmartPro 4.0. As of now, the purpose of SmartPro 4.0 is to demonstrate the current possibilities and potentials of industry 5.0 through the assembly of a pen by utilizing technologically advanced components and state-of-the-art algorithms concerning computer vision, process optimization and monitoring tools. The architecture of SmartPro 4.0 consists of seven different systems; a linear transport system, four pick and place stations, a printer and lastly the assembly station highlighted in yellow (Figure 1).

3D representation of the SmartPro 4.0

Figure 1: 3D representation of the SmartPro 4.0

 Assembly station of SmartPro 4.0

2: Assembly station of SmartPro 4.0

The assembly station is made up of a robot arm from Stäubli, a linear actuator and various contraptions required to hold the parts of the pen (Figure 2). On average the robot arm requires 58 seconds to assemble a pen. However, during this assembly process, the linear transport system is forced to wait till the robot arm has retrieved all of the components of the pen. As a consequence, the remaining stations are in an idle state during this entire assembly period. Due to this fact, the assembly station represents the main bottleneck of the production rate of the SmartPro 4.0. In addition, it also happens to be a single-point-of-failure, meaning the entire production would be on halt in case the robot arm would experience a malfunction.

Development of a PnP-System

To address this issue the IMS developed a PnP-system in the form of a mobile station to increase the current production rate and also serve as a backup to prevent outages. In terms of replicating the activities of the Stäubli robot arm, a cobot with seven degrees of Freedom named Panda from the German company Franka Emika was selected [1]. The required equipment for delivering the cobot in an operatable manner consisted of a controller for the arm, an industrial PC (IPC), and a server-grade uninterruptible power supply (UPS) for the transition phase. To house these necessities a contraption made out of Bosh Rexroth beams was assembled. For delivering this station an automatic guided vehicle (AGV) from the Japanese company Omron name LD Cart Transporter was selected. This combination of a cobot mounted on an AGV resembles a self-adaptable PnP-system (Figure 3) that can be summoned wirelessly by the SmartPro 4.0 through the OPC Unified Architecture standard [2].

Figure 3: PnP-System developed by the IMS

Management of the docking offset

During the experiments, the AGV was not capable of providing an accurate reproducible motion (transitional and rotational) of the PnP-system, despite the numerous sensors. The measured offset of 100 attempts was within ± 20 mm in the X-Y-axis and a rotational deviation of close to ± 3° around the Z-axis. The underlying reasons for this limitation in control are numerous such as sensor feedback in terms of accuracy and precision, relatively high mass (~75 kg) in terms of inertia, and also the floor properties in terms of friction. To assist the AGV a pair of guiding struts (Figure 4) were attached on the side of the SmartPro 4.0 that would mechanically reduce the deviation of the docking process to ± 2 mm on the X-Y-axis and 0.5° on the rotational axis.

Figure 4: Concept utilizing mechanical guiding struts for the docking of the PnP-System with inaccurate positioning

Figure 4: Concept utilizing mechanical guiding struts for the docking of the PnP-System with inaccurate positioning

However, this deviation was still too high for a reliable electrical power connection (230 VAC) required by the components and to keep the USP charged. Instead of using commonly available industrial power plugs and socket a more tolerant power plug and outlet was customized with Stäubli’s CombiTac tool [3] (Figure 5).

Figure 5: Docking-friendly power plug and power socket

Figure 5: Docking-friendly power plug and power socket

Further Improvements

This new concept represents an inspirational form of a self-organized group of machines and robots without human interaction. However, there are still significant shortcomings present. For instance, the guiding struts provided better handling of the AGV's inaccurate positioning, however, it could not be reduced to absolute zero which is required by certain picking and placing activities of the cobot. To calibrate the offset dynamically a solution from the realm of computer vision is being implemented. The currently considered approach is utilizing a robot guiding solution from Pickit, the Pickit 3D camera [4].

  1. A Tutorial on the Reference Platform for AI and Robotics (AIR) from Franka Emika. FRANKA EMIKA, 2022. Accessed: May 26, 2022. [Online]. Available:
  2. S.-H. Leitner and W. Mahnke, ‘OPC UA – Service-oriented Architecture for Industrial Applications’, p. 6.
  3. ‘Applications for CombiTac - Stäubli Electrical Connectors’. (accessed May 26, 2022).
  4. ‘Guide your robot with Pickit 3D vision solutions - Pickit 3D’. (accessed May 26, 2022).
Partner Name:
zhaw logo