Advertisement

Integrating Mobile Robots into Automated Laboratory Processes: A Suitable Workflow Management System

      Abstract

      The general trend of automation is currently increasing in life science laboratories. The samples to be examined show a high diversity in their structures and composition as well as the determination methods. Complex automation lines such as those used in classic industrial automation are not a suitable solution with respect to the required flexibility of the systems due to changing application requirements. Rather, full automation requires the connection of several different subsystems, including manual process steps by the laboratory staff. This requires suitable workflow management systems that enable the planning and execution of complex process steps. The integration of mobile robots for transportation tasks is currently an important development trend for realizing full automation in life science laboratories. The article “Workflow Management System for the Integration of Mobile Robots in Future Labs of Life Sciences” presents the development and application of a hierarchical workflow management system (HWMS) as a top-level process management and control system. This concept combines the typical hierarchical automation structure with novel approaches for the integration of transportation tasks with variable degrees of automation. The aim is to create a general-purpose workflow management system that can be used in different areas of the life sciences, regardless of the specific device components and applications used.

      Keywords

      Comment on: Neubert, S.; Gu, X.; Göde, B.; et al. Workflow Management System for the Integration of Mobile Robots in Future Labs of Life Sciences. Chem. Ing. Techn. 2019, 91(3), 294–304.

      Automation for Life Science Laboratories

      The automation of chemical or analytical laboratories is currently an important development topic. The samples, which have to be analyzed, show a high diversity in their structures and composition as well as the determination methods. Complex automation lines such as those used in classic industrial automation are not a suitable solution. Although they can handle highly complex processes, they cannot be easily adapted to changing requirements. Automation systems in industry are used for a special process (e.g., the packaging of certain products or the production of a certain type of car). In contrast, in the life sciences, we see a greater change in applications (sometimes weekly) that are to be carried out on an automation system. This requires very high flexibility from the systems. The laboratory infrastructure (except for drug development laboratories) is mainly characterized by a mixture of individual devices, workstations, and integrated systems currently (in contrast to industrial applications, in which we usually see large centralized systems). Laboratory personnel must transport samples, reagents, and labware between individual stations. This either limits a 24/7 automation or requires additional personnel to operate the systems in shifts. The use of mobile robots, which transport samples and labware between the individual and building-wide distributed stations, is a useful development. Full automation and thus an almost 24/7 operation are possible (constant operation is typically limited by maintenance work). In addition, laboratory staff can be freed up for creative planning and development tasks. Various mobile robot systems are currently commercially available, including the H20 (Dr. Robot, Richmond Hill, ON, Canada);
      • Liu H.
      • Stoll N.
      • Junginger S.
      • et al.
      Mobile Robotic Transportation in Laboratory Automation: Multi-Robot Control, Robot-Door Integration and Robot-Human Interaction.
      the KUKA Mobile Robot iiwa (KUKA Roboter GmbH, Augsburg, Germany);
      • Mokaram S.
      • Aitken J.M.
      • Martinez-Hernandez U.
      • et al.
      A ROS-Integrated API for the KUKA LBR iiwa Collaborative Robot.
      the Omron LD series, for example equipped with collaborative robot (Omron Corporation, Kyoto, Japan); or assistance robots LiSA (Fraunhofer IFF, Magdeburg, Germany)

      Fritzsche, M.; Schulenburg, E.; Elkmann, N.; et al. Safe Human-Robot Interaction in a Life Science Environment. IEEE International Workshop on Safety, Security and Rescue Robotics SSRR 2007, Art. No. 4381273.

      and Kevin (Fraunhofer IPA, Stuttgart, Germany). Suitable workflow management systems are required to coordinate the subtasks of all integrated devices and systems. We developed a hierarchical workflow management system (HWMS) that follows the general structure of laboratory automation systems and helps us to combine the different devices and system within a laboratory or even in a distributed laboratory environment (e.g., over different floors). The bottom layer (field layer) contains the sensors and actuators, controlled by CPUs or microcontrollers and a specific machine language. The instruments are located on the instrument control layer. They mostly consist of vendor-specific software solutions. Appropriate instrument interfaces permit the execution of functional procedures by the implemented sensors and actuators. Process control systems (PCSs) are located on the process control layer. They usually perform the control and monitoring of several substations. These can be simple devices such as balances and liquid-handling stations or more complex “robot-based“ systems. Structures combining different heterogeneous subsystems with a different degree of automation (e.g., integrated systems, workstations, and peripheral instruments) necessarily require the use of a PCS. These systems allow the generation and execution of methods and procedures using one or more substations. On the workflow control layer, the single subprocesses, located on the subordinated layers, are largely integrated by workflow management and control systems [workflow management system (WMS), process management system (PMS), method execution system (MES), and laboratory execution system (LES)]. These systems also support the planning (process definition) and execution of workflows in the sense of complex process controls of the workflow control layer.

      Architecture of a Hierarchical Workflow Management System

      The developed HWMS is located on the workflow control layer and has access to all hierarchically structured control systems of the individual stations (see Fig. 1). Thus, this approach keeps all subsystems divided and interchangeable on their layers. It uses their specific advantages in contrast to available scheduling and control systems, which often combine function of both process and workflow control layer. The individual systems can be connected to the HWMS via the common data transmission protocols TCP (Transmission Control Protocol) or UDP (User Datagram Protocol) interfaces, database interfaces, or web services.
      Figure 1
      Figure 1Typical hierarchically structured environment of the hierarchical workflow management system (HWMS).
      The presented workflow management system consists of different subsystems. A novel approach is the consideration of a transportation and assistance control system (TACS). The TACS receives transportation requests for labware or sample transports between the subsystems, and distributes them between the mobile robots and/or the human operators. The robot remote center (RRC) is the central control authority for the integrated mobile robots. It receives the commands from the TACS and translates them into path and transportation-relevant information for the respective robots. Based on the type of task, the distance between the destination(s), the charging status of the mobile robots, and the other processes in the system, the RRC delegates a task to a specific robot. The implementation of several workstations and integrated systems thus also requires the installation of several PCSs or different instances of one PCS. Currently, only the widely used scheduling system SAMI EX is integrated with three instances. So far, it does not support a standardized interface as a gateway for systems on higher layers but offers via SILAS a convenient software interface. Thus, the development of an adapter solution was possible to extend the local, proprietary interface via a network-compatible web service interface. This interface, which has to be implemented for each PCS instance, permits the HWMS to schedule and execute planned methods in the PCS and to request status, error, and setting information. The integration of further or alternative PCSs for the control of transportation units (as the RRC) or automation islands is possible with an easy adaptation of the communication protocol. Thus, any station can be integrated into the system, depending on the user requirements, and is available in the HWMS. The labware location guidance system (LLGS) is an extension of the adapter solution for the organization of the labware access by human operators. Therefore, the LLGS is coupled with an indication system based on light-emitting diodes, which shows the status of the divided labware (>400 implemented positions) on the automation islands. If labware of different positions is required for the execution of a current method, then their respective positions are marked red. The positions of labware ready for collection are marked yellow.
      To execute a process workflow, it is first necessary to create a process image in the HMWS (planning phase). This requires a prior definition of all existing devices, systems, and stations, including their positions, possible labware types, and the existing methods. These data form the basis for the mapping of the process in the HWMS and the creation of a process model. In the planning phase, the model editor establishes connections to the required substations (instruments, integrated systems, and transport systems) and requests the conditions for the subprocess execution (e.g., execution time, available methods, required labware types, and labware start and end positions). This information builds, together with the basic data, the foundation for the workflow scheduling. In addition, transport processes between the stations can be assigned to either mobile robots or laboratory personnel (integrated by smart devices) by the user. Alternatively, this option can be designed flexibly. In this case, the TACS module autonomously decides to whom a transport order will be handed over. A plausibility check of the planned process steps is carried out in parallel with the creation of the process model.
      Once the process model is created, an initial scheduling optimizes the individual process steps to combine parallel processes and minimize the overall process time. The process execution controller initiates the execution of the individual process steps related to the scheduling results. After execution, the HWMS is notified so that the next subprocesses can be started.
      The further scheduling is, however, of crucial importance. Our system enables the parallel execution of different workflows. These different workflows might partially use the same subsystems such as liquid handlers, analytical systems, or centrifuges. We integrated a dynamic scheduler to achieve optimal use of all subsystems. The requirements of the process sequences (e.g., time restrictions for the execution of individual subprocesses) were considered. The dynamic scheduler also enables suitable reactions to changes during the runtime. Unexpected changes can lead to process malfunctions and initiate a rescheduling of the process flow. This increases the stability of the workflow execution considerably.
      • Cabrera C.
      • Fine-Morris M.
      • Pokross M.
      • et al.
      Dynamically Optimizing Experiment Schedules of a Laboratory Robot System with Simulated Annealing.
      The greatest uncertainties in the overall system arise from the transport processes. This includes, for example, necessary path changes of the mobile robots due to obstacles or necessary charging procedures. In addition, malfunctions in the picking and placing of labware can occur. This influences the overall process and the execution of the different interdependent subprocesses. The number of available transport media (robots and human operators) can fluctuate during the process. The laboratory staff can be busy with other tasks or unavailable (e.g., at night or during the weekend). Robots can be logged out of the system because the batteries need to be charged. Once a new mobile robot has been registered in the system, transportation tasks can be distributed to several robots. This can speed up the process execution. If the transport tasks of a mobile robot fail, the request is transferred to an available human operator. In addition, the failure of different subsystems during the process flow requires rescheduling.
      • Gu X.
      • Neubert S.
      • Stoll N.
      • et al.
      A New Method for the Indicator of Dynamic Scheduling in Life Science Laboratories Using Artificial Neural Networks.
      The HWMS may not necessarily be the highest level of the workflow structure. Its combination with a higher level business process management system (BPMS) allows modeling the end-to-end processes, including all involved (manual) operator-based and automated subprocesses. The BPMS offers the modeling and execution of manual processes and the integration of simple automated subprocesses by the graphical representation business process model and notation 2.0 (BPMN 2.0). For complex automation processes, the HWMS provides greater process proximity and simplifies the process planning significantly, in contrast to a pure BPMS solution.
      • Neubert S.
      • Göde B.
      • Gu X.
      • et al.
      Potential of Laboratory Execution Systems (LESs) to Simplify the Application of Business Process Management Systems (BPMSs) in Laboratory Automation.

      Applications

      Distributed automation systems are used in a variety of applications in analytical measurement technology, including the investigation of environmentally relevant samples or medical samples. In the article discussed here, the authors describe inter alia the application of the system for the determination of calcium and phosphorus in bone materials using inductively coupled plasma mass spectrometry (ICP-MS).
      • Fleischer H.
      • Vorberg E.
      • Thurow K.
      • et al.
      Determination of Calcium and Phosphor in Bones Using Microwave Digestion and ICP-MS: Comparison of Manual and Automated Methods Using ICP-MS.
      Prior to measurement, extensive sample preparation is required, which includes both manual steps (e.g., weighing of sample quantities) and automated steps (e.g., dosing of digesting acids, internal standards, and solvents for dilution). The entire process is distributed at four locations (manual station, reformatter for liquid-handling processes, microwave for sample digestion, and ICP/MS for measurement of the elements). The necessary transportation steps between the substations are carried out either by a mobile robot or by the laboratory staff, who are integrated in the transportation logistic via the supported smartphone solution. Registered human operators can receive orders for labware transportations and service tasks (e.g., error handling or required preparations) via a smartphone app. Transportation tasks must be confirmed by barcodes. The entire process is planned via the HWMS, and the execution instances are monitored. In the event of errors, the described rescheduling is carried out.
      Due to the hierarchical structure of the system, simple adaptation to other applications is possible. For example, the determination of mercury in wood samples
      • Vorberg E.
      • Fleischer H.
      • Junginger S.
      • et al.
      Automated Sample Preparation for Mercury Analysis in Wood Materials.
      or incrustations in stents
      • Fleischer H.
      • Ramani K.
      • Blitti K.
      • et al.
      Flexible Automation System for Determination of Elemental Composition of Incrustations in Clogged Biliary Endoprostheses Using ICP-MS.
      uses a similar sample preparation. The method planning tool is used to adapt the method parameters. Additional stations (e.g., further analytical measuring systems) are included in the planning. For example, this is required for the determination of cholesterol in bile.
      • Fleischer H.
      • Roddelkopf T.
      • Stoll R.
      • et al.
      Automated Analytical Measurement System for Determination of Cholesterol in Pig Bile.
      This process requires the integration of different mass spectrometers for the measurement. Depending on the available laboratory resources, simultaneous planning and execution of different processes are possible.
      • Fleischer H.
      • Thurow K.
      Automation Solutions for Analytical Measurements: Concepts and Applications.
      We have developed a hierarchical workflow management system for use in life science laboratories. The system is currently limited to a few special devices and device systems. Due to the architecture used, any other device can be connected. The integration of different mobile robots is also possible and a current task for us. The goal is to integrate any mobile robot into the process. We achieve the highest flexibility if the shared use and control of mobile robots from different manufacturers are possible. Because the use of mobile robots still has many shortcomings, event-based scheduling is of particular importance. With the possibility of reacting to unforeseen events (necessary loading processes, obstacles in the way of the robot, errors in the transfer of samples and labware, etc.), actual flexible automation in complex laboratories of the life sciences is possible.
      Declaration of Conflicting Interests
      The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

      Funding

      The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by the Ministry of Economic Affairs, Construction and Tourism of Mecklenburg-Western Pomerania (Germany, FKZ: V-630-S-105-2010/352, V-630-F-105-2010/353) and the Federal Ministry of Education and Research (Germany, FKZ: 03Z1KN11, 03A1K11). This work has been supported by the European Union.

      References

        • Liu H.
        • Stoll N.
        • Junginger S.
        • et al.
        Mobile Robotic Transportation in Laboratory Automation: Multi-Robot Control, Robot-Door Integration and Robot-Human Interaction.
        IEEE ROBIO. 2014; : 1033-1038
        • Mokaram S.
        • Aitken J.M.
        • Martinez-Hernandez U.
        • et al.
        A ROS-Integrated API for the KUKA LBR iiwa Collaborative Robot.
        IFAC-PapersOnLine. 2017; 50: 15859-15864
      1. Fritzsche, M.; Schulenburg, E.; Elkmann, N.; et al. Safe Human-Robot Interaction in a Life Science Environment. IEEE International Workshop on Safety, Security and Rescue Robotics SSRR 2007, Art. No. 4381273.

        • Cabrera C.
        • Fine-Morris M.
        • Pokross M.
        • et al.
        Dynamically Optimizing Experiment Schedules of a Laboratory Robot System with Simulated Annealing.
        JALA. 2014; 19: 517-527
        • Gu X.
        • Neubert S.
        • Stoll N.
        • et al.
        A New Method for the Indicator of Dynamic Scheduling in Life Science Laboratories Using Artificial Neural Networks.
        IEEE I2MTC. 2018; : 1-6
        • Neubert S.
        • Göde B.
        • Gu X.
        • et al.
        Potential of Laboratory Execution Systems (LESs) to Simplify the Application of Business Process Management Systems (BPMSs) in Laboratory Automation.
        SLAS Tech. 2017; 22: 206-216
        • Fleischer H.
        • Vorberg E.
        • Thurow K.
        • et al.
        Determination of Calcium and Phosphor in Bones Using Microwave Digestion and ICP-MS: Comparison of Manual and Automated Methods Using ICP-MS.
        IEEE 5th IMEKO TC19. 2014; : 94-99
        • Vorberg E.
        • Fleischer H.
        • Junginger S.
        • et al.
        Automated Sample Preparation for Mercury Analysis in Wood Materials.
        IET Sci. Meas. Technol. 2016; 10: 398-404
        • Fleischer H.
        • Ramani K.
        • Blitti K.
        • et al.
        Flexible Automation System for Determination of Elemental Composition of Incrustations in Clogged Biliary Endoprostheses Using ICP-MS.
        SLAS Tech. 2018; 23: 83-96
        • Fleischer H.
        • Roddelkopf T.
        • Stoll R.
        • et al.
        Automated Analytical Measurement System for Determination of Cholesterol in Pig Bile.
        IEEE I2MTC. 2019; : 1510-1515
        • Fleischer H.
        • Thurow K.
        Automation Solutions for Analytical Measurements: Concepts and Applications.
        Wiley-VCH, Weinheim2017