Abstract
The technology and techniques used when automating laboratory activities have been developed and documented for more than 40 years. Work performed under the subject of “laboratory automation” has progressed from pioneering achievements in data acquisition and instrument control to multicomponent, fully integrated systems resembling manufacturing plants. If the field is to move forward, we need to organize the practices of those applying automation and computing technologies to laboratory activities, and to formulate a course of study. This JALA Guest Editorial seeks to initiate a dialog on the definition and development of the field of “Laboratory Automation Engineering.”
Introduction
There is widespread work underway on the application of automation technologies in diverse scientific disciplines. Chemistry, high-throughput screening, physics, quality control, electronics, oceanography, and materials testing are just a few examples. Moreover, this work has been occurring for many years. For the past 10 years, the Association for Laboratory Automation's (ALA's) annual LabAutomation conference has had presentations, papers, and short courses on the subject. Yet, in spite of the diversity of applications, one finds many characteristics in common. In fact, there are so many common traits and skills that we can say with confidence that a distinct field of Laboratory Automation Engineering (LAE) has emerged.
The following common characteristics abound.
LAE is practiced in facilities where the focus is scientific in nature. Indeed, laboratories are facilities that apply scientific methods to glean new insights.
The end product of any laboratory is information. Consequently, LAE efforts do not generally produce material products; they produce information as well—raw data that leads to new knowledge.
LAE efforts also seek to improve laboratory productivity. Whether the ultimate goal is to amplify the ability of a scientist to generate results, or free him or her from tedious work to focus on more pressing issues, frequently the end result is a productivity improvement.
Regardless of the scientific application, the practice of laboratory automation often draws upon tools such as robotics, liquid handling, computing, and knowledge from fields such as mechanical and electrical engineering.
Most importantly, the practice of LAE requires that the practitioners be knowledgeable in the scientific discipline in which technologies are being applied as well as the automation technologies themselves. It is important to be able to “speak the language” of the scientists benefiting from the result. The LAE functions in part as a translator, converting a set of scientific needs into a project plan and ultimately into a functioning system. In some respects, this is similar to the software engineering process.
As an endeavor to solve problems through an application of science and technologies such as robotics, networking and computing, and process analysis and improvement, LAE easily fits any accepted definition of an engineering discipline—the art or science of applying scientific knowledge to practical problems.
LAE is distinct from other forms of automation engineering found in areas such as manufacturing and assembly.
Most physical automation engineering efforts culminate in the production of a material product. As mentioned, LAE is unique in that it is the only physical automation field that produces no material products. Automation of sample preparation is an intermediate step to analysis and data production. The primary product of any LAE system is new information.
As compared to most automated facilities, such as manufacturing plants, laboratory automation systems tend to be small in scale. Complete laboratory automation systems are commonly found operating on top of a laboratory bench. Any practicing scientist will attest to the limited space offered by a laboratory benchtop.
Today, laboratory automation systems are commonly created as replacement processes. Manual techniques, once they mature, are replaced with automated systems. Economics and the need to improve or enable the process drive the replacement.
Emergence of Laboratory Automation
Laboratory automation was initiated by scientists with a need to do things differently. 1 Whether it was to improve the efficiency of their work, to substitute intelligent control and data acquisition systems for manual labor, or because it was the best or only way to implement an experimental system, automation moved into the laboratory. 2 Those doing the work had to be well versed in instrumentation, data acquisition, programming, and other technical aspects, as well as their primary field of science.
The first generation of laboratory automation systems focused on interfacing instruments, sensors, and controllers with early computers. As the electronic interfaces became more robust and reliable, the focus turned to the direct computer control of experiments and automated sample handling with small-scale robotics. As the quantity of experimental data increased, the need for information technology (IT) quickly became evident. Laboratory Information Management Systems (LIMS) became popular as a tool for collecting and organizing the data.
As time progressed, these core ideas evolved in new and interesting ways. Automated systems became larger, more instruments were integrated, and they moved from bench-tops to tables and dedicated automation laboratories. Later, as microfluidics and lab-on-a-chip technologies emerged, automated systems returned to smaller, benchtop formats. Electronic interfaces became even more robust and reliable, with faster and faster data rates. Modern software techniques—graphical user interfaces, object oriented programming, etc.—were quickly adopted and applied to laboratory automation systems. The diversity of data storage systems exploded, moving from general purpose LIMS to application-specific software applications. More recently, a complete replacement of the standard paper notebook with a fully electronic version has taken hold. And even in the face of dramatic technological advancements, the core purpose of the LAE field remains the same.
The Role of A Laboratory Automation Engineer
As the discipline of LAE has emerged, so has the role of its practitioner, the Laboratory Automation Engineer.
With its heavy focus on computing and IT, many naive organizations, over time, embedded their laboratory automation efforts within their IT groups. This decision has shown, time and again, to lead to poor outcomes.
It is common to find corporate IT departments at odds with those using computers in laboratory settings. From scientists, it is common to hear, “IT doesn't understand our needs,” and from IT, “we're responsible for all corporate computing.” Both sides of the debate demonstrate important valid points. IT departments are responsible for organization-wide computing; labs are part of the organization. Unfortunately, it is not uncommon for IT departments to be ignorant of the computing needs of scientists. The needs of computing on the benchtop are much different from those of the desktop. Because desktop users tend to be the larger, more vocal group within an organization, IT professionals often neglect to learn the language and understand the needs of the scientists. IT-speak and science-speak address two completely disjointed worlds. Specialized software packages used in laboratories require specialized support from IT departments. It is the responsibility of IT to be concerned with the support of these programs, and the effect they may have on the security of the organization's information. One primary role of an LAE is to bridge the gap that exists between science and IT departments.
Automation has taken hold in nearly all laboratory-based sciences. Decades of effort in applying technologies to laboratory problems resulted in a large assortment of automation products. A problem with this wide variety of products is a lack of general standards. In spite of outstanding efforts in the field, 3 the adoption of standards has not taken hold in laboratory automation. Individual laboratory tasks, also known as laboratory unit operations (LUO), have been automated with sophisticated instrumentation, but their integration is limited to vendor-specific approaches.
Another important role of an LAE is to take responsibility for ensuring that the LUOs required to automate an experimental procedure are reliably integrated into a robust system, and that they successfully perform the desired experimental task. Good LAEs are masters of modular design principles and systems engineering. Quality automated systems are designed to be gracefully upgraded rather than replaced, and less dependent upon proprietary solutions. A sound systems engineering background on the part of the LAE is required. The ability to build isolated products simply is not enough.
Today, electronic communication technologies are well developed and readily available, thanks in large part to the development and adoption of communication standards. Data storage is inexpensive, and there is an abundance of computing power available. Experimental data analysis does not need to be performed at the instrument. Data can be sent to a central data processing system for cataloging, storage, analysis, and reporting, with raw data being readily available for additional analysis as new techniques are developed. The islands on the automation map are connected with bidirectional communication channels. As a result, LAEs are trained to be adept at electronic communication technologies. They are stewards of a scientific organization's most important intellectual property—its data—and are well versed in maintaining its security.
The Current State of Lae
Laboratory automation practitioners have been successful for several decades, so why do we need a change? Among the many successes have been equally many projects that ranged from less-than-complete to outright failure. The potential for laboratory automation to assist or even revolutionize laboratory activities is too great to have project outcomes be hit-or-miss. Even projects that appear to be initial successes may lead to long-term problems with an uninformed choice of technologies.
In practice, many people working in LAE are scientists who have demonstrated a knack for technology. A tedious data processing task was answered with a Microsoft Excel script that handled the task in a blink of an eye. An arduous pipetting task resulted in a convincing argument for the purchase of a liquid-handling robot. The unruly contents of a laboratory freezer yielded bar coded containers and a Microsoft Access database to track them all. Consistent demonstration of technical prowess led first to the scientist being dubbed “the technology person” of the lab, and then to the formation of a dedicated group to manage technology in the laboratory. Unfortunately, because of a lack of formal training or the benefit of decades of accumulated knowledge and best practices, many automation projects performed by these highly skilled scientists result in failure. Too often, the hard-learned lessons of their predecessors are learned again by these homegrown LAEs. The reason for this is a lack of a centralized effort to accumulate and organize these lessons into a coordinated body of knowledge.
This is not to say the emerging field of LAE is any different than other, more established fields. In fact, LAE has moved along a familiar developmental path traveled by other fields that are now well established. In commercial chemical synthesis, aerospace, computer, civil engineering, and other fields, progress was made first by individuals, and then small groups. As the need grew to apply technology on a broader scale, training became formalized, as did methods of design and development. The methodology shifted from an art to true engineering. The end result is the ability to create buildings, bridges, aircraft, etc. on a scale that independent groups could not hope to achieve.
Benefits of Formalizing Lae
Now is the time to formalize the field of LAE. Such an initiative would have many benefits to the individual, the practicing organization, and the field itself.
Benefits to the Individual a
The creation of a thorough and systematic education in a broadly practiced field will improve an individual's ability to create successful results. The individual will benefit from the accumulated knowl edge of past activities—what has worked and what has failed. The individual will have the opportunity to demonstrate evidence of knowledge and training through a degree or certification. The individual will gain a sense of identity and community along with others in the field.
Benefits to the Organization
An organization will gain a basis for evaluating an employee's credentials and ability to work on laboratory automation projects. It also will obtain a basis for em ployee evaluation. Formalizing the field of study will obviate the need for on-the-job training, thereby reducing expenses. Projects will be implemented more quickly with a high expectation of success. As the field becomes established, LAEs can turn their at tention more frequently to the design of new laboratories with automation as part of the initial blueprint rather than use it to replace manual procedures. This will reduce the cost and improve the efficiency of laboratory operations.
Benefits to the Field of Laboratory Automation
A foundation of documented knowledge will be created that LAEs can use to learn and to improve the effectiveness of the profession. A community of people will emerge to drive the organized development of systems and technologies applied to the advancement of the practice of science, enabling the creation of reuseable resources that will eliminate the practice of repeatedly reinventing the same solutions to common problems. The field will take ownership of research of new technologies that significantly improve a scientist's ability to carry out science. A community of like-minded individuals will be formalized in which they can discuss, and where appropriate, develop positions on key issues in the field, such as the impact of regulatory requirements and standards development, and publishing position papers and guidelines on those points as warranted.
I would like to thank Mark Russo, executive editor of the JALA, for his comments and in this section, in particular, for his insights and the material he contributed.
Lae Skills
The diversity of the field of LAE requires that the skills possessed by an LAE be similarly diverse. This must be reflected in the body of knowledge and training that ultimately defines the field. Skills needed range from computer science and programming, engineering fundamentals and processes, requirements-driven project management, regulatory issues, and an understanding of core science.
Computer Science and Programming
In the mid-1960s “programming” instruments was a mechanical problem handled with needle-nosed pliers, screw drivers, and a stopwatch for adjusting cams and micro-switches. Process chromatographs made by Mine Safety Appliances and Fisher Control used cams and microswitches to control sampling and back-flush valves.
Today, automation systems are built around computer components. A strong background in computer science and programming is necessary to be successful, because all parts of an automated system demand some knowledge of programming.
Instruments are programmed to perform individual tasks.
Data acquisition devices are programmed to acquire and process data.
System controllers are programmed to communicate with instruments and coordinate their activities.
Computers are programmed to transmit data to central data stores.
Database management systems are programmed to store data in a normalized format.
Data analysis systems are programmed to collect, analyze, and report experimental data.
Engineering
Broad fundamental engineering skill requirements are necessary. These include knowledge of electronics, signal conditioning, motion control, robotics, heat and mass transfer fundamentals, control theory, machining and fabrication, and more.
Knowledge of engineering processes also is critical. As part of the introductory comments for Zymark Corporation's robotics courses, Frank Zenie (a Zymark Corporation founder) usually made the statement that “you can only automate a process, you can't automate a thing.” Recognizing that a process exists and that it can be automated is the initial step in defining a project. General Systems Theory b provides the tools for documenting a process as well as the triggers and the state changes that occur as the process develops. General Systems Theory is particularly useful when working with integrated multicomponent systems.
One part of the problem is to identify a process and document it. Another is judging whether or not the process can be automated reliably. Laboratories and laboratory equipment are generally designed for people. Using that same equipment in automated systems, including robotics, may require a significant amount of reengineering to make it reliable. An estimate of the cost of this reengineering requirement cannot be overlooked.
Process engineering also includes statistical process control, statistical quality control, and productivity analysis. Laboratory robot systems can be thought of as small-scale manufacturing systems. As laboratory activities become automated and individual systems are integrated, these systems will behave like manufacturing processes and can benefit from similar approaches to control.
Project Management
The LAE must be skilled in engineering project management tasks, including budgets, schedules, and documentation common to projects. These are well documented and well understood in the general engineering field.
The project management process is requirements-driven, beginning with an understanding of user needs, which are detailed in a document known as the User Requirements Specification. A thorough understanding of user requirements is followed by a list of the requirements for how the ultimate system will function, documented in the Functional Requirements Specification. The automation system is then designed. All design documents are collected in the design specification. Prior to completion, a plan for testing is created, and may include Factory Acceptance Test plans and Site Acceptance Test plans. User and system documentation complete a project.
Part of the project planning effort is to ensure that people understand project goals and benefits, as well as cost, schedules, risks, etc. These points and the ramifications of changes must be clearly communicated and understood. A lack of clear communications is often the basis for misdirection, delays, and frustration with projects. Communication is a two-way street. The LAE needs the ability to present project goals in clear and concise terms, and to listen to and understand the needs of scientists working in the labs.
See http://pespmcl.vub.ac.be/SYSTHEOR.html for an introduction. In particular, review the work of George Klir: General Systems Theory & Facets of Systems Science (IFSR International Series on Systems Science and Engineering).
After an automation system is developed, it should be evaluated against the performance of an existing process, or performance expectation, to show that it meets design criteria. If replacing an existing process, a seamless change-over procedure is designed to migrate from one method of operation to another.
Regulatory Issues
No matter what the industry, meeting government regulatory requirements is a consideration. An LAE must be well versed in regulations.
Companies are under increasing pressure from regulatory agencies to have their internal practices conform to standards. This is no longer an ISO, FDA, EPA, or similar agency issue that is focused on manufacturing systems, but an organization-wide set of activities encompassing financial and accounting practices, human resources, and more.
Primarily, regulations require a good system design process. For example, the requirements for the validation of a system call for the designer to show that the system works, that it is supportable, well documented, uses equipment that is suited to the designer's purpose, and is obtained from reliable sources. “Tinkering” a system together is no longer an accepted practice. It may work for prototyping, but not for production. The purpose of regulations is to ensure that systems put into use can stand up long term, and fit a well-defined and documented need. In short, regulations ensure that laboratory automation systems are well engineered.
Science
Once again, a thorough understanding of the science in which automation is being applied is crucial for an LAE. It is up to the LAE to understand the science and the scientists to translate their needs into requirements, and ultimately into working systems.
Change Management
When replacing laboratory processes performed by people with processes performed by machines, special considerations come into play.
Moving from an existing manual process to an automated process often may mean that some job functions will change. For some, it may require some minor adjustments, while for others, it may result in dramatic change—and change raises anxiety levels. If those working in the laboratory believe that an automation project will cause significant change, they may inhibit progress toward the completion of a project. Thus, possessing the necessary “people skills” to manage these inevitable changes is important for an LAE.
How Do We Proceed From Here?
The first step is for a community of leaders in the field to develop a comprehensive LAE curriculum. This will require support from both organizations like ALA and other industry stakeholders. A university cannot be expected to create such a program, even one that can be assembled from existing courses, unless it has some assurance that its students will benefit.
Curriculum contents must address the topics and skills presented in this JALA Guest Editorial, and possibly additional topics. Furthermore, it must result from candid conversations among members of the laboratory automation community. It is important that these conversations are inclusive, not exclusive, open to participation by any and all interested members. James Sterling's article 4 is one example of how to move forward.
While a university program can address the long-term needs of new LAE students, it may not address the needs of those practitioners already working in the field. Additional learning opportunities should be provided so that existing practitioners can augment their backgrounds, and possibly earn a degree or certificate. One way to satisfy this demand is to expand the short course program already offered by ALA. Another possibility is the development of new certification programs similar to those used in the field of computer science. Such a program could be sponsored by ALA with course material drawn from its existing and newly developed short courses, and organized into an “Institute for Laboratory Automation” with options such as a summer program.
Once a curriculum has been defined by the community, the next step is to outfit each part of the curriculum with supporting materials. This can include compilations of relevant texts, knowledge bases, etc. Those working in the field can be encouraged to contribute material. Publications like the Journal of the Association for Laboratory Automation can invite authors to align their articles to the curriculum, and make use of the JALA Tutorial manuscript category as a means for building the foundation of the LAE literature.
Concluding Remarks
LAE is still in the early phases of its development. Some may say that significant progress has been made, but by comparison to the rate of development of automation and information technologies in other areas, progress has been slow and incremental.
It is necessary that the user community embrace the establishment and development of a discipline focused on envisioning, creating, organizing, and improving the tools and techniques of laboratory automation. That, in turn, will allow us to realize the promise that proponents of laboratory automation have long held: enabling people to do better science.
An LAE's employment opportunities will be driven by market demand. However, the skill sets should be transferable. Just as computer science professionals have the flexibility to apply their skills in different areas, LAEs should enjoy the same flexibility to move from one scientific discipline to another, the only difference being a knowledge of the underlying science.