Validating robotic systems
Validating robotic systems is not much different from validating other types of pharmaceutical automation and much easier
than validating human interactions with a process, says Van Trieste. The validation process varies according to the product
and the process.
Robot vendors often test their products, the process, or the software, and run repeatability tests. "If the pharmaceutical
manufacturer wants to develop a new process using a robot, then they will have to qualify it themselves to their standards,"
says Carlisle. Qualification requirements vary depending on whether the process is simple (e.g., moving packages) or complex
(e.g., sample screening, analysis, and data maintenance).
Good manufacturing practice requires robots' software to be qualified, tested, and protected against unauthorized modifications.
Large robotic systems include two controllers—a programmable logic control and a robotic controller—and each must be validated,
Robots were once a source of contamination, but major manufacturers such as ABB and FANUC have adapted their products for
cleanroom use. For example, these vendors use special paint to reduce electrostatic force and particluate generation. Some
designs incorporate vacuum systems that are directly connected to the robots' joints to keep particles away from the process,
says Coste-Manière. Other robotic arms are wrapped in surgical-gown material to provide the same protection. Most cleanroom
robots have a brushless-motor design featuring components that roll instead of scraping together. This design does not generate
If a robot is intended to perform aseptic processing, the pharmaceutical manufacturer can wipe it down after it is received.
Robots now are made of materials that withstand cleaning agents used for instruments in pharmaceutical laboratories. One particular
FANUC robot resists decontamination with hydrogen peroxide. And many end effecters and actuators are autoclavable, says Barrett.
Robots are available with positive- and negative-pressure features, and they can be installed inside a laminar-flow hood,
says Carlisle. To prepare the robot, manufacturers perform a pumpdown process, the length of which varies according to the
level of cleanliness that the application requires.
The semiconductor industry has deployed robots for many years and has even stricter standards for particulates than the manufacturers
of sterile pharmaceuticals do, says Carlisle. Semiconductors are sensitive to certain chlorides and other molecules that robots
might emit. "It's possible that certain pharmaceutical processes might have that kind of requirement, but I haven't heard
of any to date," says Carlisle.
In general, pharmaceutical companies can easily find robots clean enough for cleanroom applications. "I see no increased regulatory
risk associated with the implementation of well designed, properly implemented, and validated robotic systems that operate
in a robust manner," says Van Trieste.
Incorporating robots into pharmaceutical processes has many advantages. Robots can perform operations three or four times
faster than humans and can function 24 hours per day, says Barrett. These characteristics make robots good at producing large
quantities of products efficiently. By performing simple, repetitive tasks, robots can free employees for creative work such
as developing new products.
"Robots can move to a precision of a fraction of the thickness of a sheet of paper," says Carlisle. Their accuracy in positioning,
delivering materials, and quantification exceeds humans' ability.
The typical robotic system uses fewer parts, has a quicker changeover process, requires less preventive maintenance, and entails
a lower annual operating cost than conventional equipment, says Langosch. A robot often can perform the functions of several
pieces of conventional equipment, thus reducing the footprint of the process in the manufacturing space.
Many types of automation cannot be modified easily and are limited to particular products or processes. These machines may
lose their utility if a given product is discontinued, but robots can be retooled and reprogrammed for new applications.
Variability can be greatly reduced through the use of robots, which can be programmed to perform exactly the same way every
time. This repeatability and reduced variability can lower the pharmaceutical industry's cost of making drugs and improve
product quality. If a robot is equipped with machine vision, it can provide "100% process verification and graceful error
recovery for high uptime," Langosch says.
Small contract packagers are just as likely to use robots as Big Pharma because robots provide the same advantages to both
kinds of companies, says Langosch. The only difference is that Big Pharma would likely achieve a faster return on its investment.
The pharmaceutical industry, in turn, is favorable to robotic systems because its processes generally require low forces,
take place in clean environments, and include predictable sets of operations. A robot has a lifetime of millions of cycles
in a pharmaceutical process because "there's nothing there to wear it out," says Barrett.
The increasing capabilities of computer systems are extending the applications of robotics as well. As Tallian observes, vision
systems can now perform color analysis with better resolution at high speeds, which provides a large amount of data that robots
can analyze quickly. "Ease-of-use software wouldn't be possible if we hadn't seen 10 times more power with our computers in
the last five or 10 years," says Tallian. "Computer processing power has been a huge launch of robotic systems. There has
been a lot of change in the mindset of the manufacturers, and more are now willing to consider a robotic automated system
than in the 1990s or early 2000s."
Burns says his company sees many robotic applications that include vision systems as a key component. "We see, for example,
many pick-and-place applications where you have random product and you have a vision system that detects the product position
and feeds that to the control system which then determines where to pick the part and where to drop it. And we see applications
where there are multiple robotics or cooperative robots picking product off a single line."