It is certainly true that an unconstrained robot takes up a fair amount of spce. Adding hardware motion constraints can reduce that space, but still, mostly, a robot takes up more space than dedicated automation. So that consideration must certainly be a part of the tradeoff calculations.
A two or three axis automation system dedicated to a particular process will almost always be more effective than a universal robot in that same application. There is no question about that. BUT a robot is a flexible device, while a dedicated system designed for a specific task is not. If the task changes a bit the robot only needs a program change, while the dedicated automation system may need a number of hardware modifications. The tradoff between optimization and flexibility is very real and usually recognized.
There is a company that offers a robotic test systen for automotive seats, and it costs more than any of the machines that could do any one of the 8 or nine different tests that are done on seats. BUT it is far less expensive than the collection of different testing machines needed to do all of those tests separately, and it takes up much less space.
So really, there are many applications where a dedicated automation system with fewer axis is thye only wise choice, while there are other applications where the flexibility of a robot system is the only smart choice available. The two are different and have different applications, similar to steak and athletic shoes. Each may be the best choice for a different application.
I wholeheartedly agree. Parallel systems do have their limitations, but their unique capabilities add a lot of options when designing an automated system. For example, while an Hbot is slightly larger than it's motion envelope it is nearly the same size and shape; if it can be placed over or under the motion envelop it can have the smallest impact on footprint of any system. But an Hbot can't deal with "snaking" or inserting nearly as well as a SCARA robot, or come close to handing the huge number of motion axes of a conventional articulated robot.
The big drawback to parallel-kinematic robots: they generall have to be as big as their motion envelopes. Oh, there are exceptions, but in general, if you want to "snake" something into a tight spot from a distance, an articulated serial-kinematic arm is still the way to go.
I see a lot of people trying to use 6+ axis industrial robots as "CNC" machines these days. The articulated arms are cheaper, and have a much greater motion envelope for a much smaller footprint (and price tag). But they simply can't match a gantrybot or "real" CNC machine for strength, rigidity, or accuracy.
As always, it's less about choosing the "best" robot, and more about choosing the tool appropriate to the task. A small lab with a poor equipment budget might well be better served by an articulated-arm robot that can be easily re- or multi-tasked for only the cost of a new end-of-arm tool.
The number of robots in medical assembly has gone through the roof. At the Medical Design & Manufacturing Show in Chicago today, the number of robotic systems (inlcuding Festo's) was incredible. The makeup of the exhibitors is starting to resmeble the packaging indy show, Pack Expo.
Robots and automation systems are becoming so complex, it's hard to tell one from the other. Where does the robot end and the automation system begin? Sometime sit seems the whole automation operation is one big robot.