David McMillan is an automation application engineer from the Detroit, Michigan region with over 20 years of experience in automotive, aerospace and general industry manufacturing. Opinions expressed are his own and may not represent those of his customers or employers.
Investing in a robot, especially for a small or medium-size enterprise, is a big project. While it may be tempting to start at the bottom line, the first consideration should not be cost. At least, not in isolation. I've seen customers burned by just buying "the cheapest robot." Granted, this is a lot worse when buying second-hand machines, but it can happen with brand-new units as well.
The first consideration should be capability. Can this robot do the job you're buying it for? A very close second should be support: what are the post-purchase costs (in money, time, and aggravation) to get this robot set up to do the specific job you're buying it for? (and, later, to keep it running).
Obviously, #1 and #2 are not completely separable. Sometimes, the cheapest robot turns out to be the most expensive for option packages, tech support, and spare parts.
Tip 1: Find the Right Integrator
For an inexperienced customer, solving both questions is often a matter of finding a good integrator and partnering with them. Of course, this can be like choosing a general contractor to build your house -- a good one is great to work with, a bad one can end up costing you more than if you tried doing it yourself.
A good integrator should provide:
- good local support,
- capable of handling multiple robot and PLC brands,
- able to do mechanical, electrical, and pneumatic design & build,
- willing to offer training.
- Full release of source code, mechanical drawings, and electrical diagrams are a must.
The more exacting an integrator is about getting process details and site surveys before buying hardware, the better, usually. Some integrators are so accustomed to working with "experienced" customers (Big-3 auto companies being the prime example) that they may not realize just how much hand-holding an inexperienced customer will need.
Tip 2: Which Robot is Right?
With the rise of cheaper, lighter, safer collaborative robots like the Universal Robots and Rethink machines, some of the need for high-end integration is reduced. But there can still be "gotchas", and there are a lot of jobs these cobots are not really suited for.
With or without an integrator, a customer new to robots should seriously consider starting out with an ‘easy’ task, rather than trying to roboticize their most complex process. Keep in mind, jobs that are easy for humans (and thus look easy for robots) can often turn out to have hidden complexities that make automation difficult, expensive and even impractical. Often, the best way to automate a process also involves making adjustments to the manual processes upstream and downstream of the automation or adjusting part tolerances -- pieces that people can fit together "with just a bit of fiddling" probably need their tolerances tightened to make automating that task successful.
It's also important to realize that robots are much more repeatable than accurate. If you present a part to a robot in exactly the same spot every time, the robot can do surprisingly precise work on it. But if the robot has to try to adjust to parts that are not presented repeatably, positional error accumulates rapidly.
It's also important to realize that automation generally does not have a quick ROI -- it's much more a long-tail item. There are exceptions to this, but a good piece of automation generally has a high startup cost and pays for itself over several years.
Tip 3: Plan Your Project for Murphy’s Law
For small shops, it's important to have fallback plans. Sometimes, a robot can fail and need a spare part with a lead time measured in days or even weeks. Even simple, easy-to-fix failures can cause severe downtime if your support provider has all their personnel out on other calls. So, if a broken robot can block your entire production line, you want to think about having the means to get the robot out of the way and "limp" along in the interim.
Well-designed and well-maintained automation rarely fails, but the key word is "well." A lot of Big-3 automakers tend to beat their robots to death, but that's because they often intend to replace them every 5 years or so (partly for tax reasons). Plus, they often have lots of spare parts, and "retired" automation to scavenge spares from. A small-timer operator probably won’t have that luxury, so an ounce of prevention is worth a pound of cure.
In the same vein, record keeping matters. That means:
- keep all the drawings and details secure, where you can find them years after the fact.
- Take regular backups of any programs, and keep them secure as well.
I can't count the number of times I've been called in to a new customer because their 10-year-old (or more) automation broke down, their original tech support (in-house or integrator) is dead/retired/out of business, and can't find any of their archived information ("The electrical diagrams? I think old Joe had those in his toolbox, but he retired a few years ago.")
Finally, for first-time buyers of automation: plan for difficulty. Not failure, but expect problems, pain points, and lots of tripping over things that "everybody knows," but you don't... yet. Treat your first foray into automation as a loss leader, a chance to learn how to do the next one right. Even a painful and difficult integration can (and often eventually will) produce a system that will pay for itself, but you can expect the ROI to be more extended. You'll need to learn things about buying automation, about what processes are easily automatable and which aren't, your maintenance staff will need to learn new skills, and you'll need to learn how to translate the knowledge your production staff have into programs the automation can use.
The best results come not from replacing people with automation, but supplementing them. The people who know how to build the widget and can tell what's right or wrong with it at a glance will often be the people you want overseeing the robots doing that process. Leverage the robots maximally for what people aren't good at (or tasks that are dull, dirty, or dangerous) and leverage the people for what they do best: making judgements, noticing subtle things, figuring out the ‘why’ behind what's going on. The most successful automation is often less about reducing headcount, and more about about increasing productivity per head.
For more on this topic, check out the article How to Pick, Pitch and Purchase Your First Robot.