Why we're starting here
A few years ago we put a project out to tender. We knew what we wanted to automate, we had a rough idea of the solution, and we expected a reasonable quote back. What we got was a number that made our eyes water — and when we pushed back, the quote came back even higher. That was the moment we decided to do it ourselves.
What followed was one of the best decisions we've made. Not easy, not cheap, but far more achievable than any integrator will let on. This series exists because we want smaller businesses to know that.
So — where do you actually start?
Cell layout principles
Before you open any software, get something down on paper first. It doesn't need to be a technical drawing — a rough sketch of the machine you're trying to automate, where the robot needs to pick from, where it needs to place, and what's in between. That sketch becomes your brief and your brief becomes your cell.
Our first layouts were done in basic 2D drawing software — nothing complicated. We drew the floor plan to scale, added the standard components, and drew a circle representing the robot's reach. If everything the robot needed to interact with sat inside that circle, the layout worked. That's the simplest and most effective sanity check there is, and it costs nothing.
LibreCAD and DraftSight (free tier) are both more than capable at this stage. Don't overcomplicate it — a clean 2D floor plan with a reach circle drawn on it will tell you most of what you need to know before you touch anything more sophisticated.
Involve operators in the design process early
Before you finalise your layout, talk to the people who are going to use it. Operators often know things about a machine or a process that aren't written down anywhere — access points that need to be reached regularly, quirks of the existing process, consumables that need changing, adjustments that get made on every shift. If you design around those things from the start it'll save you significant rework later.
It also means operators feel ownership of the cell from day one. A team that helped design something is far more likely to look after it and far less likely to blame it when something goes wrong.
Keep the operator in mind above everything else
This is probably the most important principle and the one most often overlooked during the design phase. The temptation when building a robot cell is to optimise for the robot — cycle time, reach, efficiency. But the main cause of downtime in most cells isn't mechanical failure, it's operator error. And operator error is almost always a design problem, not a people problem.
The more complex a cell is to operate, the longer it takes operators to learn, the more mistakes get made during shift changeovers, and the more dependent you become on a small number of people who actually understand it. That's the opposite of what automation is supposed to achieve. Automation should make things easier — for everyone, not just the engineer who built it.
Any time spent during the build process that reduces operator input or simplifies the interface is time well spent. If you can eliminate a button press, do it. If you can make a status clearer on the HMI, do it. If you can design the cell so the operator only needs to do one thing to start a cycle, do it. The build might be slightly more complex as a result — that complexity belongs in the engineering, not in the operator's daily routine.
Keep it simple — especially on your first cell
The more complex the cell, the more that can go wrong during commissioning and the harder it is to maintain. Every time you're tempted to add something clever, ask whether a simpler solution achieves the same result.
This is even more important on your first cell. Make it as simple as possible, learn from it, and make improvements on the next one. A lot of complexity and cost in automation projects comes from trying to automate the last 10% of a process.
We had a project that required a machine to change inlet size for each order. We put it on a servo drive controlled from the HMI — it seemed like the right solution. We ended up with significant downtime because product jammed and caused impact damage to the servo. It cost a lot of money to fix. The solution? We put a handle on it. The operator winds it in and out for each order. Cost less than £100 and has never failed. Sometimes the right answer is the simple one.
Use standard components wherever possible
Conveyors, hydraulic lift tables, pneumatic grippers — buy off the shelf where you can. Custom fabrication costs time and money and introduces risk. Standard components have known dimensions, known lead times, and usually come with documentation.
Think about access
The robot needs to reach every position in the cell. So does the maintenance engineer. So does the operator. A cell that's efficient for the robot but impossible to work on safely is a problem you'll live with for years.
Think about guarding early
Safety guarding isn't something you bolt on at the end — it affects the footprint of the cell, the position of access doors, and where you locate the control panel. Think about it from day one.
Taking it further with RobotStudio
Once your 2D layout makes sense on paper, RobotStudio is the next step. It's ABB's simulation and layout software and the Basic version is completely free with no time limit. You don't need an ABB robot to use it — it's simply the best free tool available for validating a robot cell in 3D.
We'll be honest — we started out on KUKA and KUKA is still our main area of expertise. But the free resources ABB make available and the accessibility of RobotStudio put it ahead for anyone starting from scratch. The programming also seems more intuitive. That might ruffle a few feathers but it's our genuine experience.
The Premium version unlocks full offline programming and simulation at around £1,200–£1,500 per year — you don't need that yet. Basic is enough for layout and reach checking at this stage.
The robot library in RobotStudio is extensive. Drop in different robots, check reach, check payload, check interference contours. You're looking for the smallest robot that comfortably reaches every position with some margin — don't spec tight on reach.
At this stage you're answering three questions:
When you can answer yes, yes, and no — you've got your cell concept and you're ready for the next step.
Download RobotStudio freeA note on robot brands and buying options
We work primarily with KUKA and ABB. Both are excellent. Both will do the job. The decision at this stage often comes down to personal preference, pricing, local suppliers and support, or any robots you may already have in your facility.
We'd stress the importance of good local technical support — particularly on your first robot. In the event of a breakdown or a problem you simply can't solve, you may need to call someone in. Make sure that support exists before you commit.
If you're starting completely from scratch with no existing preference — RobotStudio is free and ABB's learning resources are excellent. That's a reasonable place to start.
Consider buying second hand
New robots are expensive. But the second hand market is worth serious consideration — particularly for a first cell where you're still learning. Online second hand suppliers carry a wide range of used robots in good condition, and both KUKA and ABB typically hold lists of ex-demo robots with very few running hours available for nearly half the cost of a new unit.
It's worth contacting both manufacturers directly and asking what they have available before committing to a new purchase.
Path2 generates production-ready programs for both KUKA KRL and ABB RAPID — so whichever direction you go, we've got you covered when you get to the programming stage.
Need the robot program too?
Path2 generates production-ready KUKA KRL and ABB RAPID palletising programs in seconds — layer patterns, separator sheets, interrupt handling, all included.
Launch the generator →