FEATURE Robotics
BEFORE COMMITMENT BUILD CONFIDENCE
Nick Thompson, co-founder and CEO of OLO Robotics, discusses the stumbling blocks to robot implementation, and explains how AI has its role to play in simulation capital investment is made
A
sk any innovation team in manufacturing whether robotics is on their radar and you’ll get an enthusiastic ‘yes’. Ask them why
they haven’t done more with it, and you’ll hear the same three things nearly every time. It’s too expensive, too complicated, and takes too long to get anything working. Those aren’t excuses, they’re an accurate description of how traditional robotics implementation has always worked. I came into robotics from software. After spending twenty-odd years building development teams and custom software businesses, I started getting interested in what was happening at the intersection of software and physical automation. What struck me wasn’t how advanced robotics remained to deploy.
The problem isn’t the robots The standard framework for programming industrial robots is ROS2 (the Robot Operating System). It’s powerful, open source, the foundation the industry has built on. It’s also extraordinarily hard to learn. You essentially need to know everything about ROS2 before you can make a meaningful start. Most full-stack developers, people who are excellent at building software, are not going to spend months acquiring that foundation just to prototype a use case. The in the domain of specialist roboticists, and specialist roboticists are expensive and hard
When things become inordinately complicated, they become the preserve of specialists, and when that happens, costs go up and lead times stretch out. Manufacturers currently face two options: hire a roboticist (and a sizeable implementation team) or buy a proprietary robot and use whatever tools
10 March 2026 |
upfront expenditure. Both lock you into decisions before you’ve validated whether the automation works the way you need it to. There’s a lot of noise at the moment about AI
things. Manufacturing needs deterministic every time, without ‘fuzziness’. A drilling robot needs to hit a millimetre-accurate point. A pick-and-place arm needs to move to an exact coordinate. These tasks must be repeatable to nanometre tolerances, and that requirement is not going away.
The industry is moving from ‘robot
programmer’ to ‘automation platform user’ and that shift matters enormously for manufacturers. The emerging model looks more like software PhDs, uses simulation and accessible tooling to prototype use cases and validate feasibility before Platforms built on ROS2, but with the complexity abstracted away, are making this possible. The key capability is simulation. Experiment with industrial robot arms before you own them, verify code visually before it runs on real hardware, catch problems before they become expensive or dangerous. LLM-based code generation means developers describe what they want, the AI writes working code, and they verify it in simulation. Robotics prototyping becomes something an innovation team can actually do.
NVIDIA’s ‘ChatGPT moment for robotics’ models on vision and movement data so robots can learn generalised behaviour. Genuinely fascinating, and it will matter eventually. But it isn’t ready for industrial deployment yet, and it wouldn’t replace deterministic precision even if it were. Where AI helps right now is in vision. A AI determines what it is and where it is, and deterministic code takes over. Move the end cog, execute the placement. Intelligent perception feeding precise execution and achievable today, for mainstream developers. That’s where the interesting work is happening.
grower wanting to automate crop monitoring. Historically, this meant hiring a roboticist, engaging a systems integrator, and spending months on a bespoke build with no guarantee of the outcome. The same use case, approached with a domain expert with software skills can explore without all of that apparatus. The economics and The barriers haven’t completely disappeared, safety considerations remain, as they should. But the ability to experiment cheaply, validate early, new. For manufacturers who’ve watched robotics sit perpetually on the ‘future plans’ list, that change is worth paying attention to.
automationmagazine.co.uk
BUILD CONFIDENCE BEFORE COMMITMENT
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40
