Watch Out, a Montreal-based company with operations in Switzerland, France, and Canada, develops container-sized manufacturing cells that operate with minimal human intervention, addressing critical labor shortages while enabling domestic production.
This rendering shows Watch Out’s vision for integrating multiple autonomous workcells in a manufacturing facility requiring minimal human intervention. Image: Watch Out
More than 10 years ago, Sébastien Laporte had a vision based on the realization that it was becoming increasingly difficult to find skilled labor, despite technological advances in CNC machines. While machine builders tried to help operators make better decisions with the latest Industry 4.0 technologies, Laporte went in the opposite direction. He asserted that operators were so scarce, and would continue to be scarce, that he needed to build something completely autonomous.
This vision required two fundamental innovations: digitalizing the accumulated knowledge of experienced operators into the machine itself, and achieving 100% automation.
Laporte partnered with Frédéric Perret and launched Watch Out in 2012. They started out automating inspection for Rolex watches, then scaled their algorithm into a microfactory for precision machining, and are now in production for the aerospace industry.
“The heart and the brain of the system are not the mechanics. The mechanics are just the support, moving the tools. It’s the database, the digital system,” said Bernard Mariette, executive chairman at Watch Out. “Number one, you capture the data with all the sensors — optical, electrical, the whole lot — and they all have been developed by us. Number two, we organized the data system to be very frugal because we’re working in microns and nanoseconds. Our scales are very, very small and very, very fast. So you cannot have millions of data that take hours to analyze.”
The idea is not to reinvent the wheel or accumulate data that doesn’t add any value to the system. Mariette argues that deep learning isn’t necessary for every situation, problem, or decision, and therefore doesn’t require collecting the millions of data it would otherwise ingest and store. Instead, as Mariette explained, “you just have to reinvent what you don’t know.” In Watch Out’s system, if the data received doesn’t fit the algorithm, it is sent to a deep learning system, which runs the data and finds a solution. Then, the solution is put into action to machine a piece and check it. After about 100 pieces, the deep learning outcome becomes an algorithm.
“It’s fully autonomous because we catch that at the beginning, and we control 100% of the data, which goes back into the data model. It’s a data-centric system,” said Mariette. “We try to make the mathematical model, the digital model, exactly the same as the real model of the parts.”
The AI-driven microfactory monitors its handling, machining, and inspection modules in real time. Image: Watch Out
As Olivier Chéret, Watch Out’s chief strategy and growth officer, explained, the first microfactory the team developed is for precision turned parts, including aerospace fasteners, and consists of three process modules: handling, machining, and inspection.
“When you push the button, if you will, the cell itself is performing a ton of checks autonomously, including its geometrical checks. The reason we can kick them off so fast in a new factory is because the cell is going to check that the positioning of its different parts is aligned with its model. There is no human intervention required to kick off the operations,” said Chéret. “Then, when you give a new step file for the cell to produce, the AI software is going to write the machining program autonomously based on its experience of other parts and the expert system.”
From there, an “AI tower” communicates with the microfactories and operators so that operators bring the right tools to the right workcells.
“Typically, we put tags on the tools so that when the operator is bringing a new tool to the microfactory, the microfactory digitally recognizes the tool that has been loaded in the cell. It’s an AI-led decision process to make sure that we control the end-to-end,” said Chéret.
Once the microfactory is ready to start production, the operator places the tool in the handling module, gives the green light, and the cell automatically checks, handles, and machines the part.
“One of the big innovations is, thanks to the optical cameras we have, we know in real time the positioning of the parts and the positioning of the tool,” said Chéret. “In real time, the software is always able to match the tip of the tool with the part, and check every new part that’s coming into the machining cell. And because we have this vast amount of data and cameras monitoring the tool, the software is able to adapt its machining program to the state of the tool. As the tool wears, we have numerous rules to adjust the machining parameters. So, we are regaining a ton of efficiency from that.”
Once the part is machined, it returns to the handling module and then to inspection, which determines whether the part is compliant or not. Feedback from the inspection module is used to adjust parameters for subsequent parts, if required.
“The reality is, because we have so many sensors throughout the process, we already have clues of what’s going on when the part is machined. We don’t need the inspection to know if something is wrong. We might know much earlier. That’s the richness of this ecosystem, and the possibilities are quite limitless,” said Chéret.
The microfactory continuously monitors part and tool positioning and automatically adjusts its machining parameters in response to tool wear. Image: Watch Out
What was once a vision has become limitless due to the convergence of multiple advanced technologies that various companies and industries have developed to solve different engineering problems over the past decade, and AI is the pinnacle solution.
“Until now, we’ve tried to hide for two reasons. One, everybody was telling our founder, Sébastien, it’s just impossible. It will not work. The reason it’s possible is because it’s at the cross of many technologies — optical, electrical, algorithm, and AI. Without AI, we would not have been able to do it, and it’s improving every day,” said Mariette. “The technology to measure is also, from a mechanical point of view and a digital point of view, totally new. We can position the part and the tool anywhere. If the machine is not flat, it doesn’t matter. It autocontrols itself, not mechanically, digitally.”
Mariette explained how painful it has been to watch as others try to reinvent instead of innovate. Now, with AI, more stakeholders are getting on board with this new concept that is autonomous, economical, and frugal in terms of data and physical space.
“Our microfactory, the full system, is nearly the size you will find of a normal CNC machine. It’s ecological and economical,” said Mariette.
Watch Out estimates that the microfactory has a 30 to 50% lower carbon footprint than typical machines. Real-time autonomous monitoring and adjustments throughout the entire process reduce scrap. The small size also requires less hardware, less building space, and fewer workers driving cars to the factory, among other benefits. Additionally, it is easily transportable, reducing the costs and carbon footprint associated with delivery and installation.
“When you buy a traditional machine, it goes in many containers, and you cannot drive it with a truck. You need a special convoy, and in many cases, you need to change the big doors you have for a normal truck; you need to cut them. That’s what we’ve done. We had a factory with doors for a normal truck, but the machine was too big, so we had to cut the wall,” said Mariette. “Our microfactory fits in a container — in one container. We can install it in about three hours. A normal state-of-the-art machine is between three days and a week.”
Watch Out’s microfactories are currently producing parts in Europe and for LISI Group, which ranks third worldwide in the creation of aerospace fasteners and assembly components.
“It’s very flexible, totally autonomous. It’s a way of cutting costs and waste because you don’t have downtime. It’s fully data-centric, and you don’t need to stop the machine. The machine can change its tools, models, everything,” said Mariette. “You have 100% of the information about the part that comes out of the machine, and there’s no human intervention.”
Watch Out’s autonomous microfactories represent more than an incremental improvement in manufacturing automation. By combining advanced sensing, AI-driven decision-making, and complete process integration in a transportable package, the technology addresses multiple industry challenges simultaneously: skilled labor shortages, supply chain vulnerability, and the need for flexible, responsive manufacturing capabilities.
For aerospace manufacturers facing increasing pressure to reshore production while managing costs and quality requirements, such systems could provide a viable path forward. The technology’s ability to operate with minimal skilled labor makes domestic manufacturing economically feasible even in high-wage markets, while the autonomous operation ensures consistent quality and productivity.
Watch Out
watchout-corp.com
Filed Under: Aerospace + defense, Factory automation, Digital manufacturing, AI • machine learning
