Digitalisation for a Sustainable Manufacturing: the tool I wanted at TESLA

November 24th 2022

Catch & Do

How Luc LeroyFormer head of Core Automation Engineering at Tesla and CEO and co-founder of Full Speed Automation has developed a tool enabling Sustainable Manufacturing

Play Video

Since the manufacturing industry is one of the largest contributors to greenhouse gas emissions worldwide, sustainability has become an important objective for many companies producing goods. The regulatory constraints of ESG goals in Europe works as a driver for change, but more than just strengthening a brand, it has become a strategic element for growth and competitiveness.

For Luc Leroy, sustainable manufacturing is deeply rooted in the DNA of Full Speed Automation, the company he founded in 2020. Luc joined us for a “Catch and Do” session where we discussed how the shop floor digitalisation can contribute to this effort by increasing operational efficiency, reducing costs and waste. The carbon footprint of goods produced far from the consumers is also a critical element in the equation. The no code software of Full Speed Automation empowers the users and delivers an extra level of agility to enable a competitive local production.

Luc, could you introduce yourself and how you founded Full Speed Automation ?

I started my career with a regular job actually in the 90s until 98, and it was great starting my first company with two friends in 98 and one after another we were finally successful in 2004 when our company was acquired by Adobe which is what brought me to the US. I spent indeed a bit more than 10 years with Adobe, I was ultimately responsible for the applications for acrobat and reader as well as PLM. I was very interested by manufacturing and I realized that there was a gap between the world of software mobile development and the manufacturing industry. We could tell that cycles were so much longer than in the consumer software world and I was intrigued.  I started to work with some of our customers and realized that everything we learned at University a couple decades ago, which are just-in-time, zero stock, lean manufacturing etc. where dreams that were pretty far away from being achieved. I thought it was important to be involved in manufacturing to move the needle to improve things. That’s how ultimately after a few more startups in the US, I joined Tesla. My mission here was pretty clear. I accelerated the transition to sustainable transportation and ultimately sustainable energy. My role was originally to help integrating autopilot and VMware in the vehicle and very quickly I ended up being involved in manufacturing and long story short that’s what got me involved in working with Factory software and Automation and ultimately identify the opportunity for another startup and that’s how Full Speed Automation was born.

let's talk about Full Speed Automation, what is your product, what is your value proposition?

At Tesla I ended up being the leader for core automation engineering which is the automation team, responsible for developing all automation practices software. So I was very fortunate, I was at the confluence of anything that was manufactured, for two year and a half we were helping also other companies such as SpaceX, that were in that ecosystem. There was clearly a gap in the field with a missing abstract platform like a digitization platform like we have seen for about 20 years in the enterprise software. It became really clear to me that all our vendors and partners were extremely envious of what we were doing at Tesla, they did not realize how much work and how many engineers we had working on these issues. That’s a model that most companies, mostly SMEs cannot afford and therefore cannot be successful. Full Speed Automation is the first horizontal no code factory platform. Basically, we had a Bottom-Up approach, we digitized every component on the floor and then, in the digital world you manipulate these Primitives, these devices, in order to gather intelligence but also to orchestrate what happens on the floor.

Luc, in the poll we just did, many people think it is important to keep the human in the loop in manufacturing. Do you agree?

Indeed, HIL or Human In the Loop is very important. If you checked on our website and what I have posted previously on LinkedIn, indeed I’m always trying to educate our stakeholders and customers that automation does not mean necessarily replacing humans with robots. It’s about making our life better. Basically, as humans we have eyes, fingers and brains that are extremely powerful and there are tests that shows that humans will always have more value-added than robots. However repetitive, extremely predictable and boring tasks should definitely been fulfilled by robots and it takes data to understand where these Investments needs to be done. Also, we need to not forget that companies are here to serve humans and not the opposite.

This is indeed important; this is the S in the ESG goals. Which leads us to our topic today “sustainability in manufacturing”. According to you, is there such a thing as sustainable production and what should be the good practice in this area?

Yes of course, I mean, we will be doomed otherwise, so this is definitely possible. I think everything starts with data! We need to understand where we are, where we stand and what we need to do and this requires good datasets in order to make decisions.

What do you consider as being the good practices in a manufacturing environment?

The trend that we are observing is that efficient throughput should look slow. Actually it’s very interesting, every time we visit factories that are extremely efficient, everything looks boring and slow and I think that’s a key success. The ability to adapt and iterate rapidly is also very interesting, small and medium industries that have not adopted industry 4.0 practices, I believe they have an unfair advantage today compared to early adopters. Decentralized software practices can be more accessible, more affordable than they were 10 years ago, without having to do the heavy investments that were done a decade ago and that unfortunately can be seen sometimes as a golden jail. There is an opportunity thanks to the technological progress that we have witnessed for the past five years. Another option may be choosing the right robots, there is always a dilemma between what the company is trying to achieve right now and the capex that comes with it and looking forward, we tend to supersize the robots or have more general-purpose robots.  There are some practices where maximum throughput is not always the best. You will be willing to look into the lifespan of the robot but also the power consumption of a robot, so for instance to move things at maximum velocity may not be the most efficient way from both a capex perspective but also on energy consumption. One example that I have in mind in one of my previous jobs we were moving entire cars with very large yellow robots and you can imagine the amount of energy that you need to to move a three tons vehicle like the height of one story. By distributing the load on multiple axis on multiple joints of the robot with the right velocity, with robot controllers are relatively intelligent and they can regenerate the energy, so basically, we are using each joint as a damper and instead of turning that into heat basically and waste that energy you can reuse it on a different axis. So we could cut by 30% the amount of energy that was used to move that car, knowing that the intelligence in the robot controller is already capable of optimizing, so there is already like a 20 to 30 percent energy savings. In the end you can cut as much as 40 to 45 percent energy doing the same motion in the same cycle time by being a bit clever. But again it starts with the ability to measure the amount of energy that is utilized.

That really sounds like Aikido to me, can you share some examples of places that you would consider as role models?

It’s actually the manufacturer of that robot we are talking about who invited me and some team members in Japan, when I was at Tesla. We had the opportunity to meet the CEO, a great person, a very impressive leader. A doctor in robotics who graduated in the 60s, the man who invented industrial robots. These facilities were actually running in the dark. As you walk through the lines you have sensors all turning the lights on so that humans can see what happens and what happens there, is that robots of a certain type build robots of the same types.  It’s very impressive to see these bits of robots going from one station to another and being assembled by the robot of the same kind and everything looks peaceful quiet and they build something like 2000 industry robots of spectacular quality every week. There was no truck there, so I believe everything happened underground on some automated conveyers or perhaps using AGVs. It was a mind-blowing experience and it’s always been an inspiration.

In our poll we asked a question to the audience about trusting the machine or the AI, what do you think about trusting the machine intelligence ?

That’s very interesting. I have a dilemma with that. We’ve just talked about keeping human in the loop. Sometimes there is common sense where human can outsmart AI, or not necessarily AI, but it can be speaking of these robot controllers that are capable of regenerating the energy. If you combine the experience of a seasoned robot engineer with a smart robot controller that’s how you get the best return on investment.  My recommendation to every business owner or any production facility will be to try to be smart as much as you can and add a layer of machine Intelligence on top of that but do not completely trust it. Human in the loop always has value.  It’s been my experience.

How to measure/calculate and also share/provide the CO2 footprint in a verifiable way during production steps?

We focus on CO2 right now, but they are many other pollutants.  I’m thinking of sulphur dioxide. I care deeply for ocean and we tend to forget about the sulphur dioxide footprint of shipping goods across the world, particularly on ships across the ocean. We’ve destroyed a significant amount of coral reefs and unfortunately, it’s not something we can revert. A coral reef takes about 100 to 200 000 years to build, it’s not something that we will see coming back so we need to act more globally and that’s the same challenge as CO2 footprint.

The value chain for car is typically eight to nine providers, each supplier will use a chain of about six to seven providers so you would have to track each component coming through hundreds of suppliers and that will be extremely difficult to just have traceability on the goods. Let’s focus first on what happens inside the facility. To do so, we need to understand the quality of the energy that we are using. In France for instance, nuclear delivers a very low CO2 footprint. There are other issues arising from Nuclear energy, from mining to recycling but let’s focus first on CO2. In this regards, it’s a pretty cheap energy, something like 23 to 29 grams of CO2 per kilowatt hour. If you look at Central Europe fossil energies are used to produce electricity it’s not that good, in the United States it’s even worse with almost an order of magnitude between France and the US. So it’s important to know where the energy comes from and there are fluctuations during the day as well.

Did you have access to all the data you need to do so? Data are locked in machines so do you face problem to collect the data?

Absolutely! One of my providers when I was with Tesla wanted us to pay to have access to the data that was generated by hardware and software that we had paid for and owned for half a decade. When you think about that, the automation controls world is about a decade maybe two decades behind when it comes to the perspective on data. When it comes to web applications hosted in the cloud, I believe that everyone understands that what is valuable is really what you do with data, it’s not accessing the data. To pay just to have access to the data should be over! In our approach, anything that is instantiated in the test comes with Telemetry and we broadcast for free in the facility. On the customer side, various teams can subscribe to that data and do whatever they want with it. I think that’s really important for the industry to adopt that Paradigm so that we can allow everyone to iterate on this data to understand their process. 

We’ve been talking about measuring; this is basically data acquisition. The next step is how do you calculate fluctuation of cost or how clean the energy is at a given point in time, and that equation can require a lot of computation and unfortunately most of small Industries don’t have the compute power on the floor to achieve that. There are many ways we could imagine, compiling the data and sharing. Blockchain is an option, but we can also use centralized authorities with anonymization. There are multiple possibilities to succeed, it will require a lot of Investments. Progress in the field of environment in the past have always been led by authorities, the catalytic converter on cars is an example.  There are some very successful wealthy high value Industries that can still afford to pollute because they simply purchase credits. That’s the downside of cap and trade, if you are extremely profitable you can trade and you don’t really care about the environment. When that happens in Europe then we encourage offshore production in countries where the requirements on pollution management and protection of the environment are not as high. That translates into putting these goods on carriers that pull it like even more and we’ve been degrading the world almost more since we’ve instantiated cap and trade actually.

One way could be to rely on Central agencies, there are some initiatives that are trying to aggregate data to suggest ways of calculating the footprint and the challenge will be to democratize this access to the data and let it go to the cloud, or maybe it’s on edge with some local authentication and some devices that will be trusted by these authorities so that we can track CO2 predictions.

In the DIH² project, we like to work with small businesses and the big question for them is often “where to start”? If I'm a small business I'm maybe lagging behind in terms of automation, how can I move into industry 4.0?

We have the case in the west of France where one CEO told me recently that it will probably take him five years to staff enough collaborators to fulfill the need of one year of recruiting.  It means that robots need to come on the floor. But where do you start?

They cannot spend hundreds of millions of dollars or euros into automating the floor so we like to start with data. I think the best way to start is to instrument the floor, it can be smart cameras, dumb sensors just to inform when material is in and out or scanners. We don’t change the way people build their goods; they’ve been very successful at that.  We don’t orchestrate, we simply gather intelligence and that helps leadership to take informed decision as to where some automation investment makes the most sense. We promote rapid iteration, and we keep investing intelligently where we think it will move the needle.

You mentioned the problem of production made very far from the point of use and that's a real problem. We hear a lot about the decentralized production but this sounds counter-intuitive because for a low-cost production everything must be centralized. What do you think about that?

It’s a matter of value added. If you want to sell one product with no customization to the entire world, intuitively, we want to believe that one huge Factory in the world is going to be more efficient. It’s not necessarily true because that means you need to have a global supply chain that is going to be unique.  All the concepts of lean manufacturing, zero stock, just in time and so on… had been developed very intelligently. If you look at car manufacturing, again, one Factory was basically a city and you had raw rubber raw steel coming in local shops to build the entire supply chain. It was basically within a few hundred acres. It totally made sense to have zero stock, and just in time, and be super lean because everything was local. As we globalized the supply chain, the Covid crisis has demonstrated that it was more of a challenge, it was extremely stressed so it can be and it should be an advantage to have local production, manufacturing local goods that will be optimized for local consumption using local supply chain. I have a mouse right here, this thing could be made entirely of plastic in a region of the world that has unlimited access to oil and gas, in other parts of the world it could be produced with more recycled material and maybe somewhere else half of the parts will be made of cast or printed through additive techniques, and that’s how you will get the best return on investment locally. I’m a strong believer of that.

 CEO and co-founder

Luc Leroy CEO Full Speed Automation

Former head of Core Automation Engineering at Tesla, Luc Leroy is the CEO and co-founder of Full Speed Automation.

Luc was instrumental in solving the ‘production hell’ of the Tesla Model 3 by fixing overall automation controls issues and developing new automation paradigms that since became standards at Tesla and in other industries.

Before Tesla, Luc was already an experienced engineering leader in the fields of collaboration, security, CAD and PLM.

About Full Speed Automation

Full Speed Automation is the New Standard for Manufacturing Software.

At Full Speed Automation, they believe it is time for Manufacturing to have access to modern software practices and to stop depending on expensive and time-consuming software developments. Combining modern software practices including artificial intelligence with engineering excellence leads them to develop the most reliable and cost-efficient digitization and automation platform for manufacturing in green and brownfields. 

About DIH²

DIH² logo orange
A Pan-European Network of Robotics DIHs for Agile Production.
DIH² believes in the power of robotics to transform the agility of manufacturing in Small and Medium-sized Enterprises (SMEs) and drive economic growth across the European Union. Our role is to facilitate the connections that will enable agile production in factories where speed and versatility are essential to satisfy customer demand. Learn more
 

The DIH² project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 824964. 

CATCH & DO TANK

Catch&DO by BLUMORPHO is a series of inspiring sessions which invite the audience to shape its mind on key innovative topics. Too often Innovation is packaged into keywords, if not buzzwords. But what does it mean in real life? What does it mean in terms of execution and value ? We invite you to CATCH the idea to pull the trigger and turn concepts into action and DO the right thing.