Navigating Uncharted Waters: a Robotic Startup Journey

Each year, about 400,000 new industrial and enterprise robots hit the market, with over 3 million robots operating globally (as of February 2022). As per McKinsey, 88% of businesses worldwide plan to adopt robotic automation into their infrastructure in the coming months. While the stats look bright and promising, there’re still a lot of challenges and bumps in the road that robotic companies, and startups, in particular, should overcome to ensure their business sustainability, competitive advantage, and growth.

To better understand the current market for enterprise robotic solutions in Europe and explore a typical robotic startup journey, I’ve talked to Mihai Crăciunescu, Founder and “Robot Spawner” at Adapta Robotics, and his colleagues Diana Baicu and Cristian Dobre, Co-Founders and robotics engineers.

Adapta Robotics is a unique example of how a corporate robotic department can spin off into a standalone robotic scale-up.

Enjoy the conversation below!

Hi guys. Let’s start with a general robotic industry trends discussion. What sectors and industries use enterprise robots most actively today? 

Mihai Crăciunescu:

Though many different industries use enterprise robotic solutions, most use cases are basically the same. As such, robotic use cases in automotive can be replicated in other sectors, from manufacturing to aviation and beyond. The need is the same throughout industries. At Adapta, we’re working with the medical, manufacturing, R&D, aerospace, and automotive sectors.

However, the retail industry badly needs robots today to ensure automated solutions and customer retention. The level of adoption is still low, as retail businesses mostly automate simple processes like check-out while keeping many critical operations unautomated. This can drastically affect their user retention, new user acquisition, and top-line growth.

Speaking about the automotive industry, in particular, do you agree with a statement that robots will drive forward the E.V. revolution? What do you generally think about the current level of robotic adoption in automotive?

Mihai C:

If you look at electric cars, they have a lot of sensors and actuators to make intelligent, automated decisions. So, they're robots per se. The assembly of such vehicles and their testing and validation are highly automated with robots, too. I believe automotive is already the most robotized industry, so it’s an excellent example of a sector that actively uses robots to reduce costs and increase productivity and efficiency. 

From your personal perspective and experience, what factors or challenges prevent enterprise robots from going mainstream? Is it cost, interoperability and compliance issues, lack of industry standards, talent shortage, or anything else?

Mihai C:

First, it’s a lack of understanding and resistance to change.

Usually, non-technical founders and leaders have little or poor understanding of robotic technology and how it works. They’re used to traditional approaches, have legacy systems that somehow work, and view robotic automation as a high-risk investment that may not pay off well. So, to improve productivity and increase bandwidth, they prefer to hire and put more people and resources instead of building custom robots. They’re afraid that new technologies will undermine their status quo and bring about new challenges they don’t want to cope with. As such,

“unwillingness to take a risk and navigate uncharted waters of automation is another blocker of mass adoption of robots.”

Thirdly, that’s the cost. Building prototypes and manufacturing robots is a complex and high-investment endeavor that not every company can afford. External factors like a global shortage of semiconductors we’re experiencing now make the process even more expensive.

And when you source engineers for your projects, do you face a shortage of qualified talents specialized in robotics?

Mihai C:

I believe this is a general problem all tech companies are facing now, especially in Europe. Hiring robotics engineers here in Europe is tough, as some time ago, local companies started outsourcing their hardware manufacturing and software solutions development to Asia or other countries. This created a massive gap in the market for robotic engineers. The most qualified engineers from Europe are typically recruited by the U.S. companies, which makes the competition for specialists even more fierce.

And how do you solve this issue?

Our approach to solving this issue is to grow talents internally as much as possible. We start by hiring A.I., electronics, or hardware engineers and continue building new skill sets on top of their existing ones to turn them into robotic engineers.

As more robots enter different industry sectors, what new skills should employees pursue to handle them properly? What’s the best way to educate staff on how to work with robots?

Mihai C:

Our job as robot creators is to build and deliver robots that can be used literally by anyone. We always keep in mind that people who’ll be operating our robots aren’t necessarily technical, so we aim to deliver user-friendly solutions.

Of course, when we at Adapta Robotics deploy our robots on the client’s site, we need to do some training, demo the features and show how the whole system works. However, it’s not rocket science to use them. I believe everyone should be able to use robots without any special knowledge. That’s the point.

Since you’ve mentioned your company name, can you tell us your brand story? How did Adapta emerge, and why?

Mihai C:

Our journey started many years ago. Cristian and I met and teamed up while studying at the university. Our goal was to build robots for the university competition. With high-performance robots designed and built by our university robotics team, we competed internationally and came first in numerous national and international contests. 

Diana joined our team a bit later, in 2015, and helped us improve the performance of our competition robots and win the biggest robotics competition in Europe at that time – Robotchallenge Vienna. We experimented a lot with robotic tech, built prototypes of self-driving cars and other projects, which eventually helped us gain indispensable experience and robust internal expertise.

At the same time, we started thinking about building MATT – our flagship robot and the ultimate solution for testing touchscreen devices. To attract external investments, we applied for and attracted E.U. Funding. 

In 2017, our well-knit robotic team joined a custom software development company rinf.tech, as its Robotic division. Having received the funding, we managed to build our first full-fledged robot MATT. This allowed us to explore potential markets to tap into, identify the needs of diverse industries, and propose testing solutions for each of them.

To give MATT all features and capabilities it has today, we worked continuously to improve it. We enhanced its software suite by adding a user-friendly GUI and used new computer vision methods to perform on a new use case basis.

As we customized and expanded MATT, we went on a roadshow and had much success at the Embedded World Conference in 2019.

In 2021, we made a decision to spin off a rinf.tech robotic division into a standalone robotic brand independent of rinf.tech. 

Why?

Mihai C:

There were several reasons for this. 

First, we gained recognition from our customers who got confused with rinf.tech being a software services company that builds robots. 

Second, this move was required by our potential investors. We realized that scaling up internally at rinf.tech would be more difficult than scaling up a startup, so we decided to separate.

What does the process of making enterprise robots look like? Does it follow a traditional software development lifecycle or take a completely different path?

Mihai C:

Regarding the software side, we follow software engineering standards and methodologies: we develop custom U.I., modules, and add-ons like in any other software product. But besides that, there’s a completely different area of logistics, choosing the right components, building hardware, prototyping, and testing.

Our approach is to start with a robot design that can take from a couple of months to a year. Then we start developing small components to be used for robot testing. We create small prototypes that we test and develop firmware for.

After that, we produce a robot prototype based on the chosen design and software modules we write. Then we start making changes and optimizations based on the prototype feedback and performance. 

As we work on the prototype improvement, we start seeing things we hadn’t seen before. After creating a second or sometimes a third version of the robot, we go for certification. It’s an extremely important step. If you plan to enter certain markets, certification is a must for your robot to operate legally and be monetized. To get certified, your product needs to be well engineered and tested. Poorly tested solutions can fail the certification. 

The certification process puts an emphasis on the safety of your robot. For example, if your robot needs 230V to operate, a certification process will prove that your robot is safe to be used alongside humans. Sometimes certification is required to ensure your robot can be used alongside other measuring devices and that it will not interfere with the data gathered by those tools.

We ensure each of our products follows standards to pass certification with flying colors.

Cristian Dobre:

I want to add that while it’s easy to follow software development methodologies and best practices, it’s trickier for the hardware side. With hardware, you have several revisions that need to be properly tested. What’s unique about Adapta is that we all work on-prem to avoid miscommunication between remote teams and make sure software and hardware integrations work in concert. 

Such an approach also helps us minimize bugs and errors at production and reduce the time to build a robot, as we can have fewer revisions compared to other companies that build robotic solutions in a distributed manner.

Working in-house as a well-oiled machine allows us to quickly fix issues, align better in real-time, and lower our clients’ costs.

Mihai C:

Such an approach also helps our robotic team a great deal. If at some point we want to make changes or follow a different direction, or if someone on our team wants to tap into a different technology (e.g., from embedded to A.I. development), they have a good opportunity to see the whole process, how things are put together and provide input or initiate product changes.

We always work based on the team feedback, and having an entire team work on-premises means more efficiency and better commitment. 

What are the key challenges or bumps in the road you’re facing as a robotic scale-up?

Mihai C:

It all comes down to the specifics of the European startup and scale-up market. Over the past twenty years, European investors have focused primarily on funding software solutions. The average investor has little to no experience with hardware startups. So, the market is shaky when it comes to investing in robotic or hardware solutions in general. If you really want to grow and scale up, you need to move your business to the United States.

Besides those challenges, we’re facing other obstacles. We specialize in custom robotic solutions, so we need to find investors interested in such solutions or having the right contacts in industries essential for your brand evolution. But it’s not impossible, which is good news. It’s just a matter of the right outreach strategy and meeting the right people.

And while looking for investors, do you keep working on new robotic solutions? Or do you guys mainly focus on improving your existing robots?

Mihai C:

We do both at the same time. As we received feedback from the market, we updated our MATT's value proposition for customers, built different add-ons to improve performance, and created new specialized use cases. 

When it comes to ERIS, we’re working on developing new functionalities. ERIS is a robot that can scan retail store shelves and provide information about the stock and prices. We’re turning it into a real data machine because it scans a lot of products and generates a lot of data and insights that can be used by other retailers later on.

Besides this, we indulge in other robotic projects like an autonomous wheelchair.

And what piece of advice would you give to fellow robotic startup founders? What should they focus on in the first place?

Mihai C:

My advice would be to focus on the technology side of their products. They should try to develop something that would add real value to the market and try to make their solution as refined and polished as possible. This will help them find the first users, raise additional funds, or find a reliable tech partner. Technical aspects need strong feedback from the market to work properly.

If your initial product idea fails to prove feasible, don’t give up, as you can still use your technology for other use cases. As such, do focus on the technology behind your robotic solution, as it’s the backbone of your business.

Also, we have some supply chain issues, especially in the semiconductor area. So you need really good partners and vendors to help you build your stock of components. And eventually, you need a good investor to help you sustain your solution with potential clients, advice, new market penetration, and more.

Diana Baicu:

Also, when the product’s prototype is ready, it’s important to keep an open mind regarding potential clients. Most of the time, they provide extremely useful information about the industry you’re looking to tap into and about your future product. Make sure you talk to your first customers to define how your final product should be modified and fine-tuned. Improve and pivot based on their feedback.

From your personal perspective, do you believe that A.I. and robots will eventually overtake human intelligence? How will humans and robots co-exist in the future?

Mihai C:

That’s a tricky question. In my personal opinion, robots can be far better than humans in some specific tasks and jobs. However, with general AI able to understand our environment and context and make the best decision every single time, I don’t see it happening at scale yet. But I still don’t believe robots will overtake human intelligence in the future.

Diana B:

Having all these perspectives from the sci-fi world and seeing that a lot of damage can be done without proper regulations, I think that once we see more improvements and better performances of robots, many different regulations will come into effect. This will help minimize the risk of A.I. overtaking human intelligence. 

Cristian D:

My take on this would be that it doesn’t necessarily have to be a super general AI that takes over. With AI we have today, we can still misuse it and cause damage to society. So, I’d be more apprehensive about how people use A.I. against other people than if general AI will take over. 

Mihai C:

I agree with both Diana and Christian. I want to add about regulations. If we look at biotechnology, for instance, Europe and the USA ban the creation of “designer babies” or experiments with DNA of certain species, while other countries don’t. Without proper regulations, new technologies can be developed and used against us. As such, regulations really make sense, but they should be applied and enforced globally to meet their goals and prevent damage that A.I. can potentially cause. 

Videos and images courtesy of Adapta Robotics.

Please feel free to check out my interview with the Chief Innovation Officer about how to grow geeks into π-shaped R&D Engineers.