Unlocking autonomous-vehicle development: TIER IV’s open-source blueprint

Shinpei Kato, CEO of TIER IV, envisions a future of transportation in which open source revolutionizes the autonomous-vehicle industry. Such a disruption would parallel Android’s rise to dominate the mobile software market, which is a testament to the power of open source. It has been a cornerstone not only in mobile software but also in web browsing with Mozilla Firefox, in cryptocurrency through Ethereum, and in the realm of code management via Git. This collaborative and innovative foundation has propelled sectors to unparalleled growth and accessibility, posing the tantalizing question: Could autonomous vehicles be next?

In this episode of the Drivers of Disruption podcast, Kato joins Takuto Ueha, a McKinsey partner and automotive sector lead, to share with host Matías Garibaldi their insights about the potential for open source, including Autoware, in the autonomous-vehicle industry. This conversation comes at a pivotal moment: autonomous-vehicle software is witnessing both groundbreaking successes and notable setbacks. Kato aims to leverage the open-source model to navigate these challenges and accelerate development and adoption of autonomous-driving technologies. He and Ueha address whether the open-source ethos that transformed mobile software, web browsing, cryptocurrency, and code management could be the key to unlocking the future of autonomous-driving scalability.

An edited transcript of the conversation follows.

Matías Garibaldi: Welcome to Drivers of Disruption, a show covering the latest advancements in the future of mobility, current challenges, and potential solutions moving forward. I’m Matías Garibaldi, and today’s episode is on autonomous vehicles, their software, and open-source applications. Our first guest is Shinpei Kato, CEO of TIER IV and chairman of the Autoware Foundation.

TIER IV is a start-up in Japan that offers a full-stack solution for autonomous-vehicle software based on their developed hardware. Autoware is the world’s first open-source software for autonomous driving, which TIER IV built and then turned into a nonprofit. They now have two products that leverage the software. One is Pilot.Auto, which is a scalable platform with a basic functionality and capability of autonomous-driving systems, where developers can create complete systems based on their reference design. And Web.Auto is a cloud-native DevOps platform with software tools and data pipelines to help developers save development costs and time to market.

Welcome, Kato-san.

Shinpei Kato: Hi. Thanks for the invitation.

Matías Garibaldi: Our second guest is Takuto Ueha, a McKinsey partner and automotive sector lead. He’s focused on a wide range of topics including product development and procurement, as well as building new businesses within incumbents. Welcome, Takuto.

Takuto Ueha: Thanks, and nice to see you, Kato-san.

Shinpei Kato: It’ll be exciting to discuss the future of autonomous vehicles today.

Matías Garibaldi: That’s exactly what we’re going to be talking about. McKinsey recently published an article on this topic, and one of the insights is that 2023 was a very interesting tipping point for autonomous-vehicle infrastructure. There have been good and bad developments over the last year. Some leading players were able to successfully run and scale their first commercial operations and increase their funding, while others saw significant setbacks, stopped or reduced their operations, or even exited the market entirely. So, Takuto, what are some of the takeaways from that research that we published?

Takuto Ueha: I’ll kick off with what we found in our recent survey. We conduct these surveys with executives of incumbents and disruptors periodically. As you mentioned, Matías, we found some pretty interesting insights and changes from the previous survey, which was issued in 2021. One of the things we saw very clearly is that development for the autonomous vehicle is clearly extending on average. The executives’ responses indicated an average of two years’ extension in terms of how the autonomous-driving space is evolving. This hasn’t changed much, but regulation, technology, and consumer safety are the top three bottlenecks that executives have in mind. I’d love to hear Kato-san’s perspective on this later.

Specifically on technologies, our respondents often say three main technologic elements are the most critical for the space: prediction, decision-making, and perception software. We saw one more big trend from a regional perspective. In 2021, a majority of the respondents said North America would be the leading region in this space—specifically speaking, the question was how fast the L4 highway pilot would come on. And this time, in 2023, the respondents were split between China and the US. There was a clear acceleration in the China region for the market and our clients.

Matías Garibaldi: That’s really helpful. One of the interesting things as well, as you know, is that the majority of the respondents were based in the EU and the US.

I’m very interested, Kato-san, to hear your view of this world. What do you think has happened over the last two years? Why are we seeing this, and where are we going from a global standpoint? We can start from a couple of points that Tak had mentioned, like timeline to autonomy—the fact that now industry leaders are believing that it’s extending by two years. What’s the reason behind that, Kato-san?

Shinpei Kato: Let me look back into how this autonomous-driving technology started, because today’s structure of players or regional strategy is pretty much based on how this technology started. Autonomous driving has a long history, but long story short, it started in the US in what is often called the DARPA Grand Challenge, which is a university competition. And this project has been out to Google, and they made a kind of big announcement that we met the autonomous-driving test from California to Nevada, which was around 2010 or 2011 as far as I remember. Now, people started looking into this technology’s potential to be a business, and a lot of tests and pilot projects or products have emerged in the last decade. Most of the projects or products focused on a specific operational design domain, which is a kind of limited area—limited conditions, limited vehicles, and limited customers. But within these conditions, the quality of autonomous driving has been increasingly improved. Now you see in San Francisco or Shenzhen some almost-production-quality autonomous vehicles, like a taxi and buses. On the other hand, today they are still struggling to expand the regions of interest. Even Google’s Waymo has San Francisco, Phoenix, and now Los Angeles. But there are three limited areas of interest.

What TIER IV is doing is to provide the scalability of this problem, using open source. So once you deploy autonomous vehicles in some area, what TIER IV wants to do is to make it easy to expand to other regions. This means open access or transparent interoperability in technology so that not only TIER IV or limited technology suppliers but also as many players as possible will be able to access this technology. Then I expect to see an ecosystem becoming mature to be worldwide, and eventually you can see more and more players provide a variety of services, not limited to specific vehicles, specific regions.

Matías Garibaldi: Kato-san, you were mentioning Waymo in Phoenix, California, and other areas. Are you saying that the reason for the slower pace or the scaling problem is, understanding other ODDs [operation design domains] and operating the vehicles in those locations?

Shinpei Kato: Fifty percent yes. From a technology perspective, yes. I think even Google Waymo still has some challenge for their underlying technology to be more scalable. But my primary concern or primary challenge is fundamentally the fact that no other players can access that technology. Only one single company is now able to access this technology, but this fundamentally prevents an ecosystem from being scalable.

The reason why we are interested in open source is to overcome this circumstance that people cannot access technology. If more and more players are able to access the technology, then we can see more and more services all over the world. That’s the main challenge TIER IV was created to address.

Takuto Ueha: Do you see a different appetite for this open-source idea between different regions—for example, North America, Europe, China? Or is the challenge that companies don’t want to disclose?

Shinpei Kato: Today, because autonomous driving has a huge range of business implications, the market size of autonomous driving potentially is pretty big. Most of the companies developing base technology are likely to take closed solutions. But the situation is changing now because people have started understanding that autonomous driving is feasible.

Google Waymo demonstrated the difference of autonomous driving, at least in environments such as San Francisco and Phoenix, which is quite a complicated environment. But still, today’s technology is feasible to be deployed in real services. This is a really good difference. So now it’s time for other players to jump into this market, and open source can facilitate this momentum.

Matías Garibaldi: So a lot of the technology leaders have proved that it can be done or that the technology’s developing, the technology is growing. Now open source provides an opportunity for others to catch up. What are some of the other roadblocks outside of technology that you see in different regions?

Shinpei Kato: Just so as not to mislead, technology still has challenges. One example includes teleoperation. But other than technology, rulemaking needs to be aligned with the local authorities or the government, and this expands to more high-level issues such as ELSI—ethical, legal, and social issues.

These high-level issues and underlying technology issues need to be addressed with the stakeholders. This is really challenging, because each company or each stakeholder has their own agenda, so somebody needs to take the lead to address the trade-off across the stakeholders.

In Japan, the government is really strong in the sense that it provides pretty solid guidelines but is quite flexible. So almost all stakeholders in Japan, including automotive makers or start-ups, are now cooperating with each other to address the ELSI. Now each kind of city or each village or town in Japan has started accepting this technology. I think other countries would have similar issues in order to accept new technologies. I think you need to address ethical, legal, and social issues at the same time with many stakeholders.

Takuto Ueha: You have seen this for many years. Are you optimistic? You raised Japan as an example; its regulatory and ethical issues have been discussed for a decade. Do you see a sign of acceleration in a positive way, or is it stalling?

Shinpei Kato: Japan is moving ahead. I would dare to say that Japan could be a good reference in regulatory because we have taken the right steps in the white-box schemes. Everything is transparent. Still there are challenges, of course, but we never step forward without the black-box schemes. Of course, this is a really complex, time-consuming undertaking, but the reason why Japan is now moving forward is that stakeholders have some consensus in ELSI. There are no clear answers, but the stakeholders have consensus, so we can move forward.

Matías Garibaldi: Diving deeper into open source, how does TIER IV foster collaboration within the open-source community? Who is this open-source community, and then how do you ensure that the contributions are aligning with your project goals? Or is that basically a product of the open market, so people are going to be working on the things that are most important, and they in general align with what TIER IV would be interested in? How does that work?

Shinpei Kato: To be honest, I don’t have a clear answer on how we can foster open-source communities, but that’s part of what we have experienced in the last decade. The first approach, I believe, which led to a small success in the early days of Autoware, is that Autoware used to be maintained by TIER IV, but TIER IV transferred ownership of this maintenance to a nonprofit organization called Autoware Foundation. This probably opens up opportunities for other companies to feel like joining this community. They don’t want to belong to a single company, and if Autoware is maintained by TIER IV, joining this community really means they belong to a project of TIER IV. This would not be sound, in my opinion, because the spirit of open source is across countries, across politics. It should be nonprofit. Autoware accelerated because TIER IV transferred ownership of the software project to a nonprofit organization, so some credibility or reliability or transparency of the project has improved, and so more companies easily join the community.

Takuto Ueha: That’s very helpful. How do you think about safety and quality assurance under such mechanism?

Shinpei Kato: As aforementioned, the safety issues or quality issues highly depend on the business model of the users. Open source is just the way to climb a mountain halfway. Open source is the way you can start climbing, but the way you actually climb the mountain should be determined by your company. TIER IV is also positioned in this way, so TIER IV provides multiple solutions, some of which include the safety mechanisms, if needed from the customer’s perspective. But if the customer can do the safety assurance by themselves, then they can directly access the open source, and they don’t really need TIER IV’s assistance. But all in all in either way, safety issues, quality issues, any end-user issues should be addressed by each company.

Open source is just a big lake. There are many opportunities in open source, but the ultimate responsibility in delivering the product to the customers is the company’s challenge, not the challenge of the open-source community.

Matías Garibaldi: This is a question for both Kato-san and Tak. Kato-san, starting with you, are TIER IV and the open-source autonomous-driving software influencing or having an impact on traditional automotive manufacturers’ approach to product development in software? I know there’s recent news of TIER IV and Isuzu, and I would love to understand what you’ve seen the traditional automakers do with this technology.

Shinpei Kato: That’s a good question and fundamental in the sense that a majority of the market comprises automakers. I want to be sensitive to this question, because the automotive market is mature, it’s already structured. It shouldn’t be jeopardized by introducing new ideas, because this is a really solid structure. We should leverage this existing scheme so open source is what they can take in order to, say, improve incrementally the existing structure.

Some automakers may not even need open source at all if they can provide the solutions for the customers by themselves. But I believe that in this market, there are more than 200 auto brands. I would say maybe ten of them could deliver solid autonomous-driving solutions by themselves. The rest of automakers’ brands probably need to access other technologies, so open source may be a big opportunity for them because they don’t have to build new technologies from scratch. They can reference open-source software, and if needed, they can contact the tech players in the open-source community as before, so we can work together to build new solutions.

Once again, I want to be sensitive, because this scheme could jeopardize the existing structure of the automotive market. What TIER IV wants to do is to improve or enhance or complement the existing structure, and we can build our own industry next to the existing automotive market, back-to-back. For example, Level 4 autonomous driving, all kinds of mobility services, could be a different structure that is constructed from scratch. The existing automotive market needs to be kept solid, but we can still enhance, improve, and complement using open source.

So that’s my idea and TIER IV’s vision. The art of open source can reimagine the whole intelligent-vehicles market, but it doesn’t really mean we change the existing automotive market.

Takuto Ueha: I really liked your analogy of climbing the mountain: open source is not the solution to climb all the way, but it’s the enabler and the foundation to make the first step. I also really appreciate your humbleness and recognition of the existing solid maturity of the automotive industry, which I think is very important to call out.

Some automotives, as you mentioned, have already begun some of their own attempts. Leading OEMs are doing open source within their manufacturing plants. While I agree with you that maybe they can do it independently if they are in the top ten, there’s a huge value unlock if the leading OEMs also realize the benefit of the open-source vision you are talking about. And I say this because the leading OEMs are still working within very closed scope and vision when they make their own attempts individually, and they are struggling, as you know. I think there’s probably unlock potential in being more open and being humble on the thinking around “What is the foundation we can leverage?”

You use the lake analogy, which is excellent, to further accelerate their individual initiatives because speed is also a critical problem. Although OEMs are struggling with the idea of exponentially expanding their data source, they can potentially accelerate their individual initiatives much faster, and it’s something to think about. I would love for you to go see some of the clients we meet to influence them in that way as well.

Matías Garibaldi: Open source reminds me of a saying: “If you want to go fast, go alone. If you want to go far, go together.” If we want to go further in the development of autonomous vehicles, we need to go together and allow people to catch up, given that with autonomous-vehicle technology, if there’s an issue with one company, it’s almost an issue with everyone and everyone’s perception of autonomous vehicles. So it’s very important for everyone to be doing well. That really helps the point of more collaboration across companies, and so it’s very insightful.

Shinpei Kato: I also like the philosophy that if you go alone, you can go faster, but if you want to go far, then you may want to go together.

Matías Garibaldi: You can go fast but not for too long, so you need to go together. Kato-san, we talked about the last two years, and you shared the history of autonomous-vehicle development. Where do you see the future of autonomous-driving technology going in the next five years?

Shinpei Kato: That’s kind of an exciting view. In general, autonomous driving today has two kinds of streamlines. One is Level 4 commercial vehicles like robotaxis, and the other streamline is software mainly designed for passenger vehicles—privately owned cars.

In the next five or ten years, I believe that these two streamlines are going to merge, and you will see truly intelligent vehicles, which could be passenger cars or commercial vehicles. And if users request Level 4, then it can be Level 4, but for those who want to enjoy driving, they can leverage this technology to just assist their driving. This is what I hope to see or expect to see in the next five years or ten years in general.

My personal and company agenda is that we want to enable more and more new companies to join this market. If today some company doesn’t have a sufficient standard of technology to manufacture the vehicles but they leverage open-source software or open data sets, then why not become a new carmaker, as Tesla has done before? I like Tesla’s mindset that they challenge new mass production of cars.

Likewise, TIER IV wants to meet the challenge of new mass production of carmakers. This will lead to a more scalable automotive or mobility market. So I want to make some paradigm shifts: today, people may focus on mass production of cars, but I want to make mass production of carmakers.

Matías Garibaldi: Very interesting. Tak, what do you think?

Takuto Ueha: I had no idea about the real benefit of open source and struggled to understand what you are trying to achieve. But today’s conversation was very helpful in better understanding, and I’m more convinced that, yes, there’s a value not just for new challengers but also for the incumbent OEMs to consider partaking in this approach of open source.

Matías Garibaldi: Kato-san, we spoke about the five- or ten-year vision. What’s next for TIER IV? What goals do you have for the end of the year or the next 12 months that you are looking forward to or excited about?

Shinpei Kato: For the short term, we want to demonstrate the concept of what we’re doing based on open source. For example, we’d like to demonstrate that the production-quality level for autonomous vehicles is actually feasible and available to real customers. In the next 12 months, we promise we’ll deliver real Level 4 certified production-quality vehicles and services, so that these demonstrations could be a baseline for TIER IV to provide multiclass solutions. Using this experience, we can sell the whole service if needed.

Our architecture also enables us to provide components of this service and even the educational or development platform we are using. And we can open up for other partner players so they can do the same thing as what TIER IV is doing. In the next 12 months, we want to demonstrate the concept of the company, and in part, we can start the real services.

Matías Garibaldi: Thank you, Kato-san. As we wrap up, I have a question for both. I’ll go to Tak first, but then the same question to you, Kato-san. What are your closing thoughts or advice you would have for players looking to enter or already involved in the autonomous-vehicle space?

Takuto Ueha: Have a long-term vision of how your company—specifically, “your” company as in the player who is considering joining this—have a long-term vision of how your business model is going to work, but don’t get constrained with short-term return. Too many companies I observe seek an immediate return, and it’s a challenge because they have investor pressure, they have market pressure.

Both large and small companies tend to make hasty decisions. It leads back to that “If you want to go far, you’ve got to team up” mentality. Don’t get bogged down in the too-short-term return is my advice to the players considering this field.

Matías Garibaldi: Thanks, Tak. How about you, Kato-san?

Shinpei Kato: Three words: just use Autoware.

Matías Garibaldi: Perfect. I think that will wrap us up. Thank you so much. I’m looking forward to the next 12 months for TIER IV. We’re excited to read more and see that demonstration. Tak, thank you so much for your time as well.

Takuto Ueha: Thank you. Appreciate it. Thank you.

Shinpei Kato: Thank you.