Inside the Robotics Race | Boston Dynamics x Hyundai Motor Group x Bloomberg Media Studios
669,029 位关注者
22 小时前
Why will Boston Dynamics win the humanoid robotics race? We've already commercialized autonomous mobile robots, creating markets with Spot and Stretch. Now we're doing it again with Atlas: learning in real environments, designing for manufacturing at scale, and introducing a new era of physical intelligence.
Transcript
This is Atlas. The same Atlas that backflipped at CES and electrified fans at the FIFA World Cup. But this Atlas is doing something far more impressive. It's learning. And once it truly understands how to complete tasks, the scale changes. The high-stakes sprint to commercialize humanoid robots is underway, with the market for humanoids projected to reach $200 billion by 2035. Boston Dynamics and Hyundai Motor Group are betting that Atlas is the strongest contender to win that race and usher in a new generation of machines designed to power human progress everywhere. Hey, Mike. Amanda. How are you? So good. Nice to see you. Great to see you. Thanks for coming in. Of course. Can we go look at some robots? I would love to. Let's go. We've been productizing mobile autonomous robots before anyone else. We created the market for Stretch, and now we lead it. We've deployed Spot robots at over 500 customers across 46 countries. We've learned what it takes to do this really well. And we're targeting less-than-two-year ROI for all of our robots. We do that with Spot. We're aiming for that with Atlas. I truly believe we have everything necessary to be successful. Artificial intelligence has changed the game in robotics. Instead of programming robots to do tasks, Boston Dynamics can now teach them. Physical AI unlocks a future where machines can be used to solve a much larger set of tasks. The key question is how Atlas decides what to do. And we find it very useful to divide that process into two separate brains. The physical intelligence, so making sure that Atlas can step into places without losing balance, or can reach to grab something. And the second brain is reasoning intelligence, in charge of following complex instructions and having common-sense understanding of the world. When we talk about physical AI, it works similarly to a large language model. The more data you have, the better the robot's going to be at doing the task. The problem is that data doesn't exist yet. Atlas will have to leave the lab to create it. Can you explain, like, where we are and what happens here? We are at the Robotics Metaplant Application Center. It is a space for Atlas to learn how to do all kinds of different automotive tasks and to validate and ensure that the robot is as reliable as we need it to be before we actually deploy it into the real world. Working in automotive as our first industry is a great way for us to prove that Atlas can be a trusted, safe, and reliable member of the workforce. Automotive manufacturing also has rigorous security standards, safety standards, and reliability standards that we want to be held to. I spoke to you at CES when you were demonstrating Atlas to the world for the first time, and this feels like an evolutionary leap. What happens next? What's the trajectory? The Atlas that you're seeing here is designed specifically to do industrial work, to be able to lift up really, really heavy weights, and to work in really hot and harsh environments, so manufacturing environments, food and beverage, semiconductor manufacturing. This is just the beginning. You can have a great product, but without the relationships, it's almost impossible to achieve your cost targets. This is where the partnership with Hyundai Motor Group truly makes an impact. Hyundai Motor Group will build a new production facility in the US capable of producing up to 30,000 robots annually, with plans to deploy 25,000 Atlas units. With our relationship with Hyundai, we've already reduced from prototype to our first production run between 60 and 80%. That's what the partnership gives us access to. And that is really huge because certainty about scale is one of the most important things that you need to build a thriving supply chain. People come here because they want to make a difference in the real world with robots. 15 years ago, that was a dream. Nowadays, all the pieces are there. When we think about deploying our robots into the world, the focus is around elevating human potential. It is trying to get humans out of jobs that are bad for their bodies or are dull and dangerous. Anytime in history we've adopted large-scale automation, it has been good for humanity. Where we've seen the most effective implementations of these robots has been alongside humans. We are in a position to win the robotics race because we are passionate about solving the hard problems. It's not just about creating something in a lab. It's about building everything around it that's necessary. And in partnership with Hyundai Motor Group, we're leading the way. I'm excited to win.
Fascinating to see how quickly humanoid robotics is moving from impressive demonstrations toward real industrial applications. The real challenge will be making these systems reliable, safe and scalable in everyday operation. Behind every intelligent movement is a complex interaction of sensing, connectivity, power management and highly reliable electronics. This is exactly where the robotics race becomes especially exciting: turning extraordinary capabilities into dependable technology for the real world. 🤖
Bold claim, Boston Dynamics. You clearly have sophisticated simulation, real-world testing and impressive hardware. But if the claim is autonomy, I think the burden of proof has to go further. How do you establish, end-to-end, that Atlas has sufficient evidence and authority for each consequential action in the real world? Where is the auditable chain from: Objective → Reality → Evidence → Authority → Execution → Realization Verification → Mismatch → Recovery? And can that chain be reconstructed reproducibly rather than inferred afterwards from successful behavior? I have published a domain-agnostic autonomy benchmark specifically for this question. So here is an open invitation: Run Atlas against it.
This case is built on commercialization track record, not just capability. Spot and Stretch prove Boston Dynamics knows how to turn a robot into a supportable, scalable product, which is its own hard problem.
Manufacturing scale is a different milestone from a successful demonstration. Which public signal will come first for Atlas: production-line qualification, repeatable unit output, or supplier readiness? A dated milestone would help customers track the path from design intent to an industrial product.
Proven track record always beats hype. Spot and Stretch proved the commercial viability, and Atlas is about to redefine physical intelligence at scale. Exciting times ahead for robotics and mechanical engineering!
Although I cannot reveal any details, I’ve been a participant in Boston Dynamics’ progression from a research company into a serious commercial enterprise over the past four years. As the video claims, the passion to change the world is real, and the public results with Spot and Stretch show that they are a very serious contender to be a major robot supplier in the years to come.
Real world experience is a huge advantage here. Building impressive robots is one thing, but making them reliable and scalable in everyday environments is another.
The interesting part isn’t just building a humanoid robot it’s proving that it can operate reliably in real environments and create real value. The companies that connect advanced robotics with practical use cases will likely shape the next phase of physical intelligence.
成为付费用户可以阅读 Boston Dynamics 所有资料
了解更多 →