🤖"Beep-Boop-Beep"- What's so cool about robotics today?

At Tesla’s ‘We, Robot’ event, Tesla Bots busted out some really good moves, beat people at rock, paper, and scissors, and handed out some cool gift bags too. Musk even mentioned that it will “bring upon a fundamental transformation for civilization”.

While we’re confident that robotics will bring about a fundamental transformation, the real question is how easy it’ll be for startups to spearhead it. Robotics startups are growing popular, and more people building in robotics. Let’s try to understand why.

How do you define a robot?

Does a robot need to take on the form of a human? Does it need to have a certain number of actuators or sensors? Does it need to be made out of steel or some rigid material? Check out this robot below and try to answer these questions.

The lack of a concrete definition for a robot in and of itself allows robotics to span across different research and industries, further expanding the scope of problems that it can solve.

More problems → More types of robots that can be built → More robotics startups!

AI x Robotics

We’ve always known robots to perform simple and repetitive tasks like simple locomotion and picking and placing. AI enables robotics to perform complex tasks, propelling its functionality to a whole new level. Robots that are smart enough to make independent decisions are a gamechanger, and AI is actively helping robots achieve this. This is why humanoid robots are getting a lot of traction these days.

The progress that Figure AI in particular is making is spectacular, having raised $675M at a $2.6B valuation from investors like Bezos, OpenAI, and Nvidia. They are building an AI-powered humanoid robot with human-like functionality to address the labor shortages across various industries.

The rate at which AI is developing, it will be hard for hardware to keep up. However, the startups that are actively building in the AI x Robotics space can clearly help steer AI’s development to make it more feasible to build hardware and robots that can put it to use.

Open Source: Why DIY Robots are (Finally) a Thing

Let’s pivot to something near and dear to our hearts: open source. Thanks to cheaper components and a flood of available kits, you can now build your own robot without ever taking out a second mortgage. Seriously. Can you build a robot as complicated as Figure 02 or Optimus? Probably not, but you can still build some cool stuff!

In fact, research is actively trying to make robotics more accessible. Researchers at Harvard and Stanford have created some really cool toolkits to easily prototype robots and robotic systems. You can check these kits out below!

Since robotics is a very interdisciplinary field, the main issue with building robots is system integration. Outfits like K-Scale Labs are doing just this. They’ve designed and built a fully open source humanoid robot that anyone with a 3D-printer can build in their backyard. They have also integrated their designs with a software stack to support the integration of Artificial General Intelligence (AGI) in the future.

How far are we from printing our own AI-enabled humanoid robot at home? We had the chance to first-hand ask Pawel, K-Scale Labs’ software lead, at a startup event in Palo Alto.

Pawel mentioned that one of the biggest challenges for DIY humanoid robots is the supply chain. Since most of the components needed to build robots come in from Asia, you might be left waiting a year for one custom servo motor. However, he was optimistic of the US improving their manufacturing and supply chain infrastructure to support the influx of the robotics and hardware startups in the country.

Having published numerous machine learning (ML) papers himself, Pawel is a strong proponent of promoting accessibility in robotics through open source. In fact, he even mentioned that this was also one of the guiding principles at K-Scale Labs. Now that him and his team got into Y Combinator with K-Scale Labs, it’ll be exciting to see the impact they have on robotics!

No Money No Honey

Can’t build a cool robot without money. Halfway through 2024, this graph shows how funding in robotics has been consistently averaging around $10B per year since 2019.

Milo Werner, a General Partner at The Engine (a VC firm that invests in deep tech), talked about how the growth of multimodal AI can spur the creation of more multipurpose, generalist robots which can attract a lot of funding. He also talked about the impact that newer general purpose robots can have at home, especially in the kitchen (where is the Instant Pot of stir fry?).

Execution is Key

If there are so many problems that robotics can solve, and the cost of building robots is coming down, why is it that so many robotics startups still fail?

Let’s look at Zume Pizza, a startup that was using robots to cook pizza in the back of a delivery truck on route to your door. Sounds pretty cool right? Investors thought so too, that’s why they backed Zume with $445M in venture funding. However, it’s downfall can be attributed to a series of simple problems it failed to define and solve.

Zume wanted to build a really cool robot pizza shop that delivers fresh pizza; that sounds like a really cool product, but what’s the problem really? Ultimately, it turned out that customers didn’t really care for the product they built because their delivery times were long and the pizzas they delivered were worse than what they’d get from a regular pizza place. Maybe they’d change their mind if a special someone was delivering their pizzas.

After failing to deliver on the promises they made to their customers (pun intended), they also continued to fail in implementing the feedback from their customers; i.e. either make better pizzas or improve pizza delivery times. As a result they created a value gap between themselves and their customers, where they essentially overpromised and underdelivered. Customer is king, always.

Pivot hell is real. Zume pivoted from a robo-pizza shop to trying to trying to reinvent the food value chain to creating sustainable packaging. Their pivots were always in reaction to not being able to overcome a challenge they faced rather than a move based on strategy.

Zume Pizza’s startup journey is a cautionary tale that shows how bad execution can easily kill a robotics startup. Some of the mistakes they’ve made are vertical agnostic, and the subsequent lessons learned from their journey can be applied to any robotics startup in any vertical that you want to build in.

Spread the word

Share this lab note with your friends and let us know what you think of the format. We will adapt this to what our readers like.

Also if you want to chat with one of our leads, feel free to reach out here. Our door is always open 😁 .