The New Space Race: How Hardware Problems Have Been Turned Into Software Ones

Publié dans Coder stories

06 mai 2020


The New Space Race: How Hardware Problems Have Been Turned Into Software Ones
Chris Stokel-Walker

Journalist and communicator specialising in digital culture, influencers and YouTube

The US and Soviet Union’s battle for supremacy in space between the 1950s and 1980s caused a huge shift in the development of space technology. From being almost nothing in 1955, the aerospace industry increased to be a massive sector of the economy—and had successes to match. From Sputnik, the world’s first satellite, which started orbiting Earth in 1957, to Apollo 11 landing on the surface of the Moon in 1969, the development of the space industry raced ahead with unforeseen speed.

Then, everything stopped, or slowed significantly at least. The urgency disappeared, as did the funding. Advance slowed to a stumble and issues repeatedly arose.

Those under the age of 30 haven’t had that experience of exciting innovations in the aerospace industry. They’ve not witnessed the nimble, unreal development of technology at the speed of light, or the excitement of advances that redraw the boundaries of human reach on a yearly basis. In part, that’s because intra-country competition vanished. The US won the Cold War, and so some of the impetus to develop new programs and missions diminished. Big business also entered the equation: The budgets for space projects ballooned and caution took priority over innovation.

“The mindset in the 1990s and 2000s was that everything had to be perfect,” explains Lucas Brémond, Chief Architect at Loft Orbital, a space company based in the US and France. “You couldn’t screw up your mission because the launch was so expensive that, if the satellite didn’t work, it was a complete failure.” A single payload could cost $100 million and require 10 years of development. “Moon shots”—the term that has since been co-opted by start-ups to define one-off, risky attempts—literally were one-offs. If the rocket exploded on launch and you lost the mission, you lost everything.

But after decades of stagnation, during which it was weighed under by its own immense scale, the aerospace industry is evolving again. What has unkindly been called “old space” (a term that some, including Brémond, dislike) has been supplanted by a new paradigm: new space.

New space is driven by private business and borrows more from the ethos of scrappy start-up tech firms, rather than the giant multinationals that have dominated the aerospace industry for the past few decades. In the previous six or so years, names such as SpaceX, Loft Orbital, UnseenLabs, and Spire Global have started replacing Lockheed Martin, Airbus and Boeing in the conversation, and instilling a new way of working that has been injecting some of the impetus back into the world of space. And they’ve done so by hiring people who the traditional aerospace industry often wouldn’t look twice at—developers and engineers more used to tapping out code on a screen than tinkering with hydraulics and aerodynamic equations.

From old space to new space

“Loft Orbital’s adventure started three years ago in San Francisco,” explains Pierre Bertrand, who is head of its innovation center. “Since the beginning, the company has been kind of in between old space and new space.” Old space—the preserve of big firms such as Lockheed Martin and Airbus, buoyed by big contracts with Nasa and the European Space Agency—has given way to new space, and the latter is providing a shot in the arm for the aerospace industry. In 2019, Morgan Stanley estimated that the global space economy could skyrocket from about $350 billion today to more than $1 trillion in 2040, fuelled by development from new space-companies across the globe.

UnseenLabs is another new space company that has sprung up over the past five years, founded by brothers Jonathan and Clément Galic, who both have experience of working for old-space giants: Jonathan was previously at Airbus, while his sibling spent two and a half years working on a project for the French space agency, CNES. Before they joined what Jonathan calls “legacy space,” the two worked in the start-up sector. “We approached the space industry more as IT guys,” he admits. “Space is just a physical constraint we have to deliver our service.”

But new space companies don’t appear out of nowhere. They require funding and borrow heavily from venture-capital-backed peers in the start-up sector. Two of the early movers in the market are based in Silicon Valley, the crucible of technology development. Planet Labs was formed in December 2010 by three scientists from Nasa, with the aim of rewriting the old norms of space. “By reducing the cost to reach space by a factor of 10 and developing satellites at 1,000x lower mass and cost than 10 years ago, data that once was only accessible by government entities is now available to the masses—and is being utilized daily across industries to achieve great things that were never imagined,”wrote Robbie Schingler, co-founder and Chief Strategy Officer of Planet Labs, on the company’s blog at the beginning of 2020. It remains an active participant in the industry and has birthed plenty of other firms.

Another competitor in the early days was Spire Global, founded in 2012 by three graduates from the International Space University, a French research institute. Based in San Francisco, Spire has raised successive rounds of funding and launched many satellites: It currently has 88 in orbit—a 22% increase year-on-year, according to Daniel Bryce, Head of Manufacturing for the company, who is based in its offices in Glasgow, Scotland. The founding ethos of Spire is unchanged today—and borrows plenty from the mottos of start-up companies such as Facebook and Google, for whom “move fast and break things” was a guiding principle. “What we want is pretty much more, better, faster,” says Bryce. “How do we generate more data, better quality, and get it to the customer faster? That’s the driver for everybody. Everybody wants to get their information instantly. Hence all the new technologies that are coming out driving towards that same goal of more, better and faster.”

That’s a world away from the carefully cross-checked approach of old space, where any mistake could see hundreds of millions of dollars disappear in a fiery inferno. It’s now possible to break things—or send half-ready satellites up into space—knowing you can fix them more easily. “This is exactly what this current trend is about,” says Brémond. “Launch is becoming cheap, satellites are becoming a commodity, so there’s no need to over-engineer and over-test everything. You try it, and if you lose one, it’s not the end of the world. That’s been made possible by advances in technology, and a significant shift in the way the challenges of space are approached.

“During the 1990s and 2000s, we saw this heavy bloat on the ability of the space industry due to the multiplication of contractors and the more and more stringent rules that said you couldn’t fail. The price went through the roof. Even Nasa didn’t have a supply chain that was big enough to make competition happen. It just had a couple of partners that were always contracted, so they could charge what they wanted.”

How space reinvented itself

The arrival of new participants in the space industry in the 2000s changed everything. “You suddenly have a much more vivid ecosystem that is driven by supply and demand, rather than the politics or the status quo of having two big companies in the game,” says Brémond. Business models are also being rewritten: Until SpaceX came along, rockets were a single-use product, and anyone who told you otherwise was laughed out of the room. “You could hear the head of a big European space agency saying landing a rocket is never going to happen,” says Brémond. “Then they’d shift and say it’s never going to be cost-effective. And now SpaceX is using a relanded rocket four or five times without any problems.”

That’s fed through to a broader shift that borrows from software development—the idea that these satellites currently orbiting through space, and the rockets that get them there, are in a constant position of beta testing. “As we were born in San Francisco, we were initiated into the tech industry and all the tech companies that surrounded us—LinkedIn, Facebook, Google, and so on,” says Brémond. The proximity meant that, during its early years, in its development process—everything about the company, in fact—Loft Orbital looked more like the tech firms it considered its peers than the old space giants it was aiming to replace.

It is also borrowing versioning from software development for its satellites. “You’re starting to treat what fundamentally used to be a hardware problem as a software problem,” says Brémond. “It’s a much more iterative process, where bugs are tolerated because it’s software development. A zero-bug software product doesn’t exist. You release something that’s OK and you just patch it or improve it all the time. This is the same process being used by microsatellite companies—they’re not trying to catch every single issue from day one. They’re going to start by sending a few prototypes and then iterate from there, and hopefully all the flight experience they get from their potentially buggy prototypes is fully reinjected into the next versions.”

When Loft Orbital launched in 2017, most of the people interviewed and hired, as well as its investors, were children of the tech industry. “They’re used to interacting with tech companies, with technologies from the web-dev and the tech industry, and so it has been in our DNA since the beginning,” explains Bertrand. The company has now expanded to 35 employees and has a research and development base in France. The workers are drawn from all areas of expertise and come from both old-space and new-space companies, but equally they could come from anywhere. “We are also super-open to profiles that don’t know anything about space, about orbital mechanics or thermal dissipation or embedded software, but they know about recent technologies and the stack we’re using to develop our web applications,” says Bertrand.

It’s the same for UnseenLab’s new hires. “We can hire good-profiled guys from software and search, and bring them the space layer on top of their competencies,” says Jonathan Galic. “We are able to attract very good people because space is magic, but for us, very good people are not people who have 20 years of background in the legacy space industry, but are guys who know how to make very good code in a system with physical constraints.”

Writing a new legacy

Being able to hire coders who also have knowledge of legacy space may be a rarity in the world of aerospace, but the industry has been evolving nevertheless. “Usually in space we have a lot of legacy,” says Bertrand. Much of the software still in place in many parts of the industry, whether that’s test bench software or monitoring programs, was developed in the 1960s and 1970s. “It’s tough to innovate because you have all of this legacy,” adds Bertrand, who worked on OneWeb satellites before arriving at Loft Orbital, where all the test benches for satellites used Tcl, an old programming language that very few developers knew how to code in. “It was very difficult to hire for.”

So the raft of new-space companies is deliberately rewriting the rules. They’re trying to start afresh, coding in languages that are accessible to all, which allows them to hire from a broader range of backgrounds—and to compete for talent with traditional tech firms, who usually have the brightest and best candidates. This approach also includes using open-source software. “I’m a big open-source fan,” says Brémond, “so at Loft we’re also trying to do our part in open source. If you think about it, most servers in the world are running not Windows, not MacOS, but something like Linux. This is screaming evidence of the fact it works.” Loft Orbital uses open-source software in its onboard technology, taking advantage of the cost benefits of using free, regularly updated versions for its satellites. “There’s a clear shift where there’s no point trying to use proprietary software,” says Brémond. “Previously, it was a heavily guarded contractor base, and ‘open source’ was a curse word.”

Loft Orbital tests all its satellites using Python, and the web-development stack for its back end is coded in Python, using the Django framework. The front end harnesses the power of Vue.js and the API is powered by GraphQL. “The core of our business model is to be able to adapt to any type of customer. In one satellite we are flying several customers,” says Bertrand. “It means we need software to interact with a whole bunch of diverse payloads and instruments, and for the base of our satellite we use different suppliers, so we also need to interact with diverse platforms.”

The same nimbleness can be found at the core of Spire, too. “One of the main drivers behind the manufacturing model we ended up with was the ability to iterate quickly, to get new designs, new software loaded onto the satellites or the ground system or our constellation optimizer system,” says Bryce. “We vertically integrated everything so that we can be fast, and so that we can improve daily. That constant update of software is most noticeable in the frequency of updates of code on the satellite side.” The company codes a lot in Python, and Bryce is one of the oldest employees in the Glasgow office. “I’m the only person who doesn’t just write code for everything,” he jokes. “I still use spreadsheets to work things out.”

While Spire has manufacturing test suites to execute testing of its satellites, any issues are usually caught in the code and fixed there rather than through hardware. “That’s something that’s very noticeable to me coming from a manufacturing background—the proficiency with software and software development throughout the full chain of development and manufacturing and operations of the satellites as well,” says Bryce. “Really, from concept to operations, everybody involved is a software dev to a certain extent.”

Spire’s in-orbit satellites get regular software refreshes and updates pushed to them. “The operational guys are constantly improving their scheduling, their constellation optimizer software. And the data-pipeline guys are constantly updating,” says Bryce. “The speed of projects that come out is pretty frantic. It’s way faster than anything I’ve ever experienced in corporate industries, and it would be way faster than [old] space companies could actually achieve.”

And when he says fast, he means fast. Nasa’s program for Artemis, the twin sister of Apollo, has a multi-year timescale to launch, while Spire can launch a new payload into orbit within six months, and any changes to its constellation-replenishment models have a one-month build cycle. Loft Orbital works to a similar timescale. “You don’t need 10 years R&D to fly your first flight,” says Brémond. “What we’re offering is a more streamlined approach to this, where we’re avoiding a lot of the first-flight problems because we can take new payloads on board and make them fly, and provide them with the right control tooling and a safe sandbox environment, so that people can fly their stuff without having to worry too much about how the thing works.”

That agility has allowed a huge expansion in the new-space industry—and the capability to expand pre-existing capacity already in orbit. Though Spire has seen a 22% annual increase in the number of satellites it has in orbit, it has seen a 200% increase in the daily volume of data downloaded from those devices “We haven’t changed the hardware,” says Bryce. Instead, Spire has updated the software running on it to be more efficient.

What the future looks like post COVID-19

While the world, including the new-space industry, gauges what the new norms will be after coronavirus pandemic eases, those within the sector are hopeful about its future. “It’s just going to increase,” says Bryce. “Spire is doing it through this software platform for both the satellite and the data pipeline, and the constellation optimizer system is the way we can iterate quickly and get more data.”

For Brémond, it’s a really interesting time to be working in the industry: “You can certainly see a new acceleration and a new space race going on in the private sector.” However, the concern that the pandemic could stymie progress is there, of course. “It’s hard to know whether the economic change that is coming is going to kill a sector that is vivid and still young, or if it’s going to reshuffle all the cards,” he says. Whatever happens, it looks certain that the space industry’s new paradigm is here to stay.

This article is part of Behind the Code, the media for developers, by developers. Discover more articles and videos by visiting Behind the Code!

Want to contribute? Get published!

Follow us on Twitter to stay tuned!

Illustration by Victoria Roussel

Les thématiques abordées