Advances in AI-driven scientific research have the potential to propel us forwards and take advantage of human knowledge and expertise.
Open source: the new frontier for space exploration
Since Neil Armstrong and Buzz Aldrin took their first steps on the Moon in 1969, technology has undergone exponential innovation. After all, the Apollo Guidance Computer (AGC) had less computing power than a washing machine.1
Despite the fact that humans haven’t returned to the Moon since the cancellation of the Apollo program in 1972, there have nonetheless been incredible advances in space exploration in recent years. In 2019 alone, we now have multiple companies developing reusable rockets, organisations germinating plants from the lunar surface, and NASA offering private citizens the opportunity to visit the International Space Station.2,3,4
Amid all the preparations to return to the Moon, and as we look to once again take humans out beyond Earth’s orbit, what exactly are the future technologies and innovations that will enable us to accomplish the unimaginable?5
The importance of being open
Traditionally, businesses have looked to safeguard intellectual property of the software that they developed by keeping application source code top secret, however this is rapidly changing with the software revolution that is open source.
Open sourcing enables developers, entrepreneurs and organisations to participate in and contribute to a worldwide community built on knowledge-sharing and collaboration. Over the past two decades, this philosophy has gone from a largely underestimated movement, to a full blown enterprise software industry, which is expected to be worth $33 billion by 2022.6
This uptake in open source is also apparent in the research industry, where some of the most exciting space sciences are being conducted on high performance computing (HPC) systems. Evidence of this can be seen in the list of the TOP500 supercomputers worldwide, where every single system is running some form of open source operating system Linux.7
Open source in orbit
It isn’t just that open source is expected to play a role in future space programs because of its growing influence across the globe, the fact is that open source software has already run in space, on a computer system installed on board the ISS.
The idea behind HPE’s ‘Spaceborne Computer’ (which is controlled via an open source operating system Red Hat Enterprise Linux) was to test whether off-the-shelf computer systems could be successfully utilised for extended periods of time in outer space. This goes against traditional thinking, which mandated that computing systems should be highly-specialised and specifically hardened to protect against the harsh environment that exists outside of Earth’s atmosphere.
In June this year, having run computer and data-intensive processes continuously for 615 days in orbit of Earth, the Spaceborne Computer was returned via a SpaceX Dragon capsule.8 The system is now undergoing rigorous analysis by researchers, and any insights should help inform how we can use off-the-shelf supercomputers on board spacecraft in deep-space missions – as well as those used in Lunar and Martian surface bases.
Extending open source innovation to hardware design
It’s important to remember that open source innovation shouldn’t be limited to software, and it’s possible that spacecraft hardware could also be developed by collaborative teams using open source design principles. You might be surprised to learn that this concept is not quite as far-fetched as it sounds; open source hardware instruction set architectures (ISA) do in fact already exist, such as the RISC-V processor currently backed by organisations across the world.
The idea of lowering the barrier for entry for hardware design is actually one of the founding principles of a recent initiative launched by DARPA, which aims to establish a framework that will usher in a new age of innovation for the electronics and microelectronics industry.9
This DARPA project could actually bring the design point of a processor from billions of dollars in research and development costs, to tens of thousands, or even less; and from the two to three years it takes a new piece of silicon hardware to be ready for production, to delivering it in a matter of weeks.
If you consider how much and how quickly the world is changing – it’s becoming clearer that we will soon need a standard set of tools for hardware design that are applicable and accessible. As DARPA states: “we need to break away from tradition and embrace the kinds of innovations that the new initiative is all about.”
Innovation in infrastructure
Some of the computer systems on board the ISS are more than twenty years old. That means they were designed and installed before some of our youngest astronauts were even born.10 Consequently, the computers we use on Earth are several orders of magnitude more powerful than the systems that are running in space. As new technologies are sent into orbit, future space missions will have to deal with a modular approach to building computing centres.
This is where, for extended space missions, the concept of ‘composable infrastructure’ comes into its own. By treating compute, storage and network components as individual resource pools that can be provisioned in real time as needed, composable infrastructure can extend the power of systems depending on the needs of specific workloads.
You can think of this as something like a public cloud, where resource capacity is provisioned from a shared pool depending on demand. However, one fundamental difference is that composable infrastructure instead sits on-premises in a data centre – or in space exploration, on board a spacecraft.
It’s vital that spacecraft computer systems are adaptable, as their purpose is likely to change over time. For example, proposed colonisation missions – like Mars One – intend to utilise the landing module as a permanent habitation feature once it arrives on the red planet. Therefore, the computer systems that helped guide the module down to the surface must be able to reconfigure themselves to operate for its evolving role – and this requires major system components like CPUs, memory and storage to be easily composable.
Human capital and AI
All across the world, human time is the most valuable commodity – particularly when it comes to solving problems. Far from stealing jobs, the majority of advances in artificial intelligence and machine learning are designed to automate relatively ‘simple’ processes and free up time for human’s to do what they do best.
Already we have had a taste of what AI can achieve from space, even as far back as the early 2000s with the launch of the Earth Observing-1 (EO-1) satellite, which helps analyse and inform the appropriate response in the event of a disaster, such as hurricanes and volcanic eruptions. In some cases, the systems in place on EO-1 began capturing satellite images of disaster zones before ground personnel were even aware that a disaster had occurred.11
More recently, AI has been used on the Mars Curiosity rover, where AEGIS software is able to identify intriguing rock or soil patches that should be targeted for analysis.12 This significantly expedites the process of collecting data from the surface of Mars as the robot isn’t relying solely on human commands.
It is incredibly expensive to send a human into space. For example, Richard Branson’s Virgin Galactic – which hopes to soon begin conducting commercial flights – charges between $200,000 to $250,000 per ticket for a flight where passengers will experience just several minutes of weightlessness beyond the Karman line where space officially begins.13 Therefore, previous space missions have always needed to consider very carefully which astronauts they send to space, with the ultimate decision often coming down to an astronaut’s ability to act as a ‘jack-of-all-trade’.
In future, it should be possible to automate the computing and engineering tasks that historically astronauts have had to train for. This would mean sending individuals with specialised expertise in science and research in place of the all-rounder astronauts of yesteryear.
Reaching the Moon, Mars and beyond
For success in space exploration in the years to come, we will need to continue along the exponential curve of open source uptake and see advances in the approach to how spacecraft software, hardware and infrastructure is developed and deployed. I believe this is only possible with a truly open and collaborative mindset.
With this baseline in place, advances in AI-driven scientific research have the potential to propel us forwards and take advantage of human knowledge and expertise while machines handle the processes traditionally managed by engineers.
With the number of government organisations and commercial companies involved in the space economy at an all-time high, it is undeniably an exciting time to be involved in space exploration.