How crowdsourcing data analysis helps NASA defend Earth from Asteroids

By Danny Palmer
02 Sep 2014 View Comments
asteroid

Think of NASA employees and what is it that comes to mind? Expert scientists and engineers at the top of their field, no doubt. And of course, joining their ranks as a full-time member of the world's most famous space programme is no simple task.

However, with help from cloud consultancy firm Appirio and the 630,000 members of its TopCoder community of data scientists, engineers and enthusiasts, it's now possible for a much wider array of individuals to work with NASA, in tasks ranging from discovery of asteroids to developing applications for astronaut fitness, all through the power of crowdsourcing.

Further reading

Narinder Singh, president of TopCoder at Appirio, explained how the organisation is set up to be "like fruit flies for testing innovation", referencing how geneticists regularly examine the insect's DNA for the purpose of researching human diseases.

"What NASA gets with TopCoder is the same thing," he told Computing.

"They can test different patterns and really look at things like the impact of disclosure on innovation, how to heighten performance, and getting academic insight, which lets us study innovation in a vigorous way."

Working alongside Harvard University, NASA established the NASA Tournament Lab (NTL) Initiative to use the power of crowdsourced data analysis to "attack really interesting problems".

The partnership started with the TopCoder community aiding NASA to optimise an algorithm for producing medical kits for space missions. "We basically improved the algorithm from three hours to 30 seconds," said Singh.

The projects have become more ambitious over time, with crowdsourcing now harnessed in order to develop an algorithm that can validate potential asteroid discoveries - Asteroid Data Hunter - allowing scientists to concentrate on probable rather than possible discoveries.

"If you want to be dramatic, it's to save the planet; if you want to be more practical, it's to identify jumping off points for exploration," Singh said of the project.

"We've really shown great results, having identified how to position a certain set of arrays to follow the path of asteroids," he added.

With the large numbers of individuals interested in lending a hand with such programmes in their spare time, Singh suggested that NASA could move away from relying on its own supercomputers, instead "democratising participation to an individual computer level" to enable more data to be crunched.

"Our results showed we could do this with less computing power, so we want to continue with software for those algorithms, potentially making it so we can create a smaller version of what NASA does on its massive computer infrastructure on someone's laptop, so they can help participate in that," he said.

Singh explained how rather than encouraging the TopCoder community to participate by offering a large cash prize, it's the appeal of being a small part of a much bigger project that encourages both amateurs and professionals to take part in the scheme.

"That's what we've brought to the table, it's how you can think bigger by breaking down problems into smaller pieces," he said, suggesting that this results in more collaboration than schemes where prize money is offered.

"There are very few challenges where you can say ‘here's a million dollars, build me a solution' because even with that, unless someone is going to explore that area anyway, they're not going to quit their day job and do it," Singh continued.

"We can break large problems down into much smaller problems," he told Computing, explaining that having a large number of people each contributing a small amount of time maximises the usefulness of the effort.

"Like finding asteroids: we broke that down to a dozen challenges. These are bite-sized, so somebody could spend a week and participate."

The nature of crowdsourcing also means that hobbyists can also take part, something that Singh believes is good for both NASA and the TopCoder community as a whole.

"It's not just about professional firms, we have a ton of hobbyists as well as professionals," he explained.

"We have some people who are full-time doing challenges on TopCoder, and we have others who are more interested in a particular challenge or topics like asteroids or the rings of Saturn, so you get interested experts and hobbyists from outside the traditional domains," Singh continued.

"That's the kind of appeal and structure that works well for crowdsourcing projects like these," he added.

Reader comments
blog comments powered by Disqus
Newsletters
Is it time to open Windows?

Computing believes that Microsoft will start offering Windows free of charge by 2017. Is this a good thing for the enterprise?

56 %
15 %
7 %
20 %
2 %