I am constantly amazed at how suddenly astronomers learned how to discover new exoplanets. When I was in high school, planets were often assumed to exist around other stars, but nobody ever expected to be able to see one. After all, trying to see an object as small as a planet next to something as bright as a star, from light years away, is like trying to spot a moth circling a floodlight from across the city. And yet in the past decade, scientists have developed techniques and built telescopes in space that are finding them faster than we can study them. In fact, the telescopes are working so fast that NASA are recruiting people like you to sift through the data and find the planets.
One of the discoveries of the modern age of astronomy is that planets are everywhere. Current observations show that stars with planets vastly outnumber those without, and that they come in all sorts of unexpected configurations. But if we want to truly understand planets, we need to study them at all stages of their lives, from when they first begin condensing out of their protoplanetary disk to when they are finally consumed by their dying star. The the birth of a planet is the most interesting, because this is when we can see what goes into its construction, and can watch the forces that work to assemble it. Happily, we have a vast trove of data, collected by the Wide-field Infrared Survey Explorer orbital observatory over a year. Unfortunately, WISE’s data is a hugely detailed map of the entire sky, and finding protoplanetary disks amongst all that is like searching for a needle in a haystack.
To find these needles, NASA have launched a citizen science project called Disk Detectives, in which volunteers from around the world get access to the raw science data. the data is presented through an interface which allows individual users to scan through the images and find planetary nurseries at their own pace, without needing a science background or any special training. “Through Disk Detective, volunteers will help the astronomical community discover new planetary nurseries that will become future targets for NASA’s Hubble Space Telescope and its successor, the James Webb Space Telescope,” said James Garvin, the chief scientist for NASA Goddard’s Sciences and Exploration Directorate.
But why are NASA taking this approach, instead of just processing the data in-house? Computers are good at doing simple repetitive tasks, rapidly and accurately. This is why they are generally used to process large sets of data, but scientific imagery of the sort returned by WISE and other telescopes comes with a problem. It turns out that computers are very bad at understanding visual data, because an object can appear in thousands of different ways depending on how it was imaged. Imagine a cylinder: If viewed end-on, it appears as a circle, but rotate it by 90° and it looks to the camera as a rectangle. Each different angle presents an infinite variety of different shapes in the two dimensional image produced by a camera. Now add in additional complications like shadows, perspective, objects being partially hidden behind other objects, and you’ll start to see why vision is such a tough problem for artificial intelligence researchers, and why such a large portion of your huge human brain is devoted exclusively to your eyes.
The result of this is that we have images that computers struggle to understand, but that humans can sort in the blink of an eye. Traditionally, researchers have used cheap human labour (students) to handle this sort of sorting, but modern collections of data such as returned by WISE and other missions are so big that even a team of hundreds of students working all hours of the day would need centuries to get through it all. We need the capacity of computers, but the visual pattern-matching power of humans. The answer is science’s answer to crowdsourcing: Citizen Science
If you’d like to get involved, just click through to the Disk Detectives website and follow the easy instructions. Do your part, help us find new worlds, and have fun in the process!