The mission to create a searchable database of Earth's surface | Will Marshall

Four years ago, here at TED, I announced Planet's Mission 1: to launch a fleet of satellites that would image the entire Earth, every day, and to democratize access to it.

The problem we were trying to solve was simple. Satellite imagery you find online is old, typically years old, yet human activity was happening on days and weeks and months, and you can't fix what you can't see. We wanted to give people the tools to see that change and take action. The beautiful Blue Marble image, taken by the Apollo 17 astronauts in 1972 had helped humanity become aware that we're on a fragile planet. And we wanted to take it to the next level, to give people the tools to take action, to take care of it. Well, after many Apollo projects of our own, launching the largest fleet of satellites in human history, we have reached our target. Today, Planet images the entire Earth, every single day. Mission accomplished. (Applause) Thank you. It's taken 21 rocket launches -- this animation makes it look really simple -- it was not. And we now have over 200 satellites in orbit, downlinking their data to 31 ground stations we built around the planet. In total, we get 1.5 million 29-megapixel images of the Earth down each day. And on any one location of the Earth's surface, we now have on average more than 500 images. A deep stack of data, documenting immense change. And lots of people are using this imagery. Agricultural companies are using it to improve farmers' crop yields. Consumer-mapping companies are using it to improve the maps you find online. Governments are using it for border security or helping with disaster response after floods and fires and earthquakes. And lots of NGOs are using it. So, for tracking and stopping deforestation. Or helping to find the refugees fleeing Myanmar. Or to track all the activities in the ongoing crisis in Syria, holding all sides accountable. And today, I'm pleased to announce Planet stories. Anyone can go online to planet. com open an account and see all of our imagery online. It's a bit like Google Earth, except it's up-to-date imagery, and you can see back through time. You can compare any two days and see the dramatic changes that happen around our planet. Or you can create a time lapse through the 500 images that we have and see that change dramatically over time. And you can share these over social media. It's pretty cool. (Applause) Thank you. We initially created this tool for news journalists, who wanted to get unbiased information about world events. But now we've opened it up for anyone to use, for nonprofit or personal uses. And we hope it will give people the tools to find and see the changes on the planet and take action. OK, so humanity now has this database of information about the planet, changing over time. What's our next mission, what's Mission 2?

In short, it's space plus AI. What we're doing with artificial intelligence is finding the objects in all the satellite images. The same AI tools that are used to find cats in videos online can also be used to find information on our pictures. So, imagine if you can say, this is a ship, this is a tree, this is a car, this is a road, this is a building, this is a truck. And if you could do that for all of the millions of images coming down per day, then you basically create a database of all the sizable objects on the planet, every day. And that database is searchable. So that's exactly what we're doing. Here's a prototype, working on our API. This is Beijing. So, imagine if we wanted to count the planes in the airport. We select the airport, and it finds the planes in today's image, and finds the planes in the whole stack of images before it, and then outputs this graph of all the planes in Beijing airport over time. Of course, you could do this for all the airports around the world. And let's look here in the port of Vancouver. So, we would do the same, but this time we would look for vessels. So, we zoom in on Vancouver, we select the area, and we search for ships. And it outputs where all the ships are. Now, imagine how useful this would be to people in coast guards who are trying to track and stop illegal fishing. See, legal fishing vessels transmit their locations using AIS beacons. But we frequently find ships that are not doing that. The pictures don't lie. And so, coast guards could use that and go and find those illegal fishing vessels. And soon we'll add not just ships and planes but all the other objects, and we can output data feeds of those locations of all these objects over time that can be integrated digitally from people's work flows. In time, we could get more sophisticated browsers that people pull in from different sources. But ultimately, I can imagine us abstracting out the imagery entirely and just having a queryable interface to the Earth. Imagine if we could just ask, "Hey, how many houses are there in Pakistan? Give me a plot of that versus time." "How many trees are there in the Amazon and can you tell me the locations of the trees that have been felled between this week and last week?" Wouldn't that be great? Well, that's what we're trying to go towards, and we call it "Queryable Earth." So, Planet's Mission 1 was to image the whole planet every day and make it accessible. Planet's Mission 2 is to index all the objects on the planet over time and make it queryable. Let me leave you with an analogy. Google indexed what's on the internet and made it searchable. Well, we're indexing what's on the Earth and making it searchable. Thank you very much. (Applause)