More than Dots: Wayfinder and the Data

Everything is a matter of perspective … including satellites.

‍

From Earth, we only catch glimpses of them, when the conditions are just right.

‍

In Wayfinder, they appear as a cloud of digital dots.

‍

But up in orbit, their movements are very real – as is the importance of tracking and predicting their movements, to avoid threats to their operations and maximize the value of their perspective of the planet.

‍

Which is why Wayfinder is built, yes, as a world-class visualizer for everything in orbit … but also to do much more than that. And like any useful tool, its value is founded on good inputs and quality craftsmanship.

‍

But don’t take this author’s word for it. Let the engineers, data scientists, and astrodynamicists who built and maintain this dynamic space database take you behind the dots and to the information it maps.  

‍

‍

Principal Engineer Raph Martelles: “At the time I joined, it was stable but the code was ‘written under duress.’ The core functionality was there – we could represent satellites, we could do basic collision detection, we had some foundational work in terms of visualization of the scale of the space debris and space junk problem. There wasn’t a lot beyond that, but that alone was a huge accomplishment.

‍

“Since I’ve been here, we’ve instituted a massive UI upgrade and added: shortcuts to show interesting space scenarios; a tour with guidance on how it works; the ability to filter through satellites of interest based on object type and orbit regime; and good search capabilities. We’ve also made performance improvements to visualization speed.

‍

“Then, we added Crow’s Nest, which lists the 100 greatest conjunction risks in the next 48 hours or so. We added Glint Evader, which is a predictor of the brightness of satellites as they fly over your position – and you can download predictions for their area for a given time in a CSV file. We also upgraded satellite position calculations to use a commonly used algorithm called SGP-4, which feeds a more accurate orbital prediction model than anything we were using and that isn’t too computationally intensive.”

‍

Engineer Shiva Iyer: “Different providers provide data in various formats – two-line elements from the US Government, ephemeris from Planet Labs and other sources and so on. To convert it into a uniform format to store in our Amazon Cloud, ingest it, and display and render in Wayfinder is step one. 

‍

“The second step is to take some of these data products and produce our own satellite ephemeris – a file that contains the positions and velocities of each object in our catalog – and to do it at consistent (one-minute) intervals so that you get a sense for where this thing is in space and where it will be. We produce that every day and publish it on Wayfinder for the entire catalog: 23,000, 24,000 satellites. Owner-operators or enthusiasts can use this ephemeris.

‍

“The surveillance network consists of a number of sensors that we use to track these objects. These objects are moving very fast and the sensors we use all have errors, sensor noise, built in – for instance, on a telescope, there’s a shutter delay. Those are unavoidable any time you have hardware. We also sometimes have to simplify the models so we can make predictions in a reasonable amount of time. So, when we produce our ephemerides, we also characterize their uncertainty in terms of their precision, which can be incorporated into the user’s calculations. 

‍

“The third thing is because LEO is crowded, there are situations where these objects could collide or at least come within 1 km of each other. We make these collision predictions. Once a day our system takes this catalog and produces a list of objects that are predicted to come within 5 km of one another; we compute the probability of collision given the uncertainties we have. If the probability is higher than 10^-4, we consider that to be a high-probability event and display it in Crow’s Nest.”

‍

Chief Scientist Moriba Jah: “In terms of making it robust and scalable, one of the major things in Wayfinder is it is robust, it is 24/7, it has committed staff to keep it going. It’s an operationally relevant application.

‍

“The smoothness of the experience has been refined. The motion of objects has been refined to the benefit of the user. You can see not just object details but price tag on some as well, brightness to people on the ground, ask questions about probability of collision, all in one common operating picture.”

‍

Raph: “We have a lot of interactive capabilities that you can discover. You can see orbital characteristics, launch country, NORAD ID, download its ephemeris, we’ve added some different data sources like NASA SPIRE, detected fires. Just to be able to do all that is pretty amazing on a platform that began as primarily a satellite visualization tool that allows us to see the scale of the space junk problem.”

‍

Chief Technology Officer Mike Moskwinski: “Wayfinder is effectively showing you where the objects are and where they’re going to be over the next five days. We generate these predictions and cross-check them. There’s a lot of data curation and anomaly detection. The basic premise is where are these objects and where are they going to be five days from now. 

‍

“It’s a tool that can also be used for satellite services, companies like KMI, companies that want to service satellites, that will all come within the next 2-3 years as that part of the space economy builds up.”

‍

Engineer Tristan Martelles: “I started out working on the Wayfinder web app, doing fixing on performance issues – implemented modern rendering techniques and hacked them on top of Cesium to improve how satellites are calculated, how they update, display them more efficiently and quickly. We made a lot of progress on that, 4x faster in the first few months. You couldn’t run it on an old laptop and now you can.

‍

“The internal state changes made it possible to white-label the Wayfinder application for the ISS National Lab. It's the same for other white labels, with custom icons and logos and colors.”

‍

Software Engineer Sara Vass: “I have basically taken over the API explorer and docs. I’ve been doing the front-end stuff, not the Wayfinder APIs so much but the tool you interact with, run sample calls and stuff. For me, right now, I use mostly JavaScript; I’ve also done a lot of Python.

‍

“The code base for the APIs were quickly done one-off; they had demos they were trying to get things done for. My first big project was reorganizing so it was easy to update and maintain – I can add a new endpoint to the tool now in like five min.”

‍

Shiva: “How much is uncertain about our knowledge of the space environment is significant. There tend to be information silos in the SSA community, and national security policies and so forth. Although this is such an important asset, it’s a finite one. It seems vast but it’s restricted and there’s so much uncertainty, which makes it all the more finite with all kinds of threats to space missions.

‍

“It’s pretty incredible what we’ve already done in Wayfinder. There are other companies that have visualizations out there but no one else produces an entire ephemeris on all trackable objects and makes it available. No one is doing it. That’s one thing that sets us apart.

‍

“Just as important or even more, Crow’s Nest – producing our conjunction predictions and sharing it, and we do it for free. No one does that. Given the amount of work that’s gone into the infrastructure, making that a free service is amazing. The computational power it takes and the cost is significant. And yet, we are making this available as a service to the SSA community, which is really remarkable.”

‍

Mike: “When you see ISS and what’s docked with it, then zoom out and see everything around it and everything further out, the scale of what we’re doing out there really strikes you. Our flyover prediction APIs are my other favorite part, because of the practical uses in data collection and potential for dynamic tasking of satellites.”

‍

Moriba: “There are people that come from all walks of life that have questions and curiosities about things in space. Some tools satisfy scientific questions like orbital velocity, but they don’t answer policy questions or gee whiz questions. Other databases are just for policy people. Wayfinder can satisfy all, a wide variety of inquiry. That’s been the vision all along.”

‍

Sara: “I care about making tools that are helpful for people. And everyone here is so nice and humble and open to new ideas. As a woman engineer, you see a lot of ego in this space. Here, they’re over it. They just want to make the best solution. It’s really refreshing to be in an environment like that.

‍

“Being able to visualize in a way that’s really accessible is awesome. The fact that we’ve created a tool that can take such hyper-academic concepts and express them, immediately trigger an understanding, is really powerful and impressive.”

‍

Tristan: “My favorite thing is how much stuff we’re visualizing right now. From a technical perspective, I really like optimizing things. That’s my favorite part of the job, I find it incredibly fun. The type of data we’re rendering is very optimizable but hard to optimize. The technologies we’re using to optimize the data are very modern and the precision that’s required to do it, I find very satisfying to work on.

‍

“We want to be best-in-class. The way we get there is by producing applications that a) work, b) are accurate, and c) are performant. We have a special data set and we want to continue to integrate everything we have now, everything we can get our hands on as far as space data and earth data, and make it a one-stop-shop for Earth and space visualization. “

‍

Raph: “The goal is to always have the best-in-class space awareness tool. I don’t know another tool that’s as good as Wayfinder.”

‍