TellUs - Space Apps Challenge

TellUs| Home Planet at Your Fingertips

Awards & Nominations

TellUs has received the following awards and nominations. Way to go!

Global Nominee

Home Planet at Your Fingertips

Develop a user-friendly application or tool to discover, visualize, and analyze NASA Earth data for monitoring our home planet.

Tell.Us - News and Data from Space to Your Fingertips

Summary

lat. Tellus = Earth, ground -We desire to provide a mobile app, which pings entrepreneurs, academic members and the general public alike with high-level space & Earth messages and news according to their demands and interests. Those may be related to datasets, new discoveries or e.g. calls for hackathons ;)Besides this "pop-up"-space-newsletter, we developed niche ideas for each of the 3 customer types. For B2B (and NASA itself) the user data can be used to target audiences, for academia we provide ML "database-tinder"-function to match easily with datasets. For enthusiasts we want to implement 1 min. quizzes aka before entering Instagram we will challenge you with fun Earth science.

How We Addressed This Challenge

We set up a website https://tellusspaceapp.wordpress.com/ to pool all steps and achievements. For the app we developed a business scheme (monitoring business-relevant developments from space; analysis on collected space-and-Earth-science-engagement-data), the profiling queries and a prototype with a sample "1 min. quiz". We are very driven to develop the prototype further! With a bit more time, the next steps would now be to develop a recommendation algorithm which pulls information from space agency, universities, tech-company, media websites and push-notifies the user.

Harsh also developed an augmented-reality add-on, that is used to visualize some datasets in real time on a 3D model of the earth virtually using the mobile camera.

Alex looked more into the NASA datasets and how we can provide information for researchers effectively. She soon found herself quite overwhelmed, that is why we decided we need to build a "dataset-tinder". The idea is that a researcher can put in criteria like e.g. wavelength, orbit, resolution and upload sample images and is then automatically matched with NASA datasets via a machine learning algorithm. Ideally, the algorithm should output the best-fitting datasets without much "metadata" from the researcher itself and that the sample input-data can be of any format. But due to time-constraints (and being a little bit overwhelmed by the Earth Data-sets) we started to establish a small dataset from NASA Earth Observatory images and train quite successfully for such a small data set for multi-label classification.

We hope to achieve an integrative, easy and fast solution for satellite data engagement and awareness without thinking only about 1 specific target group.

How We Developed This Project

We first signed up for the hazard detection, but soon found that for most people smartphones have become more impactful than weather in their daily lives. At the same time, we saw that there is a lot of complexity in processing satellite data and understanding Earth science better - and especially on mobile devices our attention span is finite.

To solve this we need a "mediator". A platform, which delivers small but interesting & important information daily and is a portal for more support and learning. As a team of three with not so much coding experience, we developed this approach quite creatively and due to Covid-19 remote working, we even sometimes diverged quite a bit from each others ideas. Still, step-by-step we put up a wholesome approach and cheered each other up enormously. In the end even weird file-extensions like .hdf or .pkl made more sense!

From the technical side, we used basic Wordpress and web development tools. The algorithms were built with Python in Jupyter-Notebooks. For the app mockups, we used Adobe XD to make a working prototype and Spark AR Studio for the Augmented Reality demo.

For the final app we will code it separately for each mobile platform (Android & iOS) using Java and Swift programming languages. For implementing live AR visualizations of datasets we will use AR Core and ARKit respectively for Android and iOS.

How We Used Space Agency Data in This Project

We mainly downloaded NASA images from the Earth Observatory website to train the classification model. Then we peaked into some Earth Data datasets from LAADS and found a few pictures, which we could also run through our trained model to match researchers' input data and NASA Earth Data with a future fully hard-coded "dataset-tinder-algoritm" in the future more closely.