Twitter and the winter
Welcome to San Francisco, where the Polar Vortex is characterized by clear blue skies and wind strong enough to blow you off your feet, or bicycle. I learned the latter the hard way. Nonetheless, wind was not the only thing blowing our minds away this winter…
After getting insight on our final project from Google Go’s Cassandra, we turned our heads towards our noisy neighbors- Twitter. At this point, we had brainstormed a couple of ideas, evaluated their merits and demerits, and ranked our varied interests. It was time to consult software engineering experts and discuss the feasibility of our proposed projects. This led my team to Twitter’s Abdellatif Abdelfattah.
Before we even got to the nitty-gritty, Twitter headquarters swept us off our feet harder than it will take you to say ‘Jack is so cool’. Believe me, they do say that a lot! I could rant a lot of things about it- work culture, food, games, events, projects- the only missing to make it perfect was my teammate Haitham. I’ll save this for another post.
Jackie, Abdellatif and I discussed a couple of things. We reviewed our top idea- an interactive AR platform that would inform users of the impact of space missions on the invention of everyday objects around it. There are 4 ways this could work. We could build a recognition-based AR platform, a location-based AR platform, a SLAM platform, or virtual AR. Here are the differences, merits, and demerits:
- Recognition based AR
Typically referred to as Marker-based AR, here we would use a camera to identify visual objects and an overlay is created once the marker is identified. The user would then be able to observe the 3D overlay of the object and specific information about the history of the object in relation to space missions.
Merits: Works with real environment. Can be used multiple times depending on the dynamicity of the users’ environment.
Demerits: It depends on the strength of the camera of the device. The application would need a huge memory capacity. Technical backend engineering required.
2. Location based AR
Here we would use the location feature of smartphones to read the user’s location and provide information relevant to that surrounding. Thus, a person pointing at a huge piece of metal over water in Nevada would have the recognition reject the possibility of it being the Golden Gate bridge as opposed to when the location is San Francisco. More subtle differences can be made with more expertise.
Merits: Will lead to more accurate results. Can also be used in real environments
Demerits: Also depends on the strength of the camera of the device. Technical backend. Huge processing power required of devices using it.
3. SLAM Platform
It stands for Simultaneous Localisation and Mapping. Here we would do something close to what the stargazing app- Nightsky- does. We would read the user’s location then map out things onto his environment with information about their relationship to space missions. This could work with, say, popular streets such as Market, or zones such as Martin Luther King Jr memorial park.
Merits: Can still interact with the environment. Can work with lower quality device cameras.
Demerits: The choices of environments are limited without active visual recognition. Huge processing power required of devices using it.
4. ‘Virtual’ AR
Here we would create our own environment that the user can walk around in and review the objects within it. They would then be able to get more information on objects once selected.
Merits: Relatively more feasible. Can work with lower quality devices. Can work with lower processing power. Still interactive.
Demerits: Environment could be complex but not as dynamic. Is more of a VR project.
All seemed interesting and had their fair share of merits and demerits. From the software engineer’s perspective, “it would take me about 4 months to build an AR platform that could…(describing a SLAM platform)…, and I would therefore recommend you choose a project you can complete within your available time”. Our available time was 2 months and clearly, the ‘virtual’ AR idea stood out. After a long search, we finally had a lead. Jackie was evidently happy.
We left Twitter Hq happier than Svitlana at a BBQ! Now we need to work on the hardest part- getting our hands dirty! We have a couple months to go and we look forward to an epic thrill ride of failing, trying, failing, trying, failing…you get the point.
If you don’t find me at my work desk, at least you know what i’ll be working on.
Work hard blog fun,
Jack
CERES
