Shane and Ben Reflection

We started this project by attempting to implement IMU based positioning. We learned about I2C, Euler Angles, and how difficult it is to work with the Tufts wifi network. Tufts wifi prevented us from connecting multiple different microprocessors to the internet. Additionally, we had issues implementing the math that takes IMU readings and turns it into location data. We eventually ran low on time, and switched to GPS tracking which  only took us two days to implement. It was also stable enough that it performed without a hitch during the demonstration.

Our main takeaways from this project were to fail early and often. We went through numerous iterations of wifi-enabled microprocessors and IMU interpretation systems. We learned from all of our failures and in the end were able to come up with a stable system which worked. Another takeaway was to make sure our objective was inside the scope of the project.  Tackling an open area of research is outside the scope of a two week final project even though it might be an interesting and educational objective. We tried to develop an IMU-based localization system ourselves, despite knowing that that field is still an open area of research.

People Tracking!

Link

The main function of the Marauders Map is tracking people within the campus grounds of Hogwarts School of Witchcraft and Wizardry. As one of the two groups which attempted to tackle the interesting and difficult problem of localization, we wanted to design a robust system.

After analyzing the strengths and weaknesses of a variety of different positioning systems, we settled on GPS positioning, because it is a simple and robust way of getting user location. Most people have smartphones with GPS capabilities, so we decided to design a smartphone app to make use of existing sensors.

A user’s position shown on the marauders map.

We used MIT App Inventor to create an android app which polled the device’s GPS data and pushed it to the Thingworx IoT cloud. We decided to use App Inventor because it allowed us to easily create and test our app in a short period of time. App Inventor made creating a UI easy and intuitive. It also had prewritten code blocks for polling the user’s location and for pushing data to the cloud using HTTP POST and PUT requests.

The key strengths of our system were its reliability, its use of existing sensors and infrastructure, and its easy large-scale deployability. As a locally installed phone application, our system only needed a GPS signal and an internet connection to broadcast the user’s location at a rate of once per second. Since each user is assigned a unique ID, the backend can differentiate between all of them and the app can be installed on a limitless number of Android phones.

The most glaring weakness of this solution is that it does not work indoors. The buildings block any connection to the GPS satellites preventing indoor localization. Another weakness is that the user must personally install the application on their phone to use the system. This puts responsibility for the correct installation and operation of the app on our users instead of on the design team.  Finally, the high update rate rapidly depletes battery and uses up cellular data.

The main financial opportunity of this app would be to provide data for data-driven advertising. Large companies such as Amazon which have not yet entered the Big Brother track-you-everywhere industry might be interested in acquiring us and our technology.

Our largest threat is that people enjoy their privacy and don’t enjoy being tracked. Without user buy-in and consent our system won’t work. Without a large user base, companies like Amazon will not buy us.