NaviNinja

Logo

The inspiration behind the project was to help those who are lost find their way. We both have been lost inside of the CCI building a bunch of times and it's the most awkward thing when you have to ask someone for directions. As shy freshmen, the only solution would be to just not go anywhere and cry. Another example would be for guiding the elderly around the hospital. Sergio's grandmother does not speak english and she always gets lost in the hospital so a tool like that would be amazing for her. The goal with our project is to be able to help those who need it when no one is around.

What it does?

NaviNinja is a smart AI Drone which can perform various tasks in a futuristic way. The drone can recognise a face through its camera and follow the face by automatically navigating itself. Including tasks like take off, self - stabilisation and landing, the drone can follow certain pre defined hand gestures and perform assigned tasks in a give time or a condition. The drone can even recognise body postures and increase it's height and change it's position accordingly.

How it works?

The drone is coded in OpenCV (for video streaming and frame conversion), Tensorflow (To use the ML Model to analyse frames and output certain data on the given input) & DJITelloPy Library(Used specifically for this drone as it allows easier programming interface and flexibility). The drone program also uses the infamous haarcascade frontal face recognition model for analysing the image captured by drone and using them to analyse for faces in the image. The first part of the drone is featuring facial recognition which is done by using the haarcascade for facial analysis and DJITelloPy for Take off, Landing and speed control. The second and third part of the drone is featuring gesture and posture commands which used Tensorflow and Mediapipe for the analysis of the frames captured by the drone and OpenCV to perform commands accordingly. All of the features do use smaller libraries like numpy.

Challenges we ran into?

Some challenges that we ran into are :

The drone cannot perform in darker areas, we plan to fix it using a better hardware for our second prototype.
The code still has some loo-poles such as the code malfunctions when multiple postures or faces are being used in a single frame.
The battery life of the drone is only 10 minutes which is quite small for the use cases, we plan to increase up to 50 minutes for our second prototype by using rechargechable docking stations for drone to charge by themselves.

-We are proud of the facial recognition feature, this can be used in many ways for identification. -We are also extremely proud of the amount of expandability our project can go through. It can go into any field and be useful in one way or another. Wether it be security or even guidance, there are endless possibilities. -With the world of AI breaking new grounds every single day, our system will get smarter and smarter and it will become reality quicker. -Most importantly, we are proud to make a project that can help thousands of people who need it.

What we learned ?

We learned many new things not only from the coding perspective but also from a product development part.
We learned how to integrate AI into basic machines such as drones which allows us to do it with any other machine. This allows us to create prototypes that are not only tied to the body of a drone
We learned that we love working as a team to build things that would help people. There is no better feeling than helping a person in need.
We have enhanced our ability to use the internet to find valuable and trustworthy resources.

We are planning on adding many features to our product: -Speakers so visually impaired people could be guided, -Speakers also for the reason of the drone serving as a tour guide if needed. It can be used as the robot that answers basic questions on the phone, but in real life -Night vision so the drone can work in the dark -Integration with security systems so the drone can serve as a security guard -Recognition for any item, ability to connect to items and identify them ex. Printers, speakers, etc. Can be used to run diagnostics and check for faults. -Fail-Safe features such as: Longer battery life, more powerful but quiet fans, ability to work even without connection.