Hummingbird

Requesting for Drone
Home
Drone Confirmation
SOS Chat
Figma

Inspiration

In recent years, urban safety has emerged as a critical issue that affects the quality of life and well-being of people around the world. To be specific, late-night safety is still a significant concern, as individuals feel vulnerable while walking alone after dark. Statistics reveal that a presence of law enforcement can deter or prevent up to 90% of potential crimes. We envisioned a solution that can provide a sense of security to people without the constant physical presence of law enforcement. Our solution utilizes autonomous technology to offer a personal safety drone aide, ensuring that individuals are not alone in potentially vulnerable scenarios

What it does

The Hummingbird is designed to provide a discreet yet dependable solution for personal safety. Users can easily request our drone's assistance, which arrives at their location to accompany them until they reach their destination. Utilizing state of the art object detection and multiple object tracking technology, the drone ensures continuous surveillance. Once the journey concludes, the drone autonomously returns to its hotspot location, ready to assist others. If an emergency SOS is triggered by the user, the app will send details of the user's location and appearance to authorities.

How we built it

Here’s a breakdown of the core components and how each contributes to the overall system of our Public Safety Drone, Hummingbird:

Object Detection: At the heart of Hummingbird’s system is the state-of-the-art YOLO (You Only Look Once) model for real-time object detection. For our use-case, we’re using YoloV5s. This model allows the drone to reliably detect people, crucial for initiating tracking and detecting obstacles.
Object Tracking: To maintain continuous tracking of the person it is following, Hummingbird makes use of the ByteTrack algorithm, an advanced multiple object tracking technology. This ensures that once the drone has locked onto a person, it keeps them in focus.
Drone Control: The operational behavior of the drone is managed through a sophisticated control system using the DJI Tello Wrapper. This system adjusts the drone’s position dynamically to follow the person at a predefined safe distance. By utilizing a Proportional, Integral, and Derivative (PID) controller, we’ve made calculations on how to adjust the drones position in real-time based on the relative position of the person.
Front-End Interface: The user interface of the Guardian system is built using React Native, providing a intuitive user experience and the UI was designed using Figma. Key features include:
- Emergency Call-911 Button: A feature allowing users in distress to instantly call 911 from within the app.
- Green QR Code: For additional verification, the drone can linked to a user through a secure QR code. This ensures that the drone services are securely connected to the right user.

Challenges we ran into

We encountered video streaming issues from the drone, including significant packet loss that disrupted the transmission quality.
The drone lacks Wi-Fi capability, preventing it from accessing the internet during operations.
During testing, we faced challenges with broken propellers

Accomplishments that we're proud of

We're happy we made a working prototype of our drone, spotting objects in real-time and tracking users accurately. We're especially glad we could fix all the code issues to get our proof of concept up and running.

What we learned

We learned to troubleshoot hardware issues effectively, such as resolving drone connectivity problems.
Integrating the drone with our frontend taught us to adapt to devices with limited connectivity, like local area network-dependent systems.
Refining our tracking algorithms taught us the importance of precision and the iterative nature of problem-solving.
Dealing with Android build complexities taught us to anticipate and address unexpected challenges in frontend development.

What's next for HummingBird

We plan to enhance the drone's functionality by implementing an advanced collision avoidance system. Moving forward, we want to get a drone with improved battery life, extended operational range, and internet access! Additionally, on a larger scale, we hope to develop a network of multiple drones and drone centers, so that a nearby drone to be dispatched directly to users as needed. Currently, it is out of our scope to do this but eventually, we can.