Inspiration
This project was inspired by search and rescue missions where people have lost their lives. We wanted to build something that could go into dangerous places so rescuers don't have to.
What it does
BeyondBoundaries is a semi-autonomous rover that can be controlled remotely over Bluetooth. It uses AI-based object detection (YOLOv5) to help search unsafe areas for survivors and other items.
How we built it
- Microcontroller: Arduino
- Wireless control: HC-05 Bluetooth module, communicating with the Arduino over serial
- Drive: 2 modified servos
- Chassis: cardboard wheels and base
- Vision: a webcam running YOLOv5 object detection in Python
Challenges we ran into
- Our motors came from the factory running at different speeds, so we resized the wheels to compensate.
- Our original stepper motors burned out after we accidentally overvolted them.
- Power limits kept us from mounting the webcam on the rover itself, so object detection ran separately from the robot.
Accomplishments that we're proud of
- Building a working robot from scratch
- Getting YOLOv5 object detection working
What we learned
- How to adjust gearing and wheel size to balance mismatched motors
- How to communicate over Bluetooth with low latency
- How to use serial communication with an Arduino
What's next for BeyondBoundaries
- Switching to a Raspberry Pi so the webcam can be mounted on the rover and run detection on board