PHOTOGRAPH AN ICEBERG!
Inspired by similar projects such as Solar-voyager and Seacharger, I wanted to build a autonomous boat but instead of sailing across the Atlantic I wanted to up the challenge. The Arctic! One of the most inhospitable places on earth! So the objective is to sail to the Arctic, take a photo of an iceberg and return home, simple!
Why
It’s mainly an educational project and challenge for myself. The ultimate goal is to learn more about machine learning, and with the data being collected from a robotic boat, it’s a good platform to implement some of these techniques. Using cameras, I’ll be able to use this data for object recognition and collision avoidance, hopefully so it avoids icebergs or anything else in the water! Anomaly detection can be ran against sensor data to detect problems before they get worse, and machine learning can be used to optimise the running of the boat. But before the fun with machine learning can begin, the boat has to be built. It’s a multifaceted project that involves the understanding of hull design, hydrodynamics, fiberglassing, manufacturing techniques, 3D printing, as well as electronics, computing and programming to actually control the boat.
The Journey
It’ll be tested in and around the UK, and then at colder latitudes such as Norway and Iceland. I’ll be paying particular attention to how it performs in these conditions.
So how will we know when it reaches the arctic. The ultimate goal is to get a photo of an iceberg, otherwise it could just be floating in the arctic circle with miles of sea around it and that’s no fun! But navigating the boat via coordinates is tricky. If the coordinates are too far north then the boat might crash in to an ice shelf or get stuck in ice. If the coordinates are too far away, then all will be photographed will be sea! So using machine learning, along with satellite imagery we’ll train the boat to detect icebergs and ice shelves so it can navigate it’s way through the ice!