Aquasense

Introduction

The Inception (of the idea)

The idea for Aquasense stuck our team as soon as we saw one of the themes. Swimming is something that is highly physical and reliant on senses. With this project, we wanted to conceptualise a wearable which a visually impaired person can wear to swim autonomously, without any aid.

The initial design was as follows:

• A camera module (preferably integrated with the microcontroller to make the footprint smaller) was to be used in a small enclosure, that can be worn around the chest on a band.
• This enclosure was also to have the battery pack that powers the entire setup. Portable power banks seemed like a viable option, although we were also looking into rechargeable lithium ion solutions.
• The MCU on board was supposed to run a small tinyML based model for all the work behind the scenes.
• This setup was also proposed to have an audio output, and we were thinking about various ways to give more feedback to the user.

The challenge

Most of the assumptions we made were based on one assumption -the hardware being given for the competition. And with our requirements, the Google Coral Dev board was the hardware of the choice. Unfortunately, that ran out of the giveaway, and we were given the Person Sensor by Useful Sensors. This was a little difficult for us, since now we had to go with an alternate approach.

Another issue was that the Person Sensor came without any connector. And we tried hard to find it in our local market, but unfortunately we couldn’t find it. We tried to import it from overseas, but the shipping alone was very inconvenient and expensive.

Thanks to the awesome discord admin for this challenge, we are still able to publish our next steps on how we’d like to proceed with this project. On contacting them, they advised us to let know on how we intend to move forward with this project once we get the connector.

• Since we now have a person sensor, we intend to use it for all its intents and purposes -navigating the pool area for a visually impaired person.
• Lane detection still needs a camera. And while we are looking into smaller modules, we really would like the overall footprint of the device to be the same.
• Beeps can be used instead of audio prompts, as that would add simplicity to the project.

The code we started working on was based on Google Coral Dev, so that also needs a whole new level of upgradation. And we realise that this setback effectively might set us out of the prize pool, but completing the project with proper utilisation of the project is more important to us.

We would again like to thank the discord admin for this competition to hear us out and support us for this!