So after the initial test in my backyard’s jacuzzi, I learned that the balance of the drone was totally off, making the drone’s front float. Strapping some weights I found around my house to the drone actually seemed like a decent solution, so I decided to stick with the practical route; and I bought a container of small stainless steel beads that were both dense and easy to work with. 

I started by pouring exactly 1 pound into two plastic ziplock bags, which I then insert into some empty space available in the drone. Upon dropping the drone into the water again, the buoyancy was noticeably more even, but now it slightly sank at the front. 

So I put some of the foam I took out from the last test back into the front, which balanced the drone well enough. I was now ready to test the drone for real in a larger environment.

The same day, I luckily already had plans to visit my grandparents’ house, which had a decently large swimming pool. I figured that would be a great place to test the drone’s performance in greater depths and with more space to move. 

So I brought all the equipment I needed—my laptop, hex keys, spare beads and screws, controller, and of course the drone—and set it up by the pool in my grandparents’ backyard. 

I turned the drone on, dropped it in the pool, and the rest is history:

So, although I clearly wasn’t the greatest driver, I noticed some key things about the drive that could definitely be improved in the future. Firstly, the drone actually seemed to have a bit of a power imbalance between the two main thrusters, as it would constantly turn a bit to the left side as I tried to drive forward. I had to compensate by constantly stopping to yaw a bit to the right so I wouldn’t hit any walls. Secondly, I noticed that the drone was pretty susceptible to rotating, usually from the resistance of the tether or just from floating around. I didn’t have a roll control implemented on the controller, which also made it a bit more difficult for me to control the drone. In the future, because there isn’t really a reason for a drone like mine to roll in actual application, I could potentially implement automatic leveling using the Pixhawk’s IMU. 

But overall, I really enjoyed this project—I learned a lot about making drones, got some great practice designing more complex systems, and have discovered a new passion for underwater applications of engineering.