Need for a Waterproof UAV

A major research focus of our lab is utilizing UAVs for remote sensing in aquatic environments. We have been developing sensors (dissolved oxygen (DO) probes), sensor networks, and data processing techniques that integrate with UAVs. However, we have always used commercially available UAVs as our vehicle platform.

Until 2023, we were using the SplashDrone 4 made by SwellPro. This particular drone was capable of carrying a 2kg payload and could land and take off from the surface of the water. The following images show our DO sensor and winch assembly mounted to a Splashdrone 4.


New Laws = New Opportunities … ?

Florida passed a law going into effect in 2023 that prohibits state funded institutions or agencies from using any drone with critical components manufactured in China. This law grounded our lab’s entire fleet of drones (including the SplashDrone) and nearly all of the University’s UAV assets. While this was an initial setback, the new rule gave us an opportunity to design a purpose built UAV, or set of UAVs, for our research. It also created opportunities for us to collaborate with other labs on campus. (like using our UAVs to film whales in the Arctic!!)

Requirements of the Platform

We needed a UAV with the following features:

  • Support an autopilot system that allows for easy modification (Ardupilot, PX4)
  • Waterproof (can be fully submerged with no issues)
  • Land/Takeoff from the surface of water (bonus if it can handle the occasional small wave).
  • Carry and power external hardware (sensors, winch, camera, etc)

Fortunately, there are a few NDAA compliant autopilot manufacturers for UAVs that support a robust ecosystem. We ended up sticking with CubePilot components for our flight controllers, telemetry radios, and GPS because of their moderate pricing and access to a healthy number of resellers in the States.

Buoyancy and Surface Stability

The most challenging goal of making a waterproof drone is making it stable when sitting on the water. As weight is the enemy of performance, adding structures like foam to make a drone buoyant and stable in the water can have a large negative effect on either flight time or payload carrying capacity. A waterproof shell like the one used in the Splashdrone is an interesting option because shell serves as both the airframe and source of buoyancy for the drone. However, manufacturing a similar shell for our lab was beyond our manufacturing capabilities.

Finding an airframe design that is easy to waterproof and is intrinsically stable and buoyant would be the ideal candidate for our application. This is where the coaxial drone could make a lot of sense. In the most basic view, the coaxial drone is a flying spar buoy. If all of your components can fit in a pipe, waterproofing the system only requires the design of special endcaps to seal off the internals. Also, if the center of gravity is adequately below the center of buoyancy, the drone would sit upright in the water.

Design

Due to the inherent risk of designing a novel drone platform, we wanted to rapidly prototype the vehicle so we could quickly find out if the coaxial drone would be a suitable (or even possible) candidate for our sensing platform. In a little under a month, I was able to design the entire system in Fusion 360, 3D print all of the key structural components, and tune the system to achieve a somewhat stable flight with a controlled landing.

Thrust Vector Control

While the opposing rotation of two stacked motors provides yaw control, the vehicle has to change its direction of thrust to control both its roll and pitch axes. Most existing coaxial drones (like the Mars Helicopter) implement a swashplate to control the pitch of at least one set of propellers. To reduce mechanical complexity, we opted to rotate the entire motor assembly along 2 axes to provide pitch and roll control. I was able to design a ball and joint style thrust vector mount that is printed as a single piece (the ball printed inside of the joint). This led to a durable, lightweight, and easy to manufacture control system.

Work In Progress

As of writing this article (Sept 2023), I am still deep in the development of this platform. While I was able to achieve stable flight, there are a few issues that need to be addressed before we can use the drone in an aquatic environment.

  1. Flight Time: The drone is currently not very efficient. While reducing the weight by 20% has lead to a 30% drop in power consumption, the main culprit for the poor performance is the motor selection.
    • The coaxial drone design uses contra-rotating motors (the shaft of the bottom motor passing through the top motor). Most contra-rotating motors on the market are designed with high KV ratings and small propellers. For efficient hover, we need a low kv-rated motor that can handle larger propellers.
  2. Stability in Autonomous Flight: I am learning that the coaxial copter does not behave at all like a quadcopter. While a typical quadcopter can hold a fixed angle, the coaxial copter tends to swing back to horizontal even when applying thrust to move horizontally. While the drone is perfectly stable when controlled manually, the position holding portion of the autopilot’s control system becomes unstable when the drone cannot maintain the desired roll or pitch angle.