On May 6th, 2017, I graduated from the University of Minnesota, Duluth with a degree in Mechanical Engineering. I had been looking forward to that day for a long time, but not entirely because I would be graduating. While working on a quadcopter earlier that semester, I had the idea to make my graduation cap fly — an idea that had me grinning mischievously for some time.
To achieve this, I needed to make significant modifications to the cap including reinforcing the structure, routing and mounting the electronics, and improving safety. The graduation cap is made of cardboard, so I designed and manufactured a dimensionally similar aluminum frame to which the motors, flight controller, and other components could be securely attached. The aluminum frame was then fixed to the top of the cap at multiple locations to increase the overall stiffness.
The electronics required for a quadcopter include power distribution board (PDB), lithium-polymer battery (LiPo), speed controllers (ESC), receiver, brushless motors, flight controller, and their respective connections. I have made several quads over the past three years, many of which were decommissioned, leaving me with a healthy parts bin from which to begin this build. The only purchase made specifically for this project was the 4-in-1 ESC that would help the overall aesthetic of the cap and the ease of assembly.
The frame was designed with hole-spacings that are standard to quadcopter parts. The flight controller, PDB, and ESC all have the same hole locations allowing for the stacking of these components in the center of the frame. Several 5mm M3 nylon standoffs were used to provide the clearance between the boards and allowed discrete wire routing, improving the overall look of the cap. With these components consolidated, the motors themselves only required longer wires to reach the ESC.
Mounting the battery required cutting away parts of the cap to allow a battery strap to secure it to the frame and another section of the cardboard was removed for the power cable to pass through. With the battery installed, the physical structure of the cap was finished, leaving only tuning and testing.
Tuning a quadcopter means adjusting the PID controller gains until the desired system behavior is achieved. Knowing that I would be flying in somewhat close proximity to others meant that I was looking for a very stable platform that responded predictably, but would not be characterized as particularly agile. The prop-wash generated by the top surface of the cap made this quadcopter more difficult to tune, although current flight controllers are capable handling the additional disturbances.
I finished the quadcopter about a month before the commencement ceremony, so there was sufficient time to test and make changes as necessary. Once I could consistently do flips with the tassel attached, I considered it complete and ready for the ceremony.
My flight inside the AMSOIL Arena at the 2017 UMD Commencement Ceremony (30 sec)
Bonus: Video of my cap doing flips and nearly taking me out
After the ceremony, I converted the quadcopter into an all-wheel drive electric RC car by forming the aluminum frame arms down and swapping out the propellers for wheels. Modifying the hardware only required a vice, but the software was much more challenging. The flight controller and firmware I used don’t have parameters for ground-based vehicles, so I had to adapt the variables to accommodate the very different configuration of an RC car.
Instead of turning by pivoting the front wheels, this car would have to use the relative velocities of opposing sets of wheels and overcome the frictional forces between the tire and the ground. This was by far the most difficult attribute to tune and required significant trial and error to learn specifically how flight controller parameters translate to ground-based turning.
The motors used in the drone/car are rated at 2,600 rpm/V, so the 14.7 volt battery I used would spin the motors roughly 38,000 rpm without a load. The inexpensive wheels I purchased from Amazon have a 3″ diameter resulting in a theoretical top speed of 340 mph (ignoring almost everything important).. Obviously this can’t be achieved, but it still was a fun thought and gave me some idea of what percent to set the max throttle. I assumed 10% would be fine. I assumed wrong. The rubber tires separated from the wheels at around 5% throttle, resulting in extreme vibration and a scary hand-held test. I finally settled on 3%, but even that vibration is violent when holding the car.
After testing in an empty parking lot, I made additional adjustments to the turning behavior and was satisfied with the overall performance. All said, it’s not as fun to drive as it was to fly, but it was an enjoyable modification to make and worked better than expected. I’ll likely be cannibalizing the car for another quad soon, but the experience has me thinking about making a vertical takeoff and landing (VTOL) version.. we’ll see where that goes.