Hovercraft

This was the first project I worked on as an undergraduate at the University of Maryland, College Park. The Intro to Engineering class (ENES100), taken by all freshmen in the A. James Clark School of Engineering, is a semester-long interdisciplinary project to build an autonomous vehicle that can perform some sort of manipulation task. The idea is to introduce freshmen engineers to teamwork, give them a broad sense of the discipline, and get their brains pumping. Basic information relevant to the project is taught, but mostly it's just about the project.

When I took the class, the goal was to build an autonomous hovercraft that could navigate into a testing space, find a block designated as a goal, and drop a ping-pong ball onto one of four squares on top of the block. The autonomy required wasn't complex - navigation could be accomplished by either following a black tape-line on the floor or by following an IR beacon, and the correct block could be determined by easy color-sensing or read via radio-frequency.

Together with a team of eight other students, I built the hovercraft pictured above. We built a double-deck body out of Styrofoam and glued on a skirt made of vinyl sheeting. Wiring was done in between the two decks, allowing a sleek but still functional design.

A single powerful fan served to lift the hovercraft. We were unable to find a fan with the appropriate power, so we chose to use a more powerful one and pulse it to create a “hopping”-style motion. Two medium fans on the rear provide forward propulsion, and four smaller fans provide attitude control. An Arduino Uno (hidden inside) ran the system.

Our hovercraft navigated using light sensors to detect a black tape line; sensors in the front and the back detected left-right deviation and adjusted the position of the front and back of the hovercraft independently. In order to provide robustness to differing light conditions, the sensing system is hidden underneath the skirt of the hovercraft and illuminated by several LEDs. Finally, an ultrasonic distance sensor is used to detect the goal block.

We then used a unique mechanism to drop the ping-pong ball using only a single motor. The mechanism, designed by Josh Sheldon, was mechanically linked to rotate as a unit when spun clockwise but not counterclockwise. Thus it could be rotated clockwise until the ball was over the the correct square, and then the claw withdrawn via a counterclockwise rotation to release the ball. The simplicity of this mechanism (whereas many other teams used complex multi-motor designs) still inspires my current research on underactuated systems.

Overall, our design was effective and performed precisely, but at the cost of speed. As is evident in the video, the pulsing of the lift fan created a hopping motion, meaning that every time the hovercraft rose the propulsion fans would have to accelerate it from a stop. The attitude control fans also suffered from this issue, making the overall motion very slow.

The final evaluation for the project was a class-wide competition, with three runs for each hovercraft. Our hovercraft placed second, completing two perfect runs and one almost perfect run despite its slow motion, and we were given an award for Best Design, for our sleek, functional appearance.