This Single-Motor Drone Flies Like a Maple Seed

Most drones have four or more motors that can be independently adjusted to keep the aircraft stable and steer it through the air. A new drone called SAM that was developed by researchers at the Singapore University of Technology and Design has just one. SAM spins like a maple seed, using a single motor to generate lift while also controlling where it goes. That makes for an exceptionally simple aircraft, but keeping it on course is anything but simple.

SAM, or Samara Seed-Inspired Single-Actuator Monocopter, takes inspiration from the autorotating seeds produced by maple trees. Its motor rapidly spins the entire aircraft around its vertical axis, generating the lift needed to remain airborne. Changes to that same motor are also used to control its direction.

That simplicity has some advantages. SAM is lightweight, inexpensive, and mechanically robust, making the design potentially useful for applications where large numbers of small aircraft need to be deployed. But the spinning motion also creates complex gyroscopic forces that become increasingly difficult to manage as SAM flies faster or attempts more aggressive maneuvers.

To give SAM better control, the researchers developed a system built around Nonlinear Model Predictive Control (NMPC). Rather than simply reacting when the aircraft moves away from its intended path, NMPC predicts how SAM will move over a short period of time and calculates motor commands accordingly.

Doing that accurately required a better model of the aircraft itself. The researchers developed a hybrid rotational model that accounts for the interaction between SAM's spinning body and its roll and pitch movements. The predictive controller then works alongside Incremental Nonlinear Dynamic Inversion (INDI), which uses real-time feedback to correct for wind, modeling inaccuracies, and other disturbances that the prediction doesn't account for.

The researchers tested their controller on three SAM designs with short, long, and ultralight wings. The short- and long-wing versions were flown through circular, figure-eight, and elevated circular trajectories, including paths that sometimes required more actuator input than the aircraft could physically provide. They also subjected the drones to wind generated by fans, while the ultralight version was tested carrying an additional 5-gram payload.

Compared with an existing control method, NMPC + INDI reduced positional error by as much as 39.5% on the long-wing SAM and 37.2% on the short-wing version. The ultralight SAM saw velocity error reduced by as much as 43.5%.

SAM isn't ready to head outdoors on its own just yet. The experiments used an eight-camera OptiTrack motion-capture system to track the aircraft, with the control algorithms running on an external computer. Future work will focus on reducing that dependence on external tracking and moving more of the computation onboard. Doing so could turn SAM's extreme mechanical simplicity from an interesting research challenge into an advantage, particularly for applications such as environmental sensing where lightweight, inexpensive aircraft could be deployed in large numbers.