Previous Project

Please refer to this page for detail of the flying squirrel drone. This project focust on more advanced trajectory tracking control algorithm based on previous research.

Abstract

Drones, like most airborne aerial vehicles, face inherent disadvantages in achieving agile flight due to their limited thrust capabilities. These physical constraints cannot be fully addressed through advancements in control algorithms alone. Drawing inspiration from the winged flying squirrel, this paper proposes a highly maneuverable drone equipped with agility-enhancing foldable wings. By leveraging collaborative control between the conventional propeller system and the foldable wings-coordinated through the Thrust-Wing Coordination Control (TWCC) framework-the controllable acceleration set is expanded, enabling the generation of abrupt vertical forces that are unachievable with traditional wingless drones. The complex aerodynamics of the foldable wings are modeled using a physics-assisted recurrent neural network (paRNN), which calibrates the angle of attack (AOA) to align with the real aerodynamic behavior of the wings. The additional air resistance generated by appropriately deploying these wings significantly improves the tracking performance of the proposed "flying squirrel" drone. The model is trained on real flight data and incorporates flat-plate aerodynamic principles. Experimental results demonstrate that the proposed flying squirrel drone achieves a 13.1% improvement in tracking performance, as measured by root mean square error (RMSE), compared to a conventional wingless drone.

TLDR; Flying squirrel drone + Neural net wing dynamics = Enhanced trajectory tracking performance

Video

Neural network based wing dynamics

We collect real world dataset by flying the robot in the wild.

Complex aerodynamics of the wing is captured through RNN based neural network.

Detailed model architecture. Since we asume the model as flat plate with angle of attack we embed this physical property in the neural network architecture choice.

Comparison with baseline

We compare our algorithm with flat wing dynamics and conventional drone without wings. Our approach enables 90.5% success rate where wingless drone achieves 9.52%, and flat wing dynamics drone achieves 19.0%.

Real world deploy

We deploy our controller in the outdoor environment with desired speed of 7.3 m/s. We achieve more superior performance compared to baseline.

Corresponding video.

Citation

The website template was borrowed from Michaël Gharbi and Ref-NeRF.