Abstract

Typical drones with multi rotors are generally less maneuverable due to unidirectional thrust, which may be unfavorable to agile flight in very narrow and confined spaces. This paper suggests a new bio-inspired drone that is empowered with high maneuverability in a lightweight and easy-to-carry way. The proposed flying squirrel inspired drone has controllable foldable wings to cover a wider range of flight attitudes and provide more maneuverable flight capability with stable tracking performance. The wings of a drone are fabricated with silicone membranes and sophisticatedly controlled by reinforcement learning based on human-demonstrated data. Specially, such learning based wing control serves to capture even the complex aerodynamics that are often impossible to model mathematically. It is shown through experiment that the proposed flying squirrel drone intentionally induces aerodynamic drag and hence provides the desired additional repulsive force even under saturated mechanical thrust. This work is very meaningful in demonstrating the potential of biomimicry and machine learning for realizing an animal-like agile drone.

Video

Flying squirrel inspired design

Flying squirrel can fold and unfold its wing to control aerodynamic force. We follow this starategy from nature and mixed with conventional quadrotor system. Our flying squirrel drone can operate as conventionial drone but can unfold its wings to gain aerodynamic force when needed.

Fixed control test

Silicone mebrane wings

Custom built elastic silicone membrane wings that mimics the flying squirrel in real world.

Transition of wing

When drone following acceleration trajectory it fold its wings to accelerate more quickly. And to follow sudden deceleration trajectory it automatically unfold its wings to gain more aerodynamic drag force.

Learning from human demonstration

We first obtain offline dataset to be pretrained using supervised learning from human demonstrated data. Next we train with simulator using pretrained policy as base controller and learn the auxiliary policy for stable and fast learning performance. Star marked encoder, decoder is fixed during RL training.

Citation