Hybrid Drones Become more Efficient Using AI

Hybrid Drones Become more Efficient Using AI

Being capable of converting between fixed-wing and VTOL flight, Hybrid aircraft are able to offer many of the advantages of both, which makes them potentially useful for both civilians who can fly them from a farm without needing a landing strip and military which can fly them from mountains, ship decks, dense forests, and urban battlefields.

FREMONT, CA: Hybrid drones that can convert from vertical take-off and landing (VTOL) to fixed-wing flights are becoming more popular for civilian and military applications. Being capable of converting between fixed-wing and VTOL flight, Hybrid aircraft are able to offer many of the advantages of both, which makes them potentially useful for both civilians who can fly them from a farm without needing a landing strip and military which can fly them from mountains, ship decks, dense forests, and urban battlefields. Nevertheless, controlling such designs can be challenging because of how the aerodynamics of such aircraft change drastically during flight. Scientists are now taking the help of artificial intelligence (AI) to help automatically design remote controls for such hybrid drone technology.

An AI Solution: Neural Network

Jie Xu, a doctoral student in computer science at MIT, and his colleagues have devised a way to automatically design controllers for hybrid UAVs. Their developed system can design a single controller for a hybrid UAV’s all different flight modes and apply it to any type of hybrid aircraft.

The researchers used a kind of AI system referred to as a neural network, in which components dubbed “neurons” are data-fed and can cooperate to solve a problem, like recognizing faces. The neural network continuously adjusts the connections between its neurons and checks if the resulting patterns of behavior are better at solving the problem. Eventually, the network discovers which patterns are best at computing solutions. It then adopts these as defaults, mimicking the human brain’s learning process. In this new system, users first design the geometry of a hybrid unmanned aircraft by choosing components from a data set. The system then runs this design via a simulator to compute this design’s flight performance. The simulator considers realistic problems like random sensor noise and delays in control signals. Next, the neural net automatically starts learning how a controller for the UAV can achieve the best performance in the simulation.

See Also: Top Drone Tech Solution Companies

Weekly Brief