Welcome Your IET account
Drones flying in the sky

Researchers develop ‘neural lander’ to land drones smoothly

Image credit: Colicaranica | Dreamstime.com

Researchers at the California Institute of Technology (Caltech) have developed a system that uses a deep neural network to help autonomous drones ‘learn’ how to land more safely and quickly, while using less power.

The system was created by Caltech’s Center for Autonomous Systems and Technologies (CAST) in a collaboration between artificial intelligence (AI) and control experts. The “neural lander”, is a learning-based controller which tracks the position and speed of the drone, and modifies its landing trajectory and rotor speed accordingly to achieve the smoothest possible landing.

“This project has the potential to help drones fly more smoothly and safely, especially in the presence of unpredictable wind gusts, and eat up less battery power as drones can land more quickly,” said Soon-Jo Chung, a professor of Aerospace at the institute.

For many experts developing unmanned aerial vehicles, landing multi-rotor drones smoothly remains a challenge. This is due to complex turbulence being created by the airflow from each rotor bouncing off the ground as the ground grows ever closer during a descent.

According to experts, such a turbulence is not well understood not is it easy to compensate for, particularly for autonomous drones, hence why take-off and landing are often the trickiest part of a drone flight.

Typically, drones wobble and inch slowly toward a landing until power is finally cut, and they drop the remaining distance to the ground.

Deep neural networks (DNNs) are machine learning systems inspired by biological systems such as the brain. The “deep” part of the name refers to how data inputs are churned through multiple layers, each of which processes incoming information in a different way to tease out increasingly complex details.

These DNNs are capable of automatic learning, which makes them ideally suited for repetitive tasks.

The Neural Lander system is tested in the Aerodrome, a three-story drone arena at Caltech's Center for Autonomous Systems and Technologie

The Neural Lander system is tested in the Aerodrome, a three-story drone arena at Caltech's Center for Autonomous Systems and Technologies

Image credit: Caltech

To ensure that the drone flies smoothly under the guidance of the DNN, the team employed a technique known as spectral normalisation, which smooths out the neural net’s outputs so that it doesn’t make wildly varying predictions as inputs/conditions shift.

During the tests conducted by the team, improvements in landing were measured by examining deviation from an idealised trajectory in 3D space.

Three types of tests were conducted: a straight vertical landing; a descending arc landing; and flight in which the drone skims across a broken surface – such as over the edge of a table – where the effect of turbulence from the ground would vary sharply.

According to the results from the tests, the new system decreases vertical error by 100 per cent, allowing for controlled landings, and reduces lateral drift by up to 90 per cent.

Furthermore, their experiments showed that the new system achieves actual landing rather than getting stuck about 10 to 15cm above the ground, commonly found in unmodified conventional flight controllers.

“With less error, the Neural Lander is capable of a speedier, smoother landing and of gliding smoothly over the ground surface,” said Yisong Yue, assistant professor of computing and mathematical sciences.

Also, during the skimming test, the “neural lander” produced a much smoother transition as the drone transitioned from skimming across the table to flying in the free space beyond the edge.

The new system was tested at CAST’s three-storey-tall aerodrome, capable of simulating a nearly limitless variety of outdoor wind conditions.

“This interdisciplinary effort brings experts from machine learning and control systems. We have barely started to explore the rich connections between the two areas,” added Caltech AI expert Anima Anandkumar.

Other than its commercial applications, the team at CAST have filed a patent on the system, adding that the system would prove crucial to projects currently under development at the institute, including an autonomous medical transport that could land in difficult-to-reach locations (such as a gridlocked traffic).

“The importance of being able to land swiftly and smoothly when transporting an injured individual cannot be overstated,” said Morteza Gharib, Hans W Liepmann professor of Aeronautics and Bioinspired Engineering and the director of CAST.

Sign up to the E&T News e-mail to get great stories like this delivered direct to your inbox every day.

Recent articles

Info Message

We use cookies to give you the best online experience. Please let us know if you agree to all of these cookies.


Learn more about IET cookies and how to control them