Autonomous aircraft offer valuable ways to inspect dangerous areas — as long as they can fit.
What if the drone can't get itself inside a tunnel? What if a building collapses, and the only way in is through a small crack?
The drone’s standard configuration is an “X” shape, with four arms stretched out and the propellers set at the widest possible distance from each other.
When faced with a narrow passage, the drone can switch to a “H” formation, with all arms lined up along one axis, or an “O” shape, with all arms folded as closely as possible to the body.
A “T” shape brings the onboard camera, mounted on the central frame, directly to the objects requiring inspection.
Inspired by birds that fold their wings in mid-air to cross narrow passages, the morphing aerial vehicle can retract its propeller arms in flight and shrink itself to fit through thin gaps and holes.
The quadrotor's four propellers rotate independently, mounted on mobile arms. The arms use servo motors to fold around the mainframe.
The control system adapts in real time to any new position of the arms, adjusting the thrust of the propellers as the center of gravity shifts.
The Zurich and EPFL team ultimately wants to develop algorithms that will make the drone truly autonomous and able to automatically choose the best routes in a real disaster scenario.
“The final goal is to give the drone a high-level instruction such as ‘enter that building, inspect every room, and come back’ and let it figure out by itself how to do it,” said Davide Falanga, researcher at the University of Zurich and the paper’s first author.
In an edited interview below, Falanga spoke with Tech Briefs about how close his team is to achieving that goal of true autonomy.
The folding aircraft was the Aerospace & Defense category winner in this year's Create the Future Design Contest. (See the rest of the 2019 'Create the Future' winners.)
Tech Briefs: What inspired you to design the drone in this way?
Davide Falanga: Nature was the primary source of inspiration for this drone. Birds can fly through very tiny apertures, significantly smaller than their wingspan, as, for example, shown in this video. We wanted to replicate such a feature on quadrotors, allowing them to enter areas that would not be otherwise accessible due to the size of the vehicle.
Tech Briefs: When you say "foldable," what do you mean? What forms can the drone take (and not take)?
Davide Falanga: The word "foldable" highlights the possibility of folding the arms around the main body. Each arm is connected to the central part of the vehicle through a servo-motor and can rotate independently around an axis that is parallel to the direction of the thrust produced by each propeller. Therefore, it can assume any form that can be obtained by rotating the arms. For example, it can assume what we call "T-Morphology," where the two front arms are moved to the side, and the two rear arms point backward. Another configuration is the "H-morphology", with two arms pointing forward, two backward. Another possible shape is the "O-morphology", with all the arms wrapped around the body to assume the smallest form. However, it can also assume asymmetric morphologies.
Tech Briefs: How morphable/foldable are today's class of drones?
Davide Falanga: There exist no drones on the market that can change its shape while flying. The only possibility currently available is folding the arms to make a drone easier to transport. Still, once in flight, the morphology cannot be changed. There are scientific publications showing morphing quadrotors. However, these robots cannot fly with any other morphology than the standard "X," since those robots cannot be controlled when they change morphology.
Tech Briefs: Why is this kind of folding capability so important? What applications are possible?
Davide Falanga: Morphing is crucial to unlocking several new applications. Negotiating very narrow spaces is one, and it can be particularly useful for search-and-rescue missions, where it might be necessary to inspect a semi-collapsed building using a robot. Other applications are object transportation and surface inspection.
Tech Briefs: What is the key feature of the technology to enable this kind of folding?
Davide Falanga: The technology behind the folding capabilities of our quadrotor is the combination of servo-motors for actuation and an adaptive control scheme that is capable of guaranteeing stable flight at all times, adapting its parameters to the morphology of the vehicle, that runs onboard the robot.
Tech Briefs: How is this kind of "fold-ability" controlled? Does the drone function autonomously?
Davide Falanga: The drone is currently completely autonomous, meaning that there is no human intervention: All the algorithms required to fly (localization, planning, control) run on the onboard computer, and it is only necessary to specify what the robot should do in advance. The folding is controlled by an algorithm that decides when it is time to change shape.
Tech Briefs: With autonomous controls, can you give me an example scenario of what could be specified in advance?
Davide Falanga: The drone knows exactly where it is and how to reach a desired position; therefore, it is possible to specify in advance a sequence of waypoints that it has to reach. The system will autonomously compute a trajectory that goes through these waypoints, and execute it without requiring any human intervention.
For example, in the task of traversing a narrow gap, we use an algorithm to detect it: Whenever the gap is detected, another algorithm is responsible for computing the sequence of waypoints that lead to a successful traverse, as well as the moment when it issupposed to fold.
It is necessary to specify in advance only which morphology to assume — we are working on automating this procedure — and some parameters of the robot, such as its size, in order to compute the correct distance to the gap when the folding should take place.
Tech Briefs: If the drone is autonomous, how does it “know” its environment?
Davide Falanga: It is equipped with cameras, specifically two. One looks down at an angle of 45 degrees, localizing the robot in space. This camera analyzes the motion of some points of interest in the image (also called "Visual-Inertial Odometry"). The other camera, looking forward, can be used to map the environment (for example, to detect obstacles) as well as to let human operators inspect an area, since the image can be streamed to a laptop computer.
Tech Briefs: And how fast does the aircraft go?
Davide Falanga: The robot currently can fly up to 3 m/s, but it is mostly a hardware limitation, and we are working on realizing a faster version of this vehicle.
Tech Briefs: Where was this drone tested? What's next?
Davide Falanga: We tested this drone in a mockup post-disaster scenario used by the Swiss military for training. We showed them that our quadrotor can get into buildings that are very difficult for humans to enter, and they were very interested in this technology. We are now looking for new scenarios and use cases, not necessarily in the search-and-rescue field.
What do you think? Share your questions and comments below.