Image: Dancing with Drones/CollMot Robotics

Dances with Drones: When a Flock of UAVs Joins a Human Dancer

The dance-drone mashup reveals how both can be inspired by mathematical concepts.

|
May 12 2015, 12:45pm

Image: Dancing with Drones/CollMot Robotics

Drones aren't usually known for their dancing prowess. But a new performance presents a whole flock of autonomous drones, with no pre-programmed routes, taking their cues from a human dancer in real-time.

European artist-scientist team CollMot Robotics set out to create the piece to offer an alternative vision to UAVs used for military and surveillance purposes. "Artists and scientists can have a say in what you use technology for," choreographer Nina Kov told me. "The goal of Dancing with Drones is not only to do a show, but to compare both human and drone movement, and see how that comes together."

"Dancing with Drones" is the result of a collaborative project between choreographer Nina Kov and a team of biological physicists headed up by Tamás Vicsek from the ERC CollMot Project at the Eötvös Loránd University in Budapest. The project emphasises the peaceful and artistic applications of drones, and sees a self-organising flock of pilotless drones respond to the cues of a human dancer in an outdoor environment.

Image: Máté Nagy/EU ERC COLLMOT

For the project, the team created a system which allowed them to link human movements with the flock of flying drones. "The dancer is basically one of the drones," said Kov, who referenced a wearable device that is strapped to the dancer's hand, and used to communicate with the airborne drones.

The device is equipped with the same tech as the drones, and acts like a "wingless" leader drone, which communicates with the rest of the flock. As the dancer moves their arm, the beacon generates commands that are accepted by the drone flock as commands—encouraging them to self-organise and move accordingly.

The team's current project builds on previous research conducted by the CollMot Project, which is in fact dedicated to the study of collective animal behaviour. In February 2010, the research team published a study on the flight dynamics of homing pigeons. They decked out a flock of 13 pigeons with tiny GPS backpacks, then logged changes in the birds' flight direction.

The group transposed what they'd learned from tracking pigeon flock behaviour onto a group of drones, which displayed their autonomous flocking prowess in a Hungarian field in February 2014.

They then developed a swarm of autonomous drones that moved in a similar way. In a YouTube video, Vicsek explains that "one of the best ways to understand how animals move together is to build robotsdrones, in this case."

But he notes that the process works in both ways, with the observation and study of animals leading to the development of better machines, and the control and manipulation of machines shedding light on collective animal behaviour.

The group's drones are each equipped with a processor which acts like a "little brain", a GPS device to estimate positions, and sensors. They take no cues from a central computer, relying instead on basic flocking algorithms, which are inbuilt into each drone's onboard processor. These are based on what are known as the Reynolds rulesa mathematical concept which encompasses notions of alignment, attraction, and repulsion.

"Repulsion" ensures the drones keep their distance from one another and prevent collisions. "Alignment" helps the drones synchronise their speed when flying together, and "attraction" keeps the drones together as a flock, and prevents any runaway UAVs.

"When these drones are communicating with each other, they decide when to stop and how much distance to keep from their neighbours," explained biological physicist and CollMot Robotics team leader Gábor Vásárhelyi. Although this sounds simple enough, Vásárhelyi noted that it took the research team five years to develop a system that worked not only as a simulation, but also in practice.

The collaboration also sheds light on how a group of drones could potentially be controlled in the future. "What's going to happen when there are many drones in the sky?" asked Kov. "It's important to understand how you can control multiple units smoothly, and how you can work with the logistics of having lots of drones together," said Vásárhelyi.

Image: Dancing with Drones/CollMot Robotics

The project currently features seven to ten drones and one human dancer. Kov, however, is excited by its scalability, which could potentially see a flock of humans moving in sync to a flock of drones. According to her, the mathematical rules of alignment, repulsion, and attraction can apply to both humans and machines.

"In dance there is repulsion when two dancers push each other away. There is alignment when two dancers are dancing in the same direction at the same speed, and there is attraction when one dancer is attracted to the group, or when two dancers come together," said Kov.

She is fascinated by the intersections of human-machine dance collaborations, having previously worked on a project where a human danced with a mini-helicopter. But she remarked on the differences when it came to choreographing drones and dancers.

"When you choreograph you look for a reference point. Is this the body? Is it the theatre you're in, or is it [like in this case] the whole sky? With the drones, you use coordinates as points of reference, this is very different to having an audience in front of you, as you have to think in 3D," Kov said. "For a choreographer, it's also really interesting to have the possibility to think beyond the human body."

Watch more: Motherboard goes FPV drone racing, 'Star Wars'-style, in the snowy Bronx