These Modular Drones Self-Assemble To Build Cooperative Structures in Mid-Air
The idea of autonomous swarms of intelligent robots — whether working in concert to pollinate plants to increase crop yields, or helping to monitor and protect the environment — is catching on. But what if these robots could also swarm together to build structures like bridges or platforms?
Now that’s a tantalizing image that the General Robotics, Automation, Sensing & Perception Laboratory (GRASP) at the University of Pennsylvania is aiming to make into reality. Using modular robots that can fly, a team of researchers there have developed a system that allows them to self-assemble autonomously in a variety of configurations, in mid-air. Take a look at the ModQuad:
Decentralized Modular System
Inspired by the swarming behavior found naturally in colonies of ants and bees, and in the flocking of birds, the design for the ModQuad incorporates what is commonly called swarm intelligence: the collective behavior of decentralized, self-organized systems, whether they be natural or artificial.
“In biological systems such as ant or bee colonies, collective effort can solve problems not efficient with individuals such as exploring, transporting food and building massive structures,” explained David Saldaña, one of the researchers who worked on the project, on Bitcraze. “Some ant species are able to build living bridges by clinging to one another and span the gaps in the foraging trail. This capability allows them to rapidly connect disjoint areas in order to transport food and resources to their colonies. Recent works in robotics have been focusing on using swarm behaviors to solve collective tasks such as construction and transportation.”
The ModQuad system uses a quadrotor, the Crazyflie 2.0, which was selected due to its low cost, lightweight, agility, scalability and capability to carry larger payloads relative to its weight. Each machine is each equipped only with a single camera and inertial measurement unit as a sensor — these help to measure the craft’s velocity and orientation in real-time.
The cuboid cages surrounding the machines are made out of carbon fiber rods, which are connected via eight 3D printed ABS connectors. To get the cages to snap together snugly while in motion, there is a docking mechanism in the modular frame, made out of rare-earth (neodymium iron boron) magnets, which act to lock the entire structure together without the need for an external power source.
In addition, the team developed a docking software that allows pairs of these flying bots to align accurately and dynamically attach to each other in mid-air.
“In our approach, we control the attitude of the structure in a decentralized manner,” wrote Saldaña. “A modular attitude controller allows multiple connected robots to stably and cooperatively fly. The gain constants in our controller do not need to be re-tuned as the configurations change.”
According to the team, the system’s decentralized modular attitude controller increases stability and permits linked modules to fly in a cooperative fashion and collectively maneuver themselves in a three-dimensional environment. In addition, the system uses a “new cooperative localization scheme” that allows each drone to utilize measurements obtained by other machines.
In testing this system, the researchers developed some efficient ways to get pairs of the robots to dock together. One method involves a machine waiting in hovering mode, while a second bot executes a docking procedure. Both of these hovering and docking modes lets multiple machines to dock while in flight and appears to be one of the quickest paths to assembling massive structures in three dimensions.
During their experiments, the team also found that with a change of magnet shape, these modular bots could be arranged in different configurations that would allow for the transport of lighter objects, such as a coffee cup. Watch the “Flying Gripper” go:
Emergent Intelligence
As with the lab’s previous projects in precision farming, the overall sum of these modular robots is greater than its parts. While the individual bot might only follow a simple set of rules, rather than being governed by a centralized center of command, interesting things begin to happen when these individual agents interact with each other in what becomes a very complex system, leading to the emergence of what appears to be “intelligent” behavior across the system.
The ant colony is one example of this maxim: the individual ant itself is one relatively simple components in the complex system of the ant colony. Put these individual ants together, and complex behaviors — such as nest-building, food foraging, raising aphid “livestock,” competing with other colonies and burying their dead — arise.
One can only imagine what might emerge with a robotic collective. Besides giving a potential boost to the construction and transportation industries, or warehouse automation, such self-assembling robotic swarms could also assist in emergency situations such as when a building catches fire, or when a natural disaster strikes. Swarms of these intelligent bots could form a bridge to rescue survivors from burning or collapsed builders more quickly, and without threatening the lives of human responders. When combined with other technologies, such systems could also form mergeable, “self-healing” robotic nervous systems in the air, capable of adapting to changes quickly and autonomously. As the team works to refine this modular robotic system, more unexpected possibilities will likely materialize.
Images: University of Pennsylvania