MIT Engineers Build Safe Systems To Control 1000s Of Drones, Robots, Driverless Cars, etc.

Aerial fireworks (the ones that light up the sky) have always been mesmerizing.

From lavish weddings to big birthday parties to Independence Day or Fourth of July celebrations, the sky-lit fireworks have played an instrumental role in making celebrations feel all the more special.

However, they come with a large amount of pollution, can cause unplanned fires, and last a limited amount of time.

This is why they’re beginning to be replaced by drone shows.

These shows make use of hundreds and thousands of aerial drones, each flying in a predetermined pattern that enabled them to be aligned in a synchronous arrangement that can form intricate shapes in the sky.

These devices of spectacular sky views can truly captivate a large audience, leaving them in awe and wanting more—that is until they remain safe.

A few incidents of malfunctions in drone shows have led to concerns about the safety of the people sitting right below them, marking one of the biggest challenges faced by engineers in ensuring the safety of spectators not just for drone shows, but also other “multiagent systems”.

Multiagent systems contain multiple coordinated, collaborative, and computer-programmed agents, such as robots, drones, and self-driving cars—as per MIT’s (Massachusetts Institute of Technology) Department of Aeronautics and Astronautics—which has devised a way to guarantee the safety of such systems in crowded environments.

Through a news release published on its website, a group of MIT researchers and engineers revealed their new method, dubbed Graph Control Barrier Function AKA GCBF+, which leverages the well-established control barrier function theory for safety guarantees.

Through their study (which was published in the journal IEEE Transactions on Robotics), the group of MIT engineers noted that once the method is used to train a small number of agents, the safety margins and controls learned by those agents can automatically be applied to a larger number of agents.

This theory was put to the test, as the group trained a few palm-sized drones to safely execute diverse objectives, which included “simultaneously switching positions midflight to landing on designated moving vehicles on the ground.”

Through the experiment, the engineers found that training done on a few drones could be copied and scaled up to thousands of drones, which were able to safely accomplish the same tasks.


“This could be a standard for any application that requires a team of agents, such as warehouse robots, search-and-rescue drones, and self-driving cars,” said Chuchu Fan, Associate Professor of Aeronautics and Astronautics at MIT. “This provides a shield, or safety filter, saying each agent can continue with their mission, and we’ll tell you how to be safe.”

Other than the obvious safety benefits, MIT’s new method allows engineers to save time and computational expenses by eradicating the need for pair-wise path-planning for each drone, which doesn’t guarantee safety either way.

The method also enables agents to continually map their safety margins and boundaries beyond which they might be unsafe.

Barrier functions in robotics calculate a sort of safety barrier or a boundary beyond which an agent has a high probability of being unsafe, which is susceptible to change at any moment, as agents move among others. When designers chart the course of one agent, they have to take into account the course of all other agents in the system.

“In a drone show, each drone is given a specific trajectory — a set of waypoints and a set of times — and then they essentially close their eyes and follow the plan,” said Songyuan Zhang, the study’s lead author, “Since they only know where they have to be and at what time, if there are unexpected things that happen, they don’t know how to adapt.”

Using the traditional method consumes an unnecessary amount of time, cost, and effort—which MIT’s GCBF+ circumvents.

“Our controller is reactive,” Fan said. “We don’t preplan a path beforehand. Our controller is constantly taking in information about where an agent is going, what is its velocity, how fast other drones are going. It’s using all this information to come up with a plan on the fly and it’s replanning every time. So, if the situation changes, it’s always able to adapt to stay safe.”

Do you think this innovation by the MIT group of engineers and researchers will revolutionize the drone, robotics, and self-driving vehicles industry?

Let us know in the comments below!