Swarmanoid robots work together ... to steal books


August 15, 2011

A Swarmanoid Hand-Bot, with three Foot-Bots

A Swarmanoid Hand-Bot, with three Foot-Bots

Image Gallery (5 images)

Swarms of small, intercommunicating robots are now being eyed up for all sorts of potential uses, including the creation of communications networks for disaster relief, mapping out hazardous environments, or even perhaps helping with the colonization of Mars. Since 2007, a group of European research groups have been collaborating on the now-completed Swarmanoid project, in which a variety of purpose-specific mini robots where programmed to cooperate in order to accomplish a task. Although the bots have been perfecting their book-stealing routine since 2009, a video depicting the task won the Best Video award at last week's 2011 Artificial Intelligence Conference in San Francisco, and was many peoples' introduction to Swarmanoid.

The robots involved in the task included several Hand-Bots, Foot-Bots and Eye-Bots.

The Hand-Bots are capable of climbing up vertical surfaces, using their two mechanical arms and grippers. They are assisted in doing so by a magnetic grappling hook, that shoots out of their top and attaches to the ceiling - using a line attached to that hook, they can then winch themselves up. They also have a couple of ducted fans, that they use to spin themselves around when hanging from their grappling line.

The Foot-Bots can move across the floor on wheels, and feature side grippers that they use for grasping other objects. They have dual cameras and 360-degree infrared sensors, for finding their way around.
Finally, there are the Eye-Bots. These camera-equipped quadracopters are able to search for specific objects, and are capable of attaching themselves to overhead surfaces. They also feature infrared distance sensors and sonar.
The process starts with a group of Eye-Bots flying through a building, each one stopping to hook itself onto the ceiling at various points along their journey. When one of them spies the book in a bookcase, it relays that information back along the line of other parked Eye-Bots, which have formed a connected network. This signal ultimately reaches the deployment area, where the Hand-Bots and Foot-Bots are waiting.

Because the Hand-Bots cannot move across the ground on their own, one of them requests that the Foot-Bots pick it up, and move it to the bookcase. While two of them respond to that request, a number of others position themselves beneath the linked Eye-Bots, creating a ground-based path for the Hand-Bot-carrying Foot-Bots to follow.

Once deposited at the bookcase, the Hand-Bot uses it arms, grippers, grappling hook and fans to climb up the edge of the bookcase, grab the book, then bring it back down. The Foot-Bots then pick it back up, along with its book, and return it to the deployment area. A second Hand-Bot has meanwhile been brought to the bookcase, should it be needed.

The Swarmanoid research project wrapped up last September, although the lessons learned through it could be applied to scenarios such as search-and-rescue missions, or the replacement of human workers in dangerous locations.

The proof-of-concept book-stealing caper can be viewed below.

Source: IEEE Spectrum

About the Author
Ben Coxworth An experienced freelance writer, videographer and television producer, Ben's interest in all forms of innovation is particularly fanatical when it comes to human-powered transportation, film-making gear, environmentally-friendly technologies and anything that's designed to go underwater. He lives in Edmonton, Alberta, where he spends a lot of time going over the handlebars of his mountain bike, hanging out in off-leash parks, and wishing the Pacific Ocean wasn't so far away. All articles by Ben Coxworth

So they\'re learning on-the-fly or was the book stealing enterprise programmed?

Renārs Grebežs

The swarm is impressive: they positively feel like a gang of Daleks from Dr. Who.


@Renars Grebezs - I imagine that only an image of the target book from multiple angles was pre-programmed. The goal, of course, was to locate that target and retrieve it. As you will recall, at the point the video begins, the system had ALREADY done considerable amount of investigation of its environment, which might've taken several minutes to hours. Eventually, we are allowed to see what happens when the system has located its target and begins the process of deployment and acquisition. The entire process was performed via cooperative components that each have specific capabilities and duties. At any moment, the network decides what needs to be done and which type of component is most suitable for the current task and which of those can do the task most efficiently.


With the publication of this article, warehouses around the world have gone on lockdown.


Perfect for stealing small items, provided all the security systems, electronic and human, ignore a swarm of flying and rolling robots infesting a building. ;)

Gregg Eshelman
Post a Comment

Login with your Gizmag account:

Related Articles
Looking for something? Search our articles