Like watching your child take their first steps, the feat that the collection of "Swarmanoid" robots achieve brings tears to the eyes. Although I did not help create these bots, I can appreciate the advancement in robotic collaboration and artificial intelligence they show.
Original Swarm-Bot concept
Professor Merco Dorigo, of the Université Libre de Bruxelles, Belgium, ran a project that lasted 42 months, completed on March 31 in 2005, that was to take a new approach to design and implementation of self-organizing and self-assembling artifacts. The outcome of the program were Swarm-Bots. The ground based, cylindrical bots, had simple rules that were used to perform tasks. The bots demonstrated the ability to map an area, and move objects around in unison. With the limitations at the time in mind, Dorigo talked about what he would like to see next, "In the future we might have robots that actively seek help from others when they come up a problem they can't solve alone."
Since October 2006, Dorigo and his team set out to make the bots perform collaborative tasks into 3D space. The Swarmanoid project began. Swarmanoid's goal was the "design, implementation, and control of a novel distributed robotic system." The system was to be composed of dozens of different types of autonomous robotic platforms. Eye-bots that can fly and see, hand-bots that can pick up foreign objects, and foot-bots that can traverse the surface and transport other bots or objects.
When the Swarmanoid is first activated, they know nothing about their surroundings. They only know one thing, they need to retrieve the target, in this case a book. The eye-bots explore and search each room. Each eye-bot moves to a location that is at the extents of their communication range with the last bot, they a chain of eye-bots form. The eye-bots attach to the ceiling when done moving.
Once the target is found, the foot-bots are activated. The foot-bots form another chain of bots from the homebase(nest) to the book(target). Two foot-bots attach to a hand-bot and transport it to the target. Additional, redundant, hand-bots are also being transported.
Once at the location, the hand-bot is dropped in front of the book. Since the book is on a shelf, the hand-bot shoots a magnetic tether to the ceiling, for support, and climbs to the book's level. After grabbing the book, the hand-bot lets go of the shelf and repels to the ground. The foot-bots reconnect and transports the hand-bots and book back to the base.
By the way, each of the bots are acting autonomous.
Communication between bots is handles with different color LEDS, flashing of the LEDs, and an infrared signal. The bots are always in constant communication. This allows the group to detect and adjust to real-time errors, such as a broken bot. Dorigo stated that the communication is like talking in a crowded area. Just as a single person's voice can be isolated, the bots can detect particular counterparts. Direction and distance can also be calculated from the various types of communication.
Dorigo said the next step is real world applications, like search and rescue. But for now, the team is actively improving the communication abilities, speed, and capacity of the bot's actions. Like taking first steps, the Swarmanoid is still in its infancy.
Cabe
Swarmanoid won best video in the Conference on Artificial Intelligence in San Francisco. (AAAI, Association for the Advancement of Artificial Intelligence.)
