How is it possible that the seemingly disconnected actions of individuals add up to a complex group behavior like the schooling of fish or the erratic flight of birds?
The collective abilities of swarming animals may seem disjointed, even incredible, to the untrained eye, but biologists are in possession of some intriguing insights which help explain the managerial method behind the madness.
In general, understanding swarm behavior comes down to this; the swarm and its outcome are the result of nearly countless interactions between individuals, each of whom are acting on a few simple rules of thumb which contributes to a whole which is called 'self-organizing.'
Also called 'swarm intelligence,' it amounts to very simple creatures acting according to simple rules based on localized information. No one member of a swarm directs any other member, but through interaction, very complex behavior is coordinated without what we might consider management.
Called Collective Cognitive Robots (CoCoRo), a group of 40 3D printed mini-submarine robots have been made to work together to perform complex tasks. Their behavior is based on the concept of "swarm robotics." The idea is that a group of simple little robots, all capable of working in concert according to elemental rules, carry out complex tasks at a much lower cost than a single advanced robot.
Craig Reynolds, a researcher in the science of computer graphics, described how such rules might work back in 1986 with a steering program he called 'boids.' Reynolds' simulation used generic birdlike objects, or 'boids,' to carry out three instructions. They were: avoid crowding nearby boids, fly in the average direction of any nearby boids and stay relatively close to any boids nearby. Once the simulation was run on a computer screen, the result was a reasonably accurate simulation of flocking behavior which included seemingly unpredictable movements. Reynolds now works on gaming research and has created an algorithm to simulate – in real time – as many as 15,000 birds, fish, animals or people interacting virtually. His demonstration of self-organizing models which ape swarm behavior has been the basis for similar work by today's robotics engineers.
The work was a foundational breakthrough which led to the CoCoRo project. The CoCoRo project, which has been funded to the tune of $4.1 million over the course of the last three years, is now set for a final testing phase during September of 2014. Five European universities (the University of Graz in Austria, the University of Stuttgart, the University of York, the University of Brussels and the Scuola Superiore Sant'Anna University of Pisa) were involved in the project and handled various roles. The building of the robot prototypes was carried out by the Department of Biorobotics at the Italian partner location in Tuscany.
"We've been using 3D printers for several years," says Stefano Mintchev, one of the student engineers involved in the project.
Each of the bio-inspired, robotic structures which replicate animals and living organisms in general in their shape and function, are 10 inches in length.
The largest robotic swarm ever made, CoCoRo is a highly efficient and cost-effective system.
"It is a totally different approach," Mintchev said. "Because swarm robots react in a much more 'robust' manner to complex and variable situations such as those that they would encounter underwater – for example identifying an object or a source of submarine pollution – a single robot would be hard pressed to explore a large space and follow the track to the source, while a group of robots can much more easily follow tracks that are non-linear and in constant motion."
The CoCoRo system is actually made up of several subsystems. One of them, a floating base station, injects GPS data into the system. Another, a 'self-aware' ground swarm, performs the target task while a 'relay swarm' bridges the communication gap between the other elements.
The relay swarm consists of submarines which perform the underwater swimming and they include a range of on-board sensors which help them interact with the environment.
The developers of the CoCoRo project say the swarms will likely be used for environmental monitoring of water pollution, to detect the effects of global warming and to search out problems like lost jetliner black-boxes, toxic waste dumps and polyp fields in ocean environments.
Each robot, equipped with sensors to aid navigation and analyze its surroundings, can react to light or changes in temperature. The robots take care of business via a system of propellers and artificial swim bladders used to create the proper buoyancy. The robots also have three separate communication systems: sound sensors which act like sonar, a system of blue LED lights used for their ability to penetrate through water over large distances and bioelectric sensors which mimic those of certain fish species.
The proprioceptive and exteroceptive sensors are hooked up to a fault tolerant hardware abstraction layer, appropriately named 'HAL,' and middleware to aid in that fault identification and recovery.
Research is also being done to study whether or not information must be processed at an individual or on a collective level. The team is studying systems based on self-awareness and collective self-recognition.