{"id":35483,"date":"2013-04-01T16:28:12","date_gmt":"2013-04-01T20:28:12","guid":{"rendered":"http:\/\/planetsave.com\/?p=35483"},"modified":"2013-04-01T16:28:12","modified_gmt":"2013-04-01T20:28:12","slug":"engineers-build-first-robotic-ant-colony-video","status":"publish","type":"post","link":"https:\/\/planetsave.com\/articles\/engineers-build-first-robotic-ant-colony-video\/","title":{"rendered":"Engineers Build First Robotic 'Ant Colony' [VIDEO]"},"content":{"rendered":"

 <\/p>\n

\"robotic<\/a>
Robotic Ants following light trails (Image credit – The Swarm Lab<\/a>)<\/figcaption><\/figure>\n

Robotic engineers\u00a0have made\u00a0ant-like robots that successfully navigated through mazes using only light\u00a0sensing and a simple ‘random walk’ programming rule.<\/em><\/p>\n

When we are tasked with identifying an analogue in Nature of the “perfect society”, our thoughts quickly turn to insect societies — with bee, termite,\u00a0and ant colonies being the most common examples.<\/p>\n

When it comes to artificially engineering such a society (in the lab, not in the real world), it is the\u00a0latter (ant) insect’s behavior that has most\u00a0inspired and occupied the imaginations of certain robotic engineers. These roboticists have sought\u00a0to replicate the seemingly complex, collective behavior\u00a0of ant colony members, and in particular, the ability of ants to navigate through\u00a0complex networks of\u00a0bifurcating<\/em> tunnels.<\/p>\n

Regarding this behavior, there were two working hypotheses: either the ants have the ability to “measure” the geometry of the bifurcation (which would require relatively complex cognitive ability), or, that this ability is mostly determined by the structure of their environment and the ability to follow (blindly) chemical (pheromone) trails.<\/p>\n

However,\u00a0different species of ant\u00a0use different methods (or types of information) to accomplish these navigation tasks. Some can use visual cues such as the position of the sun or various environmental markers. Some use what is known as proprioceptive<\/em> information such\u00a0the number of steps or body rotations in space. Ants also track chemical signals and social information such as the food loads of their fellow\u00a0colony members.<\/p>\n

And, in navigating these tunnels, the ants also depend on the geometry of their connecting points, or nodes. The angle at which these nodes bifurcate (split into two) can determine the route taken by the ants. Thus, this tunnel geometry also influences the ants’ ability to track a chemical trail to a given food source.<\/p>\n

What is most fascinating about this behavior is\u00a0that ants are\u00a0always successful at collectively coordinating their movements — even when confronted with unpredictable routes — to find the most efficient and fortuitous path through the network.<\/p>\n

Biologists have suspected that some complex form of communication is going on. But if one is an engineer tasked with testing this theory, the mission becomes one of replicating this behavior using\u00a0the least complex solution.<\/p>\n

‘Alices’ – The Engineered Ant Bots<\/h3>\n

Enter Swarm Lab: a robotics research\u00a0group at the New Jersey Institute of Technology. The team, led by Simon Garnier, designed an experiment in which anti-like robots — about the size of a sugar cube or single dye — were presented with a similar navigation challenge. But instead of using the more complex chemical signaling involved in pheromone tracking, the antbots (collectively referred to as ‘Alices’) were designed to\u00a0leave light<\/em> trails that could be detected via built-in light sensors.<\/p>\n

\"robotic<\/a>
two Alices – robotic ants – about the size of a sugar cube (Image credit – The Swarm Lab<\/a>)<\/figcaption><\/figure>\n

A total of ten Alices were placed in alternating fashion in one of two\u00a0mazes: one with symmetrical bifurcations and the second with asymmetrical bifurcations (which more accurately mimics the natural structure of ant tunnels). Their task was to establish the best route from a starting point to a target point with a network of tunnels in between. The antbots were\u00a0programmed with\u00a0very simple rules for exploring their environment based upon observations\u00a0of how real ants move: specifically, a “random walk” but, in the same general direction<\/em> (no complete turn-arounds or looping back).<\/p>\n

Simon Garnier elaborates:<\/p>\n

\u201cThe robots were programmed to move in a straight line for a randomly chosen time before turning with a random angle chosen between +30 and -30 degrees. This provided some flexibility to the behavior of the robot.\u201d<\/p>\n

While observing the experiment, the researchers paid attention to the interaction between the displacement of the robotic ants, their trail laying and following behavior, and, how it all related to the physical structure of the environment.<\/p>\n

Where Ants Do Not Fear to Tred (Where Other Ants Have Been)<\/h3>\n

At first, the ants “chose” the path of least deviation from their trajectory when encountering a “fork in the road”, or bifurcation. But, when the bots detected a light trail (the analogue signal of a pheromone trail laid down by a previous ant bot’s movement), they would spontaneously turn and follow that lighted path.<\/p>\n

\u201cThe robots prefer to go where other robots have been before,\u201d said Garnier.<\/p>\n

Another surprising result of the experiment was that, unlike their previous research\u00a0involving the angles of the forking tunnels, the Alices were able to find an efficient path to the target point without any additional programming for\u00a0geometric sensing of the physical structure (the bifurcating angles) of the “tunnels”. The combination of the two programmed traits (light sensing and the directional random walk) was sufficient for the ants\u00a0to successfully accomplish the navigation challenge.<\/p>\n

A\u00a0classic positive feedback effect was demonstrated: at first, the ants might use any path equally as much as any other. But, as the shorter paths were “marked” more and more\u00a0often (with light trails), soon enough, the bots started taking these shorter paths more often, further increasing the number of light trails, and thus more ant followers, etc.<\/p>\n

Remarkably, the antbots were able to solve the problem of\u00a0avoiding getting lost (which means not finding the right path from the nest\u00a0to the food and back again), or getting stuck in an endless loop (where one follows and reinforces\u00a0one’s own tracks).<\/p>\n

Garnier, in\u00a0a recent\u00a0io9.com article (see link below), elaborated further:<\/p>\n

“Their networks of trails are not symmetrical: at a fork, coming back from a food source, the path that makes the smaller angle with their incoming direction is more likely to go toward the nest than the path that makes a larger angle. Also, because this path deviates less from their incoming direction, they are more likely to take it because it requires less effort (smaller rotation).”<\/p>\n

This “blind” behavior of the antbots — modeled after real-world Argentine ants — makes a good deal of sense. Argentine ants have poor visual faculties and move too speedily for complex decision-making.<\/p>\n

Although they refer to this as engineering an insect “society”, it is more accurately a replication of a fundamental social behavior (proving the second hypothesis, noted above); insect societies — and their collective behaviors — are actually more complex than what is reductionistically demonstrated here. Still, like many other seemingly complex, collective natural phenomena — like the flocking and schooling behaviors of birds and fish<\/a> — complex behavior emerges from a simple set of programming rules, and sheer repetition.<\/p>\n

Watch this fascinating video showing the robotic ants learning to find their way through a maze (article continues below):<\/strong><\/p>\n