Collective Robotics is an example of Ant Colony Optimization in action. Using simple rules, robots can behave in a coordinated fashion without any coordinating mechanisms (in other word, each robot acts out its own autonomous, simple behaviours, and the emergent effect is coordinated effort). Check out the video!
The objective in the transport task is to locate a brightly lit box and move it to a goal location (located at the leftside of the screen). The box is weighted such that at least two robots are needed to move the box. The goal position is indicated with an overhead spot light placed at one corner of the lab environment. The robots do not explicitly communicate and are not centrally controlled. The robots also do not differentiate between objects and other robots.
Six robots starting from a random initial position must locate the box using two forward pointing light sensors. Once a boxside is located the robot determines if the box is between the robot and the goal using an upward pointing goal sensor and either pushes or repositions itself on the box. Repositioning can (and often does) cause the robot to loose sight of the box since the field of view of the box sensors are not omni-directional. This results in the robot having to relocate the box, find a boxside and determine the goal location. Although not an optimal solution, the robots always manage to push the box towards the goal and demonstrates the utility of simple feasible solutions.