The Ant Algorithm, also often called Ant Colony Optimization (ACO) in the literature, is an algorithm for the approximate solution of complex optimization problems. This method of combinatorial optimization, also known as metaheuristics, is based on the behavior of ants in their search for food.

The model is inspired, as the name suggests, by nature. The ant just runs off and looks for food. It marks its way with volatile scents, on the one hand so that it can find its way back, on the other hand so that other ants can follow its way to the food if it is successful. If another ant now finds a shorter path, the scents on its route are correspondingly more intense because the release was less long ago. Consequently, subsequent ants orient themselves to this path, which further intensifies the scent trail.
If the original fastest path is suddenly blocked, the returning ants and the intensity of their scent trails can be used to determine the fastest path again. In this way, the best way can be found within a short period of time and communicated in a way that is comprehensible to all – without any higher-level coordination.

From nature to industry

As early as the 1990s, researchers were working on transferring this know-how from nature to industry. Today, the ant algorithm is used in many scenarios, such as route optimization in production and intralogistics, switching of communication channels such as phone lines and Internet connections, or transport management in logistics.



What all these areas of application have in common is that the highly complex route planning is not elaborately calculated and centrally coordinated, but rather each object involved is capable of independent decision-making and action and decides on the optimal solution based on the algorithm and the current data of the other objects.

Ant algorithm in intralogistics and production

The internal flow of materials is a complex network with countless influencing factors, such as incoming orders, route occupancy, cut-off time, availability of goods, or machine and personnel availability. The more variables and parameters are included in a control process, the more complex its mathematical calculation becomes.

Various mathematical methods allow these calculations, whereby a distinction can be made here between exact calculations and heuristics (analytical procedure based on scenarios that are likely to occur). While the exact calculations provide a hundred percent reliable result, so-called metaheuristics, such as the ant algorithm, only allow for a probable best result. However, the advantage of the heuristic method is the much lower computational effort, which requires only a fraction of the effort of an exact calculation.

“In the planning phase of a logistics process, when time is not of the essence, you can and should work with as many exact methods as possible,” explains mathematician Thomas Runkler in an interview with brand eins. However, if unforeseen situations arise during operation, it is important to react as quickly as possible. “And that’s just where the ants are closer to real life.”

Thus, these methods of combinatorial optimization enable a dynamic material flow that adapts to the current conditions independently. Especially in the context of Industry 4.0 developments, the trend in some areas is therefore toward decentralized control intelligence, which also provides a basis for the smart factory and various cyber-physical systems.

For information on the simulation of intralogistics systems, see Material Flow Simulation.


Image source: © fdecomite/ License (CC BY 2.0)

Also available in Deutsch (German)