The members of a swarm interact locally with each other and with their environment. These interactions lead to adaptive search behaviour and ultimately to the global optimization of the swarm. This nature-inspired optimization approach has developed rapidly in recent years and is inspiring new developments in numerous industries. But what mechanisms underlie swarm robotics, what features characterize intelligent swarm algorithms and what models and applications are currently available?

Algorithms in swarm robotics

The model of an intelligent swarm is based on a cooperative algorithm as a key component that controls behavior and interactions. Robot swarms are based on a wide variety of algorithms that can control basic functions such as the simple distribution of objects or robots in space or functions for complex cooperation such as the formation of a chain.

The basic goals of swarm robotics are often aimed at miniaturization of hardware or cost efficiency. Four of the most important characteristics of swarm intelligence are coordination, group formation, optimization and path planning. Typical characteristics of swarm algorithms are

• Low complexity: Individual robots follow simple rules.
• Scalability: The system is designed for any number of robots.
• Decentralization: Swarm robots are autonomous and do not follow external commands.
• Local interactions: The robots exhibit collective behavior through local information exchange.

Overview of existing models

Many different swarm algorithms have been developed over the years. The most commonly used models are Artificial Bee Colony (ABC), Boids (Bird-oid object) and Ant Colony Optimization (ACO).

ABC - Artificial Bee Colony

ABC was developed in 2005. The algorithm is inspired by the behavior of honey bees in their search for food sources (nectar) and the subsequent exchange of information about these food sources with other bees in the hive. The ABC algorithm consists of three phases: the Employed Bee Phase, the Onlooker Bee Phase (start phase) and the Scout Phase (search phase). Each Employed Bee selects a known food source and determines the nearest location. She estimates the amount of nectar available there and passes the information on to the other bees by dancing around the hive. The onlooker bee observes the worker's dance, selects one of the food sources on the basis of this information and goes there to view the available nectar. Old food sources are replaced by the new food sources discovered by the onlooker bee. The best food source and the position associated with it are thus preserved.

The position of a food source represents the 'target'. The amount of nectar found in this area corresponds to the 'quality/quantity', also called fitness or 'associated solution'. The number of workers thus corresponds to the number of 'associated solutions' within the swarm.

An illustrative example of an ABC algorithm can be found in agriculture. Swarm robotics is particularly suitable for precision farming, for example in large-scale agricultural operations where drones are used for topographical and thermographic purposes.

In agriculture, thermography provides important information about environmental conditions that is difficult to obtain by other means. If a drone is equipped with a thermal camera, the temperature in the area under investigation can be measured with astonishing precision. Different temperature values represent different characteristics of plant development. The use of drones in precision farming leads to high product efficiency and enables the use of minimal resources and thus savings.

Boids - Bird-oid Objects
Boids stands for a simulation of artificial life that was developed by Craig W. Reynolds in 1986. It is an optimization algorithm inspired by the natural behaviour of flocking birds. Boids is a simple algorithm with three rules that enable complex behavior in swarm robotics: Cohesion, Alignment and Separation.