Abstract
The search of a given area is one of the most studied tasks in swarm robotics. Different heuristic methods have been studied in the past taking into account the peculiarities of these systems (number of robots, limited communications and sensing and computational capacities). In this work, we introduce a behavioral network made up of different well-known behaviors that act together to achieve a good performance, while adapting to different scenarios. The algorithm is compared with six strategies based on movement patterns in terms of three performance models. For the comparison, four scenario types are considered: plain, with obstacles, with the target location probability distribution and a combination of obstacles and the target location probability distribution. For each scenario type, different variations are considered, such as the number of agents and area size. Results show that although simplistic solutions may be convenient for the simplest scenario type, for the more complex ones, the proposed algorithm achieves better results.
Highlights
From the very beginning of the development of swarm robotics, researches have focused on implementing behaviors in robots to solve simple tasks [7]
We propose an algorithm based on a behavioral network, made up by different behaviors that act together to obtain a common decision in order to lead each agent
In order to organize the search, most of the past works based on patterns have made use of dividing the area into cells, taking into account if they have been visited or not
Summary
As well as in our societies, it is easy to find individuals that act together to achieve a specific goal. With the development of robotics, it has been possible to cover increasingly complex tasks by using groups of intelligent agents that coordinately act as a team. Some other tasks might not benefit from the use of more than one robot, either because they are simple or because the development of the coordination and the communication between the members is a barrier not worth overcoming. Communications are achieved mostly by pheromones [2], chemicals spread in the environment used to mark trails, warn other members of the nest and confuse enemies It is obvious how difficult it would be for them if they had to survive individually. Swarm behaviors are a particular case of collective behavior, in which a large number of members is involved, requiring very limited communications between them and being individually simple. Hardware miniaturization and cost reduction have made possible new research lines in this direction, and hundreds of papers developing all sorts of algorithms have been published
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have