Swarm intelligence (SI) systems exploit artificial or physical agents to solve optimization tasks without the need for a supervisor. Can swarm robots benefit from SI? To answer, this paper surveys how the search-and-tracking problem can be solved for robot swarms, where robots are simple, with local controllers, no central memory, and limited communication.

The challenge in using swarm robotics is that the behavior of the swarm emerges during activity, more than being fully designed a priori. The drawback is that it is difficult to obtain a stable behavior and to guarantee formal properties of the solution.

Search-and-tracking is central to many robotics applications, such as surveillance and monitoring. Multirobot systems, according to the authors, can outperform a complex and large robot in robustness and efficiency. However, there are so many algorithms for multiple robots that making a choice is not straightforward, hence the importance of a review, addressed to new users.

The authors characterize the search-and-tracking problem by number of targets, mobility of targets and trackers, complexity of the environment, prior knowledge, and cooperation and coordination among swarm members. Then, they focus on SI algorithms, such as particle swarm optimization, bees algorithms, artificial bee colony optimization, ant colony optimization, bacterial foraging optimization, glowworm swarm optimization, firefly algorithms, and biased random walk. Other non-SI methods, such as Kalman filter, potential fields, and formation-based target following, are also reported.

Two tables summarize the comparison. One table compares the problem characteristics considered by six papers, showing that mobile targets and obstacle avoidance are not commonly solved. The other table qualitatively compares eight papers on the target-following problem, showing that one approach (from 2010) fulfills all but one of the requirements of swarm robotics and has mathematically proven properties: it is a formation-based method, which unfortunately can require complex computation. Therefore, the answer about the importance of SI in allowing better target tracking for swarms of robots is partially unanswered.