Swarm Robotics (SR) is an extension of the study of Multi-Robot Systems that exploits concepts of communication, coordination and collaboration among a large number of robots. The massive parallelization yielded by the robots working together can make a task faster than in the case of the usage of a single complex robot. One of the main aspects in robotic swarms is that the control is decentralized by definition and distributed among the robots of the swarm, improving the system robustness and fault-tolerance. Furthermore, this characteristic often allows the emergence of collective behaviors from the robot's interaction with each other and with the environment through their embodied sensors and actuators. In most cases, the number of inputs from sensor readings turns analytical solutions hard or even impossible. Thus, many ad-hoc approaches are contributed to deal with the situation at hand. The main goal of this review is to find out, through the study of existing research works of the field, the reason behind the lack of exploitation of swarm robotic systems in real-world applications. For this purpose, we first review the different possibilities of study in SR: physical and simulated robotic platforms, development methodologies and the variety of basic tasks and collective behaviors. We then briefly describe some fields related do SR that have a big impact on the development of SR. After that, based on existing taxonomies found in literature, we categorize existing research works regarding SR in two large main groups: those that deal with SR design and those that deal with tasks as required in SR. The review of both existing robots and techniques in the literature show a diversity of approaches to discuss SR issues. Nonetheless, it is easily noticeable from these works that there is a clamant absence of solid real-world applications of SR. An analysis of the interests and bottlenecks of this field indicates that the number of research works is smaller than those in other related areas. This suggests that, even though with many research studies, the field of SR is not yet mature enough, mainly due the absence of a universal methodology and generic robots that can be used in any, or at least in many, applications. Thus, we emphasize, discuss and analyze the urgent need for standardization of many aspects in SR, including hardware and software, as to allow a possible flourishing of SR applicability to real-world applications. This standardization could accelerate a great deal the field of SR, thus facilitating the development of SR solutions for applications that impact our daily life.

Swarm Robotics（SR）是对多机器人系统研究的扩展，该系统利用了大量机器人之间的通信，协调和协作概念。与使用单个复杂机器人的情况相比，机器人共同工作所产生的大规模并行化可以使任务更快。机器人群的主要方面之一是控制通过定义分散，并在群体机器人之间分配，从而提高了系统的鲁棒性和容错性。此外，该特征通常允许通过机器人彼此的相互作用以及通过其具体体现的传感器和致动器与环境的相互作用而出现集体行为。在大多数情况下，来自传感器读数的输入数量使分析解决方案变得困难甚至无法实现。因此，许多特设采取了各种措施来应对眼前的局势。这篇综述的主要目的是通过对该领域现有研究工作的研究，找出在实际应用中缺乏利用群体机器人系统的原因。为此，我们首先回顾一下在SR中研究的不同可能性：物理和模拟机器人平台，开发方法以及各种基本任务和集体行为。然后，我们简要介绍与SR相关的一些领域，这些领域对SR的发展有很大影响。然后，根据文献中发现的现有分类法，我们将有关SR的现有研究工作分为两个大类：处理SR设计的工作和处理SR所需任务的工作。对现有机器人和技术的回顾表明，讨论SR问题的方法多种多样。尽管如此，从这些作品中很容易注意到，确实缺少可靠的SR实际应用程序。对这一领域的兴趣和瓶颈进行的分析表明，研究工作的数量少于其他相关领域的研究工作的数量。这表明，尽管进行了许多研究，但SR领域还不够成熟，这主要是由于缺少通用方法和通用机器人，这些通用方法和通用机器人可用于任何或至少许多应用程序。因此，我们强调，讨论和分析了对SR的许多方面（包括硬件和软件）进行标准化的迫切需求，以使SR在现实世界中的适用性可能蓬勃发展。