China
Africa
Explore chapters and articles related to this topic
Understanding Digital Transformation
Published in Antonio Sartal, Diego Carou, J. Paulo Davim, Enabling Technologies for the Successful Deployment of Industry 4.0, 2020
Swarm robotics is the study of how to design groups of distributed robots that operate without relying on any external infrastructure or on any form of centralized control. A collective behaviour (Garnier et al., 2005) emerges from the interactions between the robots and interactions of robots with the environment in which they are operating. Swarm intelligence and biological studies of insects, ants, bees and others are the main source of swarm behaviour. In swarm robotics, automatic design has been mostly performed using the evolutionary robotics approach (Nolfi and Floreano, 2000). Evolutionary robotics has been used to develop several collective behaviours including collective transport (Groß and Dorigo, 2008) and development of communication networks (Huaert et al., 2008). A swarm robot team is fault tolerant, scalable and flexible. The robots in a swarm environment are able to perform different activities concurrently. More importantly, swarm robotics promotes the development of systems that are able to cope well with the failure of one or more of their constituent robots. That is to say that the failure of an individual robots does not imply the failure of the whole swarm (Fault tolerance) as the swarm does not rely on any centralized control entity, leaders or any individual robot playing a predefined role.
A System of Intelligent Robots–Trained Animals–Humans in a Humanitarian Demining Application
Published in Thrishantha Nanayakkara, Ferat Sahin, Mo Jamshidi, Intelligent Control Systems with an Introduction to System of Systems Engineering, 2018
Thrishantha Nanayakkara, Ferat Sahin, Mo Jamshidi
Swarm robotics is a related area of research, where a complex collective behavior is emerged in a group of relatively simple robots through interrobot and robot-environment interactions (Goldberg and Mataric, 1997; Hayes, 2002; Jones, and Mataric, 2003; Krieger and Billeter, 2000). A given robot is generally very simple and inexpensive, capable of eliciting a limited array of primitive behaviors, and equipped with a limited number of sensors and actuators. Given a task to be accomplished, such as walking over a gap that none of the individual robots can accomplish, the simple robots may share their diverse sensor information, actively support each other by joining hands, and coordinate movements to achieve the common goal. This concept can be useful in a task to find a hazardous object in a cluttered environment where one robot cannot carry all the required sensors. Yet, in that case, there is nothing that improves the behavior of an individual robot. The focus is to emerge a collective behavior based on the rules that apply to each robot to interact with other robots and the environment, and not to improve individual robotic behaviors through peer interaction. To the best of our knowledge, there has been no work done to study a scenario where each robot is directly influenced by a real animal as a peer who can sniff for the target objects and, at the same time, support the robot to learn how to navigate in a cluttered environment.
A Playbook-Based Interface for Human Control of Swarms
Published in Mustapha Mouloua, Peter A. Hancock, James Ferraro, Human Performance in Automated and Autonomous Systems, 2019
Phillip M. Walker, Christopher A. Miller, Joseph B. Mueller, Katia Sycara, Michael Lewis
Robot swarms consist of autonomous or semiautonomous robots that coordinate via local interaction rules. These laws are based on the robot's current state and surrounding environment (Brambilla, Ferrante, Birattari, & Dorigo, 2013). The primary advantages of swarms, often touted, include robustness to failure of individual robots and scalability (Bayindir & Şahin, 2007; Şahin, 2004), which are due largely to the distributed nature of their coordination and lack of globally specified plans. Conversely, multi-robot systems are distinct from swarms in that the individual members have explicitly represented goals, knowledge of the overall group plans, and can be controlled directly by a human operator (Farinelli, Iocchi, & Nardi, 2004; Lewis, 2013; Parker, 2008). Such robots could act independently without coordinating, for example, in a scenario with multiple robots searching different areas via independent routes for victims in a search and rescue scenario, or they could cooperate as a team in which all members work toward a known goal. Swarms, on the other hand, almost exclusively operate “collectively” but not necessarily “as a team” in that none of them need be aware of the overall goal or even of the team's progress toward it. Instead, coordination among robots relies on autonomous distributed algorithms and information processing to give rise to global emergent behaviors. Individual, pairwise, and small group actions (e.g., coordination of spacing and following behaviors) may accomplish an overall goal (e.g., searching an area or moving to a new location), but no individual agent is aware of that goal for the team as an entity. Instead, the overall goal is held by an outside entity (e.g., a human controller or supervisor), or abstractly (by “the hive”) and is monitored, if at all, by that outside entity.
Obtaining emergent behaviors for swarm robotics singling with deep reinforcement learning
Published in Advanced Robotics, 2023
Pilar Arques, Fidel Aznar, Mar Pujol, Ramón Rizo
Several swarm control techniques have now been developed, such as controller learning, the use of neighborhood topologies, artificial potential functions, and different forms of adaptive control. As discussed in [1] designing models from swarm observation or learning them from swarm data, although a valid alternative, can be problematic since the human perception can be biased. For example [3] demonstrated characteristic differences in the perception of swarming that raise the question of whether models driven by human perception of swarming are biased or incomplete. Furthermore, there is no guarantee that the human-designed models reviewed in this paper would be the most efficient way to drive an artificial agent to shepherd, nor that it is the right and/or only way. A robotic swarm has constraints on memory, processing resources, and energy that must be taken into account.
“Is something amiss?” Investigating individuals’ competence in estimating swarm degradation
Published in Theoretical Issues in Ergonomics Science, 2022
August Capiola, Izz aldin Hamdan, Elizabeth L. Fox, Joseph B. Lyons, Katia Sycara, Michael Lewis
Swarms comprise robotic assets operating via local control laws (Kolling et al. 2016). Coordinated asset behavior, such as flocking, can be generated with simple rules such as: 1) move away from an asset sensed closer than a defined distance (i.e., d1), 2) move toward an asset sensed further than d1, and 3) adjust heading to the average of those sensed assets (Amraii et al. 2014). Autonomous robotic swarm behaviors result from underlying algorithms which emulate natural swarms (e.g., bees, birds) found in nature (Brambilla et al. 2013). Specifically, swarms are unique compared to other robotic systems in that swarms are scalable (i.e., perform well with different group sizes), flexible (i.e., cope in different environments and tasks), and robust (i.e., cope with the loss of individual assets). Of particular interest, research shows that swarms are often robust to asset loss (Walker et al. 2012), which suggests that some swarms can maintain an acceptable level of task performance despite the loss of some of their individual assets. This places a great challenge on human operators to understand the various performance implications of swarm degradation in order to make responsible and ethical decisions regarding swarm behavior (which might include human inputs to the swarm behaviors). However, as assets operate based on information sensed from neighboring assets, swarms can be affected by degraded assets they comprise (Liu et al. 2019). As such, robust and reliable swarm technologies are necessary before swarms are deployed alongside humans in real-world missions.
Partition of a swarm of robots into size-balanced groups in presence of line obstacles
Published in International Journal of Parallel, Emergent and Distributed Systems, 2022
Arun Kumar Sadhu, Srabani Mukhopadhyaya
Swarm robotic research considers a group of very simple robots who collectively perform various tasks by their continuous cooperation and coordination. Direct communication among the robots makes the system susceptible to error or malicious attack. Researchers in this field prefer to consider indirect communication among the robots during the design of possibly bio-inspired algorithms [1]. Getting a job done by a group of robots sometimes required human intervention [2,3]. However, in this paper, we consider a group of completely autonomous robots that execute a specific algorithm independently to achieve their goal. In this paper, robots do not have any direct communication among themselves during the execution of the algorithm. Each robot is capable of observing its surrounding, carrying out simple computations, moves freely over the plane. This behaviour of the robots is designed by monitoring the activities of insects, birds, etc., for their survival. Some of these behaviours, like foraging, vigilance, flight, etc., help researchers to design basic algorithms [4] for swarm robots, which can also be used to develop a low-cost multi-robot system. Multi-robot systems are helpful in various applications like exploration, rescue operation, target searching, surface cleaning, etc., where human intervention is difficult. To address these multi-robot applications, researchers usually design distributed algorithms to solve basic problems in this domain, like pattern formation, flocking, partitioning, searching, gathering, convergence, aggregation, etc., by a swarm of robots [5–11].