and Birattari (2016)). Lacking a well-established engineering methodology to design swarm behavior, machine learning approaches to automatically build swarm robot control systems appear as a promising alternative, and many of the proposed approaches belong to the field of Evolutionary Robotics (ER, Nolfi and Floreano (2000)) and Evolutionary Swarm Robotics (ESR, Trianni (2008)). ER approaches use Evolutionary Algorithms (EA) to design robot controllers, and are of particular interest since those methods do not require complex in- formation to guide the search for behavior, only an overall evaluation of the performance is required. Among these, EmbodiedEvolution (EE, Watson et al. (2002); Eiben et al. (2010a)) is a family of algorithms that take place online, once the robots are already deployed for operation. Evolution of behaviors is carried in a decentralized manner: each robot runs an Evolutionary Algorithm onboard, robots exchange genetic material when meeting (i.e. with robots in a close vicinity), and selection and variation are performed locally by the robot (Bred` eche et al. (2015)). As such there is no central authority orchestrating the evolutionary process, in contrast with traditional ER approaches, and learning in the swarm emerges from interactions among the individual agents.
Unlike the microscopic model, a macroscopic model directly describes the overall collective behaviour of the sys- tem. In general, macroscopic models are more computationally efficient than their microscopic counterparts. One of the fundmental elements of the macroscopic probabilistic model are the Rate Equations, which have been suc- cessfully applied to a wide variety of problems in physics, chemistry, biology and the social sciences. For instance, Sumpter and Pratt  developed a general framework for modelling social insect foraging systems with gener- alised rate functions (differential equations). Sugawara and coworkers [6, 7] first presented a simple macroscopic model for foraging in a group of communicating and non-communicating robots, with analysis under different conditions; further study can be found in . Lerman and Galstyan [9, 10] proposed a more generalised and fun- damental contribution to macroscopic modelling in multi-agent systems. In , they presented a mathematical model of foraging in a homogeneous multi-robot system to understand quantitatively the effects of interference on the performance of the group. In , they developed a macroscopic model of collaborative stick-pulling, and the results of the macroscopic model quantitatively agree with both embodied and microscopic simulations. Agas- sounon and Martinoli  use the same approach to capture the dynamics of a robot swarm engaged in collective clustering experiments.
Whilst we highlight the evolutionary utility of global assess- ment in swarm robotics and use this as a primary motivation, the experimental task of foraging has no interdependency between operating robots. In the future we intend to utilise this methodology to investigate the possibility of a robot simulating itself interacting with other encountered robots. In this way, the embedded simulators will be collectively evolved, and the swarm behaviour will be virtually evolved specifically in terms of interacting robots. We tentatively hypothesise that when a robot cannot accurately perceive the actions of another robot, an embedded simulation may provide an alternative to develop coordinated and cooperative robotic behaviours in real time.
Mathematical modelling and analysis offers both an alternative and complement to ex- periments and simulation, and attention has been direction in recent years to addressing the modelling problem in swarm robotics using probabilistic approaches. One such approach is macroscopic modelling, which aims to directly describe the overall collective behaviour of the system. One of the fundamental elements of the macroscopic probabilistic model are the Rate Equations, which have been successfully applied to a wide variety of problems in physics, chemistry, biology and the social sciences. For instance, Sumpter and Pratt (2003) developed a general framework for modelling social insect foraging systems with gener- alised rate functions (differential equations). Sugawara and coworkers (Sugawara and Sano, 1997; Sugawara et al, 1999) first presented a simple macroscopic model for foraging in a group of communicating and non-communicating robots, with analysis under different con- ditions. Lerman and Galstyan (2001, 2004) proposed a more generalised and fundamental contribution to macroscopic modelling in multi-agent systems. Lerman et al (2001) devel- oped a macroscopic model of collaborative stick-pulling, and the results of the macroscopic model quantitatively agree with both embodied and microscopic simulations. Lerman (2002) presents a mathematical model of foraging in a homogeneous multi-robot system to under- stand quantitatively the effects of interference on the performance of the group. Agassounon et al (2004) used the same approach to capture the dynamics of a robot swarm engaged in collective clustering.
facilitation effect is responsible for these differences [Bartneck, 2003]. It would also mean that robots invoke stronger social facilitation effect than ECAs.
While unexpectedly participants declared the task as relatively difficult, which meant that the direction of H1 was reversed, they performed it very well hardly making any mistakes. Moreover, what is very important for the use of robots and ECAs in the educational domain, in both conditions users significantly decreased required time to solve one modular arithmetic task during the course of the experiment. In other words, they have learnt to solve problems faster after receiving supporting feedback from Nabaztag. Unfortunately, due to the lack of a control group it is impossible to say how big impact Nabaztag‟s feedback had and what was a result of simply learning better techniques for solving the task, which would also occur in absence of Nabaztag. Nevertheless, participants declared that the robot/agent helped them to focus on the main task.
transmission adds new variability so that successful behaviours, that increase the adaptation ability of agents, can evolve in the population. Acerbi and Nolfi  presented an adaptive algorithm based on a combination of selective reproduction, individual learning and social learning. They claimed that social learning provides an adaptive advantage when individuals are allowed to learn socially from experienced individuals and individually. Their results show that agents that learn on the basis of both social and individual learning outperform agents that learn on the basis of social learning only or individual learning only. Parisi  presented a method in which a neural network is trained so that it demonstrates the same behaviour as another neural network. The two networks were exposed to the same input and the connection weights of the learner network were changed so that the learner network progressively learn to behave like the teacher network. He claimed that if random noise is added to this training process, some students may have higher performance than their teachers. This random noise allows the evolution of behaviours in a group of social agents as it is sufficient to trigger the evolution of useful behaviours. The work presented in this paper is also relevant to the research on the evolution of language. In a work particularly relevant to this paper, Kirby  argued that the language must be transmitted between generations through a repeated cycle of use and learning. In the evolution of language, compositional syntax may have emerged not because of its utility to us, but rather because it ensures that the language can be transmitted successfully. The process of linguistic transmission is itself an adaptive system which operates on a time scale between individual learning and biological evolution. In another research, Kirby et al.  claimed that languages, as they are culturally transmitted, evolve so that they can be transmitted with high fidelity. In an experimental scenario, they showed how a basic artificial language became easier to learn and structured as it was transmitted in a group of human participants. In this research, it is shown that behaviours that are copied from one robot to another evolve and adapt during multiple cycles of iterated learning and these evolved behaviours are better fitted to the environment of the robot collective and the robots themselves. Limitations and heterogeneities in the real robots’ sensors and actuators give rise to variations in imitated behaviours and these variations allow better adapted behaviours to emerge and evolve during multiple cycles of imitation.
The mechanisms driving the evolution of gigantism and foraging capacity in cetaceans remains poorly understood. Here we provide different hypotheses that can be tested in future comparative studies. We posit that toothed whales evolved gigantism because echolocation allowed for the progressive invasion of the deep ocean that required larger size and enhanced diving capacity. In contrast, filter-feeding baleen whales evolved gigantism in response to recent changes in the ocean that provided high-quality prey patches, thereby resulting in increased foraging efficiency at greater whale sizes that could, in turn, support such extremes in body mass. With the advent of biologging tag technology, our ability to quantify the diving and foraging performance of cetaceans in their natural environment has greatly increased in the past two decades. The next great challenge is to integrate data from individual studies to test mechanistic hypotheses regarding form, function, physiology and ecology across scales. Specifically, we need to better understand the energetics of foraging and how it compares between particulate- feeding odontocetes and filter-feeding mysticetes. Furthermore, we need to test whether the estimated energetic efficiency of foraging increases with body size in both lineages and by what magnitude. Although direct oxygen consumption measurements cannot yet be made on cetaceans in natural environments, a comparative analysis of feeding rates and foraging performance in species ranging from the small harbor porpoise to the huge blue whale will provide key insights into the physiology and functional ecology of cetaceans. Moreover, these studies will clarify the processes that may have driven body size evolution in a group of mammals that ultimately led to the largest animals ever.
Abstract – When ships suffer hull damage at sea, quick and effective repairs are vital. In these scenarios where even minutes make a substantial difference, repair crews need effective solutions suited to modern challenges. In this paper, we propose a self-assembly algorithm to be used by a homogeneous swarm of autonomous underwater robots to aggregate at the hull breach and use their bodies to form a patch of appropriate size to cover the hole. Our approach is inspired by existing modular robot technologies and techniques, which are used to justify the feasibility of the proposed system in this paper. We test the ability of the agents to form a patch for various breach sizes and locations and investigate the effect of varying population density. The system is verified within the two-dimensional Netlogo simulation environment and shows how the system performance can be quantified in relation to the sizes of the breach and the swarm. The methodology and simulation results illustrate that the swarm robot approach presented in this paper forms an important contribution to the emergency ship hull repair scenario and compares advantageously against the traditional shoring methods. We conclude by suggesting how the approach may be extended to a three-dimensional domain to aid real-time implementation in the future.
CHAPTER 1. INTRODUCTION AND LITERATURE REVIEW
This chapter discusses and compares the various published works in which evolutionary algorithms have been used to optimize some or all aspects of a robot’s 1 morphology in addition to its control policy. There
are numerous dimensions by which this work may be classified, including but not limited to: (1) how much of control and morphological structure was placed under evolutionary control; (2) the tasks that robots were evolved to perform; (3) the sorts of genotypic encodings employed; (4) whether experiments were carried out entirely in simulation or whether physical robots were constructed or otherwise utilized; (5) whether sensory systems were included such that controllers received feedback from their environments; and (6) the sorts of control policies used. The first of these dimensions has been chosen to determine the high level structure of this chapter, though all dimensions will be discussed. The following section presents several experiments in which only morphology was under evolutionary control (either for unactuated objects or for robots with a predefined control policy). After this, experiments in which morphology and control were jointly evolved are discussed, starting with examples in which some parameters of the morphology were evolved and then mov- ing on to experiments in which the topology of the morphology and controller were evolved, thus allowing for the creation of robots of arbitrary shapes and behaviors. Following these discussions, a brief overview of some of the questions surrounding the evolution of complexity are discussed, along with how evolutionary robotics has been able to contribute to this debate. Finally, this chapter ends with some conclusions that may be drawn from these various studies along with a brief outline of the remainder of this dissertation.
can work on either single or multiple nodes. Rune provides a reasonably abstracted interface for easily implementing emulation targets as spaces without much concern about the interaction between emulation nodes. Rune has the following roles: (i) experiment environment setup/cleanup and progress management; (ii) procedure invocation; (iii) interaction be- tween spaces; (iv) time synchronization; (v) mutual exclusion. Figure 5 shows the structure of an experiment implemented using Rune. The ”Rune Master” module manages the config- uration of each experiment, and controls the progress of the experiment. The execution of all spaces deployed on multiple nodes is initiated by Rune master via modules called ”Rune Manager”. The Rune manager is deployed on every emulation node and mediates communication between them through ob- jects called ”conduits”. Spaces implementing emulation targets exist on emulation nodes in the form of shared objects, loaded dynamically by the Rune manager.
This paper focuses on Padlet and presents some of the uses of this web application that support collaborativelearning in higher education. In particular, we summarize the process of collaborativelearning through Padlet in the Master's Degree for Secondary Education, Vocational Training and Language Teaching at University Jaume I (Spain). One of the primary purposes of this Master’s Degree is to shape the idea that teaching is a thoughtful process and that future teachers should introduce reflection in their daily activities (i.e. they should not act based on impulse or intuition) and that they need to interact with other teachers to improve the quality of their teaching. We believe that in this context, collaborativelearning may help students to retain information, increase their motivation and improve critical thinking skills (Cavanagh, 2011) all of which are of crucial value for future teachers.
Abstract--The advancement of industrial robotics has caused robots to become more widespread across various industries ranging from manufacturing to health care. Robots offer speed and accuracy that can’t be achieved with human labour. Robots can also reduce operating costs, reduce scrap and are flexible for future changes. Few other manufacturing solutions can reduce waste as well as robots when designed into the system properly. Robotics’ capabilities have only increased with time, while costs have continued to fall. Major robot manufacturers are constantly upgrading their robots with increased payload capacity, greater accuracy, increased reach and range of motion, improved speed and acceleration, faster communication with external equipment, better safety features and lower operational costs. Many people believe the misconception that robots have taken away jobs from workers, but that is not necessarily true. Robots have created new jobs for those who were once on production lines with programming. They have pulled employees from repetitive, monotonous jobs and put them in better, more challenging ones. Today robots are user-friendly, intelligent, and affordable. This paper briefly describes robot technology and goes into more depth about where robots are used. A chronology of robotics technology and use of future robots is also given.
In Chapter 5, an ILSA is proposed. This algorithm takes into account the random node deployment in defining the active CB nodes to take part in CB. The ILSA algorithm is discussed in details. In addition, series o f simulations are conducted using different number o f nodes and objectives to evaluate the performance o f this algorithm, along with other companion algorithm in the literature that are designed for WSNs. Comparisons are made between the algorithms in order to show the benefit o f using ILSA in handling the random node with desired objectives.
The e-Learning is expected to bring a revolution to the education system by providing different opportunities to share abundant information and knowledge . IROBI was the human friendly Intelligent home Robot developed under the e- Learning technology that provides home monitoring, entertainment, messaging, home tutoring and security. They provide facial expressions like happiness, calm and sadness. In comparison with other Medias, these robots make differences in the effect of learning . Most of the children reacted to the motions and facial expression of the IROBI. Figure 1  shows the interaction between the IROBI and the child.
items of quality Q i . The quality is a numerical indication of the importance of the resource
with respect to the task that will be performed; this is similar to the nutritional value of food items in animal foraging. In this work, we are interested in the foraging process at steady state; therefore, we assume sources which never deplete. If a robot enters a source area, it immediately collects one virtual item (or object) and returns it to the central circular depot (of radius 10 cm). We do not take into account any handling time of the resource item. Also, we do not consider the time spent in the resource patch, as the robot immediately finds an object and returns to the depot (no exploration within the source area). The load carried back to the nest site is always one item at a time. Travelling takes place with the same speed independent of the load carried (i.e. either unloaded or loaded with one object). Keeping these aspects in abstract terms helps to focus the study on the collective motion aspect and allocation of robots to source areas. In fact, this study focuses on strategies to coordinate the robot motion between depot and source areas through decentralised self-organising mechanisms. In particular, we explore how indirect communication in the form of virtual pheromone trails can allow the robot swarm to balance the trade-off between the quality of resource items and the distance between the source area and the central depot.
Algorithm Switching Multiple behaviors within a single algorithm are well- known inside the swarm algorithm community . Matari¢ and Arkin have worked extensively in behavior-based robotics [2, 4]. It is common for in- dividual robots to switch between the behaviors of food collecting, obstacle avoidance, and resting, for example. Parker has studied distributed consensus in a swarm setting, using it to enable the swarm to move between subtasks in an overall task . We focus on foraging, and do not use formal distributed consensus (or assume that our robots are `well-stirred'). McLurkin has de- veloped a large range of robot swarm behaviors as well as dynamic task assignment methods for individual robots within a swarm . These meth- ods focus on individuals, whereas we need a method for the swarm as a whole to switch algorithms.
Abstract— We discuss techniques towards usingcollaborativerobots for infrastructure security applications. A vast number of critical facilities, including power plants, military bases, water plants, air fields, and so forth, must be protected against unauthorized intruders. A team of mobile robots working cooperatively can alleviate human resources and improve effectiveness from human fatigue and boredom. This paper addresses this scenario by first presenting distributed sensing algorithms for robot localization and 3D map building. We then describe a multi-robot motion planning algorithm according to a patrolling and threat response scenario. A block diagram of the system integration of sensing and planning is presented towards a successful proof of principle demonstration. Previous approaches to similar scenarios have been greatly limited by their reliance on global positioning systems, the need for the manual construction of facility maps, and the need for humans to plan and specify the individual robot paths for the mission. Our proposed approaches overcome these limits and enable the systems to be deployed autonomously without modifications to the operating environment.
and current robot control.
The reasons for building multi-robot system is very different. However, one of the main motivations is the efficiency of multi-robots system. That is, compared to a single autonomous robot, a team of multiple robots can perform a mission better in terms of time cost and map quality. A team of robots could search the required environment cooperatively to directly reduce time cost of exploring; nevertheless, a team of robots usually have multiple points of view to the objectives in the environment. Environment objectives (e.g. landmarks) can be better estimated by fusing member robots' sensing data, which in turn increases the system effectiveness. Moreover, the reliability of multi-robot is higher than single robot because a team of robots could suffer one or two robots are damaged after robots begin their tasks. The rest of team members could finish those tasks that should be finished by those broken robots. Finally, instead of building a single powerful robot, building a team of robot can be easier and cheaper, can make the system tolerant to possible robots' faults, but can achieve complicated tasks as powerful as a single robot.
orientation. The size of the blob created by the filtering technique described above is directly proportional to the distance from the observer. Figure 4.6(a) shows the results of using blob size to estimate distance. Four different data sets were measured, where the observed robot was placed directly in front (Forward), 90 degrees to the left (Left), 90 degrees to the right (Right), and directly behind (Back). The data shows there are significant asymmetries in the camera which results in different blob sizes measured at the same distance. In addition to asymmetries in blob size measurements due to direction, the size can also be affected in other ways. For instance, if the observed robot is partially obscured its blob will appear smaller and thus be perceived to be further away then it actually is. The observed size of the blob could also be affected by the lighting in the environment which could change the amount of blue which passes through the colour threshold filter.