Swarm Intelligence

We study the psychophysiological state of humans when exposed to robot groups of varying sizes. In our experiments, 24 participants are exposed sequentially to groups of robots made up of 1, 3 and 24 robots. We measure both objective physiological metrics (skin conductance level and heart rate), and subjective self-reported metrics (from a psychological questionnaire). These measures allow us to analyse the psychophysiological state (stress, anxiety, happiness) of our participants. Our results show that the number of robots to which a human is exposed has a significant impact on the psychophysiological state of the human and that higher numbers of robots provoke a stronger response.

Ant-Miner is an ant-based algorithm for the discovery of classification rules. This paper proposes five extensions to Ant-Miner: (1) we utilize multiple types of pheromone, one for each permitted rule class, i.e. an ant first selects the rule class and then deposits the corresponding type of pheromone; (2) we use a quality contrast intensifier to magnify the reward of high-quality rules and to penalize low-quality rules in terms of pheromone update; (3) we allow the use of a logical negation operator in the antecedents of constructed rules; (4) we incorporate stubborn ants, an ACO variation in which an ant is allowed to take into consideration its own personal past history; (5) we use an ant colony behavior in which each ant is allowed to have its own values of the α and β parameters (in a sense, to have its own personality). Empirical results on 23 datasets show improvements in the algorithm’s performance in terms of predictive accuracy and simplicity of the generated rule set.

In the past few years, there has been much research on the application of swarm intelligence to the problem of adaptive routing in telecommunications networks. A large number of algorithms have been proposed for different types of networks, including wired networks and wireless ad hoc networks. In this paper, we give an overview of this research area. We address both the principles underlying the research and the practical applications that have been proposed. We start by giving a detailed description of the challenges in this problem domain, and we investigate how swarm intelligence can be used to address them. We identify typical building blocks of swarm intelligence systems and we show how they are used to solve routing problems. Then, we present Ant Colony Routing, a general framework in which most swarm intelligence routing algorithms can be placed. After that, we give an extensive overview of existing algorithms, discussing for each of them their contributions and their relative place in this research area. We conclude with an overview of future research directions that we consider important for the further development of this field.

In this paper, we investigate how to design an effective interface for remote multi-human–multi-robot interaction. While significant research exists on interfaces for individual human operators, little research exists for the multi-human case. Yet, this is a critical problem to solve to make complex, large-scale missions achievable in which direct operator involvement is impossible or undesirable, and robot swarms act as a semi-autonomous agents. This paper’s contribution is twofold. The first contribution is an exploration of the design space of computer-based interfaces for multi-human multi-robot operations. In particular, we focus on agent transparency and on the factors that affect inter-human communication in ideal conditions, i.e., without communication issues. Our second contribution concerns the same problem, but considering increasing degrees of information loss, defined as intermittent reception of data with noticeable gaps between individual receipts. We derived a set of design recommendations based on two user studies involving 48 participants.

In this paper, heuristic solution techniques for the multi-objective orienteering problem are developed. The motivation stems from the problem of planning individual tourist routes in a city. Each point of interest in a city provides different benefits for different categories (e.g., culture, shopping). Each tourist has different preferences for the different categories when selecting and visiting the points of interests (e.g., museums, churches). Hence, a multi-objective decision situation arises. To determine all the Pareto optimal solutions, two metaheuristic search techniques are developed and applied. We use the Pareto ant colony optimization algorithm and extend the design of the variable neighborhood search method to the multi-objective case. Both methods are hybridized with path relinking procedures. The performances of the two algorithms are tested on several benchmark instances as well as on real world instances from different Austrian regions and the cities of Vienna and Padua. The computational results show that both implemented methods are well performing algorithms to solve the multi-objective orienteering problem.

In this paper we present a procedure to automatically design and verify the local behavior of robots with highly limited cognition. All robots are: anonymous, homogeneous, non-communicating, memoryless, reactive, do not know their global position, do not have global state information, and operate by a local clock. They only know: (1) the relative location of their neighbors within a short range and (2) a common direction (North). We have developed a procedure to generate a local behavior that allows the robots to self-organize into a desired global pattern despite their individual limitations. This is done while also avoiding collisions and keeping the coherence of the swarm at all times. The generated local behavior is a probabilistic local state-action map. The robots follow this stochastic policy to select an action based on their current perception of their neighborhood (i.e., their local state). It is this stochasticity, in fact, that allows the global pattern to eventually emerge. For a generated local behavior, we present a formal proof procedure to verify whether the desired pattern will always eventually emerge from the local actions of the agents. The novelty of the proof procedure is that it is primarily local in nature and focuses on the local states of the robots and the global implications of their local actions. A local approach is of interest to reduce the computational effort as much as possible when verifying the emergence of larger patterns. Finally, we show how the behavior could be implemented on real robots and investigate this with extensive simulations on a realistic robot model. To the best of our knowledge, no other solutions exist for robots with such limited cognition to achieve this level of coordination with proof that the desired global property will emerge.

Biologically inspired collective decision-making algorithms show promise for implementing spatially distributed searching tasks with robotic systems. One example is the best-of-N problem in which a collective must search an environment for an unknown number of sites and select the best option. Real-world robotic deployments must achieve acceptable success rates and execution times across a wide variety of environmental conditions, a property known as resilience. Existing experiments for the best-of-N problem have not explicitly examined how the site layout affects a collective’s performance and resilience. Two novel resilience metrics are used to compare algorithmic performance and resilience between evenly distributed, obstructed, or unobstructed uneven site configurations. Obstructing the highest valued site negatively affected selection accuracy for both algorithms, while uneven site distribution had no effect on either algorithm’s resilience. The results also illuminate the distinction between absolute resilience as measured against an objective standard, and relative resilience used to compare an algorithm’s performance across different operating conditions.

In this paper we analyse a swarm robotics system using Bio-PEPA. Bio-PEPA is a process algebra language originally developed to analyse biochemical systems. A swarm robotics system can be analysed at two levels: the macroscopic level, to study the collective behaviour of the system, and the microscopic level, to study the robot-to-robot and robot-to-environment interactions. In general, multiple models are necessary to analyse a system at different levels. However, developing multiple models increases the effort needed to analyse a system and raises issues about the consistency of the results. Bio-PEPA, instead, allows the researcher to perform stochastic simulation, fluid flow (ODE) analysis and statistical model checking using a single description, reducing the effort necessary to perform the analysis and ensuring consistency between the results. Bio-PEPA is well suited for swarm robotics systems: by using Bio-PEPA it is possible to model distributed systems and their space-time characteristics in a natural way. We validate our approach by modelling a collective decision-making behaviour.

Multi-option collective decision-making is a challenging task in the context of swarm intelligence. In this paper, we extend the problem of collective perception from simple binary decision-making of choosing the color in majority to estimating the most likely fill ratio from a series of discrete fill ratio hypotheses. We have applied direct comparison (DC) and direct modulation of voter-based decisions (DMVD) to this scenario to observe their performances in a discrete collective estimation problem. We have also compared their performances against an Individual Exploration baseline. Additionally, we propose a novel collective decision-making strategy called distributed Bayesian belief sharing (DBBS) and apply it to the above discrete collective estimation problem. In the experiments, we explore the performances of considered collective decision-making algorithms in various parameter settings to determine the trade-off among accuracy, speed, message transfer and reliability in the decision-making process. Our results show that both DC and DMVD outperform the Individual Exploration baseline, but both algorithms exhibit different trade-offs with respect to accuracy and decision speed. On the other hand, DBBS exceeds the performances of all other considered algorithms in all four metrics, at the cost of higher communication complexity.