Robot swarms are known to be robust to individual robot failures. However, a reduced swarm size causes a reduced swarm density. A too low swarm density may then decrease swarm performance, that should be compensated by adapting the individual behavior. Similarly, swarm behaviors can also be adapted to changes in the environment, such as dynamic light conditions. We study aggregation of swarm robots controlled by an extended variant of the BEECLUST algorithm. The robots are asked to aggregate at the brightest spot in their environment. Our approach efficiently adapts this swarm aggregation behavior to variability in swarm density and light conditions. First, each robot individually monitors its environment continuously by sampling its local swarm density and perceived light condition. Second, we exploit the collaboration of robots by letting them share features of these measurements with their neighbors by communication. In extensive robot swarm experiments with ten robots we validate our approach with dynamically changing swarm densities and under dynamic light conditions. We find an improved performance compared to robot swarms without communication and without awareness of the swarm density.

Mobile sensor networks and robotic swarms are being used for monitoring and exploring environments or environmental events due to the advantages offered by their distributed nature. However, coordination and self-organization of a large number of individuals is often costly in terms of energy and computation power, thus limiting the longevity of the distributed system. In this paper we present a bio-inspired algorithm enabling a robotic swarm to collectively detect anomalies in environmental parameters in a self-organized, reliable and energy efficient manner. Individuals in the swarm communicate via 1-bit signals to collectively confirm the detection of an anomaly while minimizing energy spent for communication and taking measurements. This algorithm is specifically designed for a swarm of underwater robots called “aMussels” to examine a phenomenon referred to as “anoxia” which results in oxygen depletion in the lagoon of Venice. We present the algorithm, conduct simulations and robotic experiments to examine the performance of the algorithm with respect to early detection of anoxia while minimizing energy consumption.

Ant nest relocation is smoother and swifter than the same process undertaken by any other animal. Within the population of ants, the ratio that participates in nest relocation is only 58.0% at best and 31.0% at worst. Does such a low active ratio improve or deteriorate ant nest relocation? In this study, we use a particle swarm optimization (PSO) algorithm to simulate real-world ant nest relocation. Our PSO-based algorithm duplicates the velocity and position of an inactive particle (representing an inactive ant) with the velocity and position of an active particle (representing an active ant). The number of particles that the algorithm computes is dramatically reduced, and the global best position can be identified at an early stage. In a series of simulations, our algorithm performs significantly better and faster with active ratios of 15%, 30%, 35%, 45%, 55%, 60%, and 75%–95% than with the full 100% active ratio. We confirm the robust and stable performance of our algorithm at active ratios of 60%, 80%, and 85%. Clustering of the simulation results shows that low active ratios improve ant nest relocation. Furthermore, three field studies carried out by biology experts empirically demonstrate that we have successfully modeled and simulated real-world ant nest relocation using our PSO-based algorithm.

Animals and humans encounter many tasks which permit ritualized behaviours, essentially fixed action sequences or “scripts”, similar to options known from Reinforcement Learning, but proceeding without intermediate decisions. While running a script, they proceed in an open-loop fashion. However even when these are already known, an agent needs to decide whether to perform a basic action or to trigger a script regarding the particular task. Here we study if including such scripts (i.e. behaviour rituals) is advantageous from the point of view of the relevant information required to take the decision to start such a script depending on the tasks. To achieve this, we modify the relevant information framework including sequences of basic actions to the possible actions.

Social network analysis and agent-based modeling are two approaches used to study biological and artificial multi-agent systems. However, so far there is little work integrating these two approaches. Here we present a first step toward integration. We developed a novel approach that allows the creation of a social network on the basis of measures of interactions in an agent-based model for purposes of social network analysis. We illustrate this approach by applying it to a minimalist case study in swarm robotics loosely inspired by ant foraging behavior. For simplicity, we measured a network’s inter-agent connection weights as the total number of interactions between mobile agents. This measure allowed us to construct weighted directed networks from the simulation results. We then applied standard methods from social network analysis, specifically focusing on node centralities, to find out which are the most influential nodes in the network. This revealed that task allocation emerges and induces two classes of agents, namely foragers and loafers, and that their relative frequency depends on food availability. This finding is consistent with the behavioral analysis, thereby showing the compatibility of these two approaches.

This paper proposes an artificial pheromone communication system inspired by social insects. The proposed model is an extension of the previously developed pheromone communication system, COS-_. The new model increases COS-Φ flexibility by adding two new features, namely, diffusion and advection . The proposed system consists of an LCD flat screen that is placed horizontally, overhead digital camera to track mobile robots, which move on the screen, and a computer, which simulates the pheromone behaviour and visualises its spatial distribution on the LCD. To investigate the feasibility of the proposed pheromone system, real microrobots, Colias , were deployed which mimicked insects’ role in tracking the pheromone sources. The results showed that, unlike the COS-Φ, the proposed system can simulate the impact of environmental characteristics, such as temperature, atmospheric pressure or wind, on the spatio-temporal distribution of the pheromone. Thus, the system allows studying behaviours of pheromone-based robotic swarms in various real-world conditions.

We propose a simple decentralized control scheme for swarm robots that can perform spatially distributed tasks in parallel, drawing inspiration from the non-reciprocal-interaction-based (NRIB) model we proposed previously. Each agent has an internal state called “workload.” Each agent first moves randomly to find a task. When it finds a task, its workload increases, and then it attracts its neighboring agents to ask for their help. Simulation was used to demonstrate the validity of the proposed control scheme.

In nature, flocking or swarm behavior is observed in many species as it has beneficial properties like reducing the probability of being caught by a predator. In this paper, we propose SELFish (Swarm Emergent Learning Fish), an approach with multiple autonomous agents which can freely move in a continuous space with the objective to avoid being caught by a present predator. The predator has the property that it might get distracted by multiple possible preys in its vicinity. We show that this property in interaction with self-interested agents which are trained with reinforcement learning solely to survive as long as possible leads to flocking behavior similar to Boids, a common simulation for flocking behavior. Furthermore we present interesting insights into the swarming behavior and into the process of agents being caught in our modeled environment.

Self-organised aggregation, the formation of large clusters of independent agents, is an important process in swarm robotics systems since it is the prerequisite for more complex collective behaviours. Previous work on self-organised aggregation focused on the study of the individual mechanisms required to allow a swarm to form a single aggregate. In this paper, we discuss an analytical model which looks at the possibility to use the concept of informed individuals to allow the swarm to distribute on different aggregation sites according to proportions of individuals at each site arbitrarily chosen by the designer. Informed individuals are opinionated agents that selectively prefer an aggregation site and avoid to rest on the non-preferred sites. We study environments with two aggregation sites, and consider two different scenarios: one in which the informed individuals are equally distributed in numbers between the two sites; and one in which informed individuals for one type of site are three times more numerous than those on the other site. Our objective is to find out whether and for what range of model parameters the swarm distributes between the two sites according to the relative distribution of informed agents among the two sites. The analysis of the model shows that the designer capability to exploit informed individuals to control how the swarm aggregates depends on the environmental conditions. For intermediate values of the site carrying capacity, a small minority of informed individuals is able to guide the dynamics as desired by the designer. We also show that the larger the site carrying capacity the larger the total proportion of informed individuals required to lead the swarm to the desired distribution of individuals between the two sites.