Search

Social insects, such as ants, termites, and honeybees, have evolved sophisticated societies where collaboration and division of labor enhance survival of the whole colony, and are thus considered “superorganisms”. Historically, studying behaviors involving large groups under natural conditions posed significant challenges, often leading to experiments with a limited number of organisms under artificial laboratory conditions that incompletely reflected the animals’ natural habitat. A promising approach to exploring animal behaviors, beyond observation, is using robotics that produce stimuli to interact with the animals. However, their application has predominantly been constrained to small groups in laboratory conditions. Here we present the design choices and development of a biocompatible robotic system intended to integrate with complete honeybee colonies in the field, enabling exploration of their collective thermoregulatory behaviors via arrays of thermal sensors and actuators. We tested the system’s ability to capture the spatiotemporal signatures of two key collective behaviors. A 121-day observation revealed thermoregulation activity of the broodnest area during the foraging season, followed by clustering behavior during winter. Then we demonstrated the system’s ability to influence the colony by guiding a cluster of bees along an unnatural trajectory, via localized thermal stimuli emitted by two robotic frames. These results showcase a system with the capability to experimentally modulate honeybee colonies from within, as well as to unobtrusively observe their dynamics over extended periods. Such biohybrid systems uniting complete societies of thousands of animals and interactive robots can be used to confirm or challenge the existing understanding of complex animal collectives.

We investigate how simple physical interactions can generate remarkable diversity in the life history of social agents using data of social wasps, yielding complex scalable task partitioning. We built and analyzed a computational model to investigate how diverse task allocation patterns found in nature can emerge from the same behavioral blueprint. Self-organizing mechanisms of interwoven behavioral feedback loops, task-dependent time delays and simple material flows between interacting individuals yield an emergent homeostatic self-regulation while keeping the global colony performance scalable. Task allocation mechanisms based on implicitly honest signaling via material flows are not only very robust but are also highly evolvable due to their simplicity and reliability. We find that task partitioning has evolved to be scalable and adaptable to life history traits, such as expected colony size or temporal bottlenecks in the available workforce or materials. By tuning solely the total number of agents and a social connectivity-related parameter in the model, our simulations yield the whole range of emergent patterns in task allocation and task fidelity akin to observed field data. Our model suggests that the material exchange (“common stomach mechanism”) found in many paper wasps provides a common functional “core” across these genera, which not only provides self-regulation of the colony, but also provides a scalable mechanism allowing natural selection to yield complex social integration in larger colonies over the course of their evolutionary trajectory.

Honey bees live in colonies of thousands of individuals, that not only need to collaborate with each other but also to interact intensively with their ecosystem. A small group of robots operating in a honey bee colony and interacting with the queen bee, a central colony element, has the potential to change the collective behavior of the entire colony and thus also improve its interaction with the surrounding ecosystem. Such a system can be used to study and understand many elements of bee behavior within hives that have not been adequately researched. We discuss here the applicability of this technology for ecosystem protection: A novel paradigm of a minimally invasive form of conservation through “Ecosystem Hacking”. We discuss the necessary requirements for such technology and show experimental data on the dynamics of the natural queen’s court, initial designs of biomimetic robotic surrogates of court bees, and a multi-agent model of the queen bee court system. Our model is intended to serve as an AI-enhanceable coordination software for future robotic court bee surrogates and as a hardware controller for generating nature-like behavior patterns for such a robotic ensemble. It is the first step towards a team of robots working in a bio-compatible way to study honey bees and to increase their pollination performance, thus achieving a stabilizing effect at the ecosystem level.

Predictions of emergent phenomena, appearing on the macroscopic layer of a complex system, can fail if they are made by a microscopic model. This study demonstrates and analyses this claim on a well-known complex system, Conway’s Game of Life. Straightforward macroscopic mean-field models are easily capable of predicting such emergent properties after they have been fitted to simulation data in an after-the-fact way. Thus, these predictions are macro-to-macro only. However, a micro-to-macro model significantly fails to predict correctly, as does the obvious mesoscopic modeling approach. This suggests that some macroscopic system properties in a complex dynamic system should be interpreted as examples of phenomena (properties) arising from “strong emergence,” due to the lack of ability to build a consistent micro-to-macro model, that could explain these phenomena in a before-the-fact way. The root cause for this inability to predict this in a micro-to-macro way is identified as the pattern formation process, a phenomenon that is usually classified as being of “weak emergence.” Ultimately, this suggests that it may be in principle impossible to discriminate between such distinct categories of “weak” and “strong” emergence, as phenomena of both types can be part of the very same feedback loop that mainly governs the system’s dynamics.

Environmental monitoring should be minimally disruptive to the ecosystems that it is embedded in. Therefore, the project Robocoenosis suggests using biohybrids that blend into ecosystems and use life forms as sensors. However, such a biohybrid has limitations regarding memory—as well as power—capacities, and can only sample a limited number of organisms. We model the biohybrid and study the degree of accuracy that can be achieved by using a limited sample. Importantly, we consider potential misclassification errors (false positives and false negatives) that lower accuracy. We suggest the method of using two algorithms and pooling their estimations as a possible way of increasing the accuracy of the biohybrid. We show in simulation that a biohybrid could improve the accuracy of its diagnosis by doing so. The model suggests that for the estimation of the population rate of spinning Daphnia, two suboptimal algorithms for spinning detection outperform one qualitatively better algorithm. Further, the method of combining two estimations reduces the number of false negatives reported by the biohybrid, which we consider important in the context of detecting environmental catastrophes. Our method could improve environmental modeling in and outside of projects such as Robocoenosis and may find use in other fields.

Social insect colonies show all characteristics of complex adaptive systems (CAS). Their complex behavioral patterns arise from social interactions that are based on the individuals’ reactions to and interactions with environmental stimuli. We study here how social and environmental factors modulate and bias the collective thermotaxis of young honeybees. Therefore, we record their collective decision-making in a series of laboratory experiments and derived a mathematical model of the collective decision-making in young bees from our empirical observations. This model uses only one free parameter that combines the ultimate effects of several aspects of the microscopic individual behavioral mechanisms, such as motion behavior, sensory range, or contact detection, into one single coefficient. We call this coefficient the “social factor.” Our model is capable of capturing the observed aggregation patterns from our empiric experiments with static environments and of predicting the emergent swarm-intelligent behavior of the system in dynamic environments. Besides the fundamental research aspect in studying CAS, our model enables us to predict the effects of a physical stimulus onto the macroscopic collective decision-making that affects several crucial prerequisites for efficient and effective brood production and population growth in honeybee colonies.

Vibratory signals play a major role in the organization of honeybee colonies. Due to the seemingly chaotic nature of the mechano-acoustic landscape within the hive, it is difficult to understand the exact meaning of specific substrate-borne signals. Artificially generated vibrational substrate stimuli not only allow precise frequency and amplitude control for studying the effects of specific stimuli, but could also provide an interface for human-animal interaction for bee-keeping-relevant colony interventions. We present a simple method for analyzing motion activity of honeybees and show that specifically generated vibrational signals can be used to alter honeybee behavior. Certain frequency-amplitude combinations can induce a significant decrease and other signals might trigger an increase in honeybees’ motion activity. Our results demonstrate how different subtle local modulatory signals on the comb can influence individual bees in the local vicinity of the emitter. Our findings could fundamentally impact our general understanding of a major communication pathway in honeybee colonies. This pathway is based on mechanic signal emission and mechanic proprio-reception of honeybees in the bee colony. It is a candidate to be a technologically accessible gateway into the self-regulated system of the colony and thus may offer a novel information transmission interface between humans and honeybees for the next generation of “smart beehives” in future beekeeping.

We develop here a novel hypothesis that may generate a general research framework of how autonomous robots may act as a future contingency to counteract the ongoing ecological mass extinction process. We showcase several research projects that have undertaken first steps to generate the required prerequisites for such a technology-based conservation biology approach. Our main idea is to stabilise and support broken ecosystems by introducing artificial members, robots, that are able to blend into the ecosystem’s regulatory feedback loops and can modulate natural organisms’ local densities through participation in those feedback loops. These robots are able to inject information that can be gathered using technology and to help the system in processing available information with technology. In order to understand the key principles of how these robots are capable of modulating the behaviour of large populations of living organisms based on interacting with just a few individuals, we develop novel mathematical models that focus on important behavioural feedback loops. These loops produce relevant group-level effects, allowing for robotic modulation of collective decision making in social organisms. A general understanding of such systems through mathematical models is necessary for designing future organism-interacting robots in an informed and structured way, which maximises the desired output from a minimum of intervention. Such models also help to unveil the commonalities and specificities of the individual implementations and allow predicting the outcomes of microscopic behavioural mechanisms on the ultimate macroscopic-level effects. We found that very similar models of interaction can be successfully used in multiple very different organism groups and behaviour types (honeybee aggregation, fish shoaling, and plant growth). Here we also report experimental data from biohybrid systems of robots and living organisms. Our mathematical models serve as building blocks for a deep understanding of these biohybrid systems. Only if the effects of autonomous robots onto the environment can be sufficiently well predicted can such robotic systems leave the safe space of the lab and can be applied in the wild to be able to unfold their ecosystem-stabilising potential.

The perceptions and definitions of healthy indoor environments have changed significantly throughout architectural history. Today, molecular biology teaches us that microbes play important roles in human health, and that isolation from them puts not only us but also other inhabitants of urban landscapes, at risk. In order to provide an environment that makes honeybees more resilient to environmental changes, we aim for combining the thermal insulation functionality of mycelium materials with bioactive therapeutic properties within beehive constructions. By identifying mycelial fungi’s interactions with nest-related materials, using digital methods to design a hive structure, and engaging in additive manufacturing, we were able to develop a set of methods for designing and fabricating a fully grown hive. We propose two digital methods for modelling 3D scaffolds for micro-super organism co-occupation scenarios: “variable-offset” and “iterative-subtraction”, followed by two inoculation methods for the biofabrication of scaffolded fungal composites. The HIVEOPOLIS project aims to diversify and complexify urban ecological niches to make them more resilient to future game changers such as climate change. The combined functions of mycelium materials have the potential to provide a therapeutic environment for honeybees and, potentially, humans in the future.

In this paper a honeybee inspired collective-decision-making algorithm called BEECLUST is studied in a swarm of autonomous robots and the performance of the swarm is investigated in different conditions. The algorithm has low requirements thus it is promising for implementation in robots with low resources. Here the algorithm is applied in swarms of improved e-puck robots in three different conditions in order to study the strengths and limitations of the algorithm. The collective system demonstrated a high performance in adapting to a dynamic environment as well as a very low sensitivity to additional robots with malfunctioning sensors. On the other hand the system shows an strong response to robots that act as social seeds influencing the decision-making of the swarm.

In evolutionary robotics, an encoding of the control software that maps sensor data (input) to motor control values (output) is shaped by stochastic optimization methods to complete a predefined task. This approach is assumed to be beneficial compared to standard methods of controller design in those cases where no a priori model is available that could help to optimize performance. For robots that have to operate in unpredictable environments as well, an evolutionary robotics approach is favorable. We present here a simple-to-implement, but hard-to-pass benchmark to allow for quantifying the “evolvability” of such evolving robot control software towards increasing behavioral complexity. We demonstrate that such a model-free approach is not a free lunch, as already simple tasks can be unsolvable barriers for fully open-ended uninformed evolutionary computation techniques. We propose the “Wankelmut” task as an objective for an evolutionary approach that starts from scratch without pre-shaped controller software or any other informed approach that would force the behavior to be evolved in a desired way. Our main claim is that “Wankelmut” represents the simplest set of problems that makes plain-vanilla evolutionary computation fail. We demonstrate this by a series of simple standard evolutionary approaches using different fitness functions and standard artificial neural networks, as well as continuous-time recurrent neural networks. All our tested approaches failed. From our observations, we conclude that other evolutionary approaches will also fail if they do not per se favor or enforce the modularity of the evolved structures and if they do not freeze or protect already evolved functionalities from being destroyed again in the later evolutionary process. However, such a protection would require a priori knowledge of the solution of the task and contradict the “no a priori model” approach that is often claimed in evolutionary computation. Thus, we propose a hard-to-pass benchmark in order to make a strong statement for self-complexifying and generative approaches in evolutionary computation in general and in evolutionary robotics specifically. We anticipate that defining such a benchmark by seeking the simplest task that causes the evolutionary process to fail can be a valuable benchmark for promoting future development in the fields of artificial intelligence, evolutionary robotics, and artificial life.

[This corrects the article DOI: 10.1371/journal.pone.0188004.].

We developed an agent-based computer model of an ecosystem to predict interactions of competition and predation. In our simulations of the model, the effects of the ‘Gause law’ emerged as the results of population fluctuations and a large number of stochastic events. Small biases in life history parameters produced strong effects through the interactions of positive feedback with the population fluctuations. In a low-production environment, the smaller and faster consumer outcompetes the larger and slower one, but in a high production environment the larger and slower consumer survives. Predation hastens the extinction of one of the consumers, but niche partitioning of the consumers increases both the coexistence of consumers and the number of predators. Predators with medium efficiency are able to coexist in the system longer and in larger numbers. Besides the ecological insights this model provides, we conclude that agent-based simulations are very effective tools to explore the interactions between predation and competition interactions.

Plant growth is a self-organized process incorporating distributed sensing, internal communication and morphology dynamics. We develop a distributed mechatronic system that autonomously interacts with natural climbing plants, steering their behaviours to grow user-defined shapes and patterns. Investigating this bio-hybrid system paves the way towards the development of living adaptive structures and grown building components. In this new application domain, challenges include sensing, actuation and the combination of engineering methods and natural plants in the experimental set-up. By triggering behavioural responses in the plants through light spectra stimuli, we use static mechatronic nodes to grow climbing plants in a user-defined pattern at a two-dimensional plane. The experiments show successful growth over periods up to eight weeks. Results of the stimuli-guided experiments are substantially different from the control experiments. Key limitations are the number of repetitions performed and the scale of the systems tested. Recommended future research would investigate the use of similar bio-hybrids to connect construction elements and grow shapes of larger size.

We propose a new regulation mechanism based on the idea of the "common stomach" to explain several aspects of the resilience and homeostatic regulation of honeybee colonies. This mechanism exploits shared pools of substances (pollen, nectar, workers, brood) that modulate recruitment, abandonment and allocation patterns at the colony-level and enable bees to perform several survival strategies to cope with difficult circumstances: Lack of proteins leads to reduced feeding of young brood, to early capping of old brood and to regaining of already spent proteins through brood cannibalism. We modeled this system by linear interaction terms and mass-action law. To test the predictive power of the model of this regulatory mechanism we compared our model predictions to experimental data of several studies. These comparisons show that the proposed regulation mechanism can explain a variety of colony level behaviors. Detailed analysis of the model revealed that these mechanisms could explain the resilience, stability and self-regulation observed in honeybee colonies. We found that manipulation of material flow and applying sudden perturbations to colony stocks are quickly compensated by a resulting counter-acting shift in task selection. Selective analysis of feedback loops allowed us to discriminate the importance of different feedback loops in self-regulation of honeybee colonies. We stress that a network of simple proximate mechanisms can explain significant colony-level abilities that can also be seen as ultimate reasoning of the evolutionary trajectory of honeybees.

Learning is the act of obtaining new or modifying existing knowledge, behaviours, skills or preferences. The ability to learn is found in humans, other organisms and some machines. Learning is always based on some sort of observations or data such as examples, direct experience or instruction. This paper presents a classification algorithm to learn the density of agents in an arena based on the measurements of six proximity sensors of a combined actuator sensor units (CASUs). Rules are presented that were induced by the learning algorithm that was trained with data-sets based on the CASU’s sensor data streams collected during a number of experiments with “Bristlebots (agents) in the arena (environment)”. It was found that a set of rules generated by the learning algorithm is able to predict the number of bristlebots in the arena based on the CASU’s sensor readings with satisfying accuracy.

A computer program that tracks animal behavior, thereby revealing various features and mechanisms of social animals, is a powerful tool in ethological research. Because honeybee colonies are populated by thousands of bees, individuals co-exist in high physical densities and are difficult to track unless specifically tagged, which can affect behavior. In addition, honeybees react to light and recordings must be made under special red-light conditions, which the eyes of bees perceive as darkness. The resulting video images are scarcely distinguishable. We have developed a new algorithm, K-Track, for tracking numerous bees in a flat laboratory arena. Our program implements three main processes: (A) The object (bee's) region is detected by simple threshold processing on gray scale images, (B) Individuals are identified by size, shape and spatiotemporal positional changes, and (C) Centers of mass of identified individuals are connected through all movie frames to yield individual behavioral trajectories. The tracking performance of our software was evaluated on movies of mobile multi-artificial agents and of 16 bees walking around a circular arena. K-Track accurately traced the trajectories of both artificial agents and bees. In the latter case, K-track outperformed Ctrax, well-known software for tracking multiple animals. To investigate interaction events in detail, we manually identified five interaction categories; 'crossing', 'touching', 'passing', 'overlapping' and 'waiting', and examined the extent to which the models accurately identified these categories from bee's interactions. All 7 identified failures occurred near a wall at the outer edge of the arena. Finally, K-Track and Ctrax successfully tracked 77 and 60 of 84 recorded interactive events, respectively. K-Track identified multiple bees on a flat surface and tracked their speed changes and encounters with other bees, with good performance.

We develop a model to produce plausible patterns of task partitioning in the ponerine ant Ectatomma ruidum based on the availability of living prey and prey corpses. The model is based on the organizational capabilities of a "common stomach" through which the colony utilizes the availability of a natural (food) substance as a major communication channel to regulate the income and expenditure of the very same substance. This communication channel has also a central role in regulating task partitioning of collective hunting behavior in a supply&demand-driven manner. Our model shows that task partitioning of the collective hunting behavior in E. ruidum can be explained by regulation due to a common stomach system. The saturation of the common stomach provides accessible information to individual ants so that they can adjust their hunting behavior accordingly by engaging in or by abandoning from stinging or transporting tasks. The common stomach is able to establish and to keep stabilized an effective mix of workforce to exploit the prey population and to transport food into the nest. This system is also able to react to external perturbations in a de-centralized homeostatic way, such as to changes in the prey density or to accumulation of food in the nest. In case of stable conditions the system develops towards an equilibrium concerning colony size and prey density. Our model shows that organization of work through a common stomach system can allow Ectatomma ruidum to collectively forage for food in a robust, reactive and reliable way. The model is compared to previously published models that followed a different modeling approach. Based on our model analysis we also suggest a series of experiments for which our model gives plausible predictions. These predictions are used to formulate a set of testable hypotheses that should be investigated empirically in future experimentation.

Endothermic heat production is a crucial evolutionary adaptation that is, amongst others, responsible for the great success of honeybees. Endothermy ensures the survival of the colonies in harsh environments and is involved in the maintenance of the brood nest temperature, which is fundamental for the breeding and further development of healthy individuals and thus the foraging and reproduction success of this species. Freshly emerged honeybees are not yet able to produce heat endothermically and thus developed behavioural patterns that result in the location of these young bees within the warm brood nest where they further develop and perform tasks for the colony. Previous studies showed that groups of young ectothermic honeybees exposed to a temperature gradient collectively aggregate at the optimal place with their preferred temperature of 36 °C but most single bees do not locate themselves at the optimum. In this work we further investigate the behavioural patterns that lead to this collective thermotaxis. We tested single and groups of young bees concerning their ability to discriminate a local from a global temperature optimum and, for groups of bees, analysed the speed of the decision making process as well as density dependent effects by varying group sizes. We found that the majority of tested single bees do not locate themselves at the optimum whereas sufficiently large groups of bees are able to collectively discriminate a suboptimal temperature spot and aggregate at 36 °C. Larger groups decide faster than smaller ones, but in larger groups a higher percentage of bees may switch to the sub-optimum due to crowding effects. We show that the collective thermotaxis is a simple but well evolved, scalable and robust social behaviour that enables the collective of bees to perform complex tasks despite the limited abilities of each individual.