Search

Honeybees (Apis mellifera L.) perform an important service to the ecosystem as they function as significant pollinators for plants. Over the past few decades, honeybees have suffered a progressive decline. The ability to observe and control the activity and density dynamics of honeybees in their hives in an automated way can allow to stimulate their actions and improve the overall efficiency of the hive. In this paper we present a novel honeybee observation method that is primarily based on honeybee density estimation using a convolutional neural network. First, specially designed stationary robots were positioned inside an arena designed for young honeybees. Three robots were used, each equipped with six infrared sensors for honeybee detection. The hardware and software setup that was used during the raw data collection process is described. Using the collected data from experiments with different numbers of honeybees we tested different convolutional neural networks to evaluate the relation between the network parameters and the estimation accuracy. To obtain better results, the numbers of honeybees were grouped into four different categories. It is shown that the most influential parameters are the number of epochs and the feature map size. By using the correct parameters it is possible to obtain 100 % accuracy during network training process and 86 % accuracy during evaluation process.

The field of swarm robotics draws most of its inspiration from (eu)social animals, which leads to the creation of bio-inspired algorithms. In this study, we show that counterintuitively - seemingly asocial behavior can also lead to successful problem solving performed by a swarm. We decided to test our new algorithm at first with real robots as a proof of concept, because this approach is of high conceptual novelty. We show that our social distancing algorithm (SocDist) performs similarly effective as a simple version of the well established bio-inspired social algorithm BEECLUST in a laboratory experiment, while avoiding some of its drawbacks. In additional agent-based computer simulation experiments we show that such an ‘asocial’ component within a swarm robotic algorithm can lead to a significant performance increase. Beside its effectiveness, the SocDist approach also leads to specific spatial distributions of the swarm robots, which may be useful for practical applications.

This article introduces a novel approach towards modeling evolutionary game-theoretical systems by creating a self-regulating societal ecosystem. We here develop a very simple model of a cooperating agent population that interacts with a dynamically changing environment. These interaction patterns are initially random but adapt with time in a process mimicking natural adaptation and selection in biological evolution: Ultimately, a homeostatic system emerges from scratch that regulates the growth of populations and environmental resources in a sustainable way, exhibiting strong resilience to perturbations. Moreover, the agent population exhibits horizontal social behavior represented by the Ultimatum Game but also vertical evolution-relevant behavior represented by reproductive strategies. The interplay between these horizontal and vertical mechanisms is in the focus of the analysis presented here. We found emergence of co-dependent horizontal and vertical norms. Ultimately, we analyzed the interplay between micro-and macro-properties and found significant effects, suggesting occurrence of emergence within the system: Micro properties are local and individual while macro properties are global and population-wide. We show that both levels interact and influence each other bilaterally as a result of the nonlinear and complex system behavior.

Bio-hybrid systems in which living organisms interact and self-organize with multi-robot systems are a novel approach in engineering and biology. We show here how a group of honeybees and robots can interact in collective decision making and how computer code that adds feedback loops to the system may affect the global system properties. This study contains a series of experiments with living honeybees and robots as well as a cellular-automaton inspired model that is simple, yet still in good agreement with the empirical findings presented here. Using this model, we explore the most likely candidates for local parameters in the proximate mechanisms of the animals, thus further the understanding of this natural system in a context that is relevant for such bio-hybrid manifestations. We identify positive feedback based on bee-to-bee collision and temperature as an important factor governing collective decision making and found the stopping probability after close-encounters among bees as a crucial local parameter. This study is the first step towards using computer and robotic technology to monitor and control complex animal societies like honeybees.

Fundamentalism and extremism are behavioral traits prominently observed today, ranging from religion and ideology to product choices and even dietary preferences. This study develops a multi-agent model depicting the exemplary case of preventing to eat meat (vegetarianism) or animal-derived products (veganism) as a set of “memes” that can evolve within a society. Behavioral traits can develop over time by means of (local) individual-based adaptation to an ultimate societal phenomenon by cultural (global) learning. The multi-agent model presented here predicts that fundamentalist strategies emerge in significant amounts only if certain extreme starting conditions of the environment (market) or of the initial population are met. When starting from non-extreme conditions, the population develops towards moderate behavioral traits and the market adapts to comfort these society-level adaptations. The underlying social network structure has a significant effect on those processes, as fundamentalism occurs only in fragmented subpopulations as longer-lasting phenomena. A very simple mechanism of social interaction is capable to capture general principles of the emergence of fundamentalist traits in societies.

In recent studies, researchers can easily record the behaviors of animals by digital video cameras, which provide high functionality at reasonable cost. However, it is a laborious and time-consuming manual task for them to extract the useful behavioral data from these videos. We recently proposed a tracking method for unmarked multiple honeybees in a flat arena, named the “K-Track” algorithm. The algorithm can successfully identify and track nearly 90% of interaction cases of targets. In this study, we proposed an improved method for the existing K-Track algorithm by tracking results using backward-play movie. If the tracking results differed between the forward and backward episodes, one of them had probably resulted from correct tracking. Therefore, by comparing the forward and backward trajectories of the same interaction, we assumed that there is the potential for an increase in tracking accuracy. In the experiments, K-Track using backward movies successfully tracked four out of five situations that was failed by original K-Track and we confirmed that the method has the potential of improved tracking accuracy.

As a part of ASSISIbf project, with a final goal of forming a collective adaptive bio-hybrid society of animals and robots, an artificial neural network based on LSTM architecture was designed and trained for bee density estimation. During experiments, the bees are placed inside a plastic arena covered with wax, where they interact with and adapt to specialized static robotic units, CASUs, designed specially for this project. In order to interact with honeybees, the CASUs require the capability i) to produce and perceive the stimuli, i.e., environmental cues, that are relevant to honeybee behaviour, and ii) to sense the honeybees presence. The second requirement is implemented through 6 proximity sensors mounted on the upper part of CASU. In this paper we present estimation of honeybees (moving agents) density in 2D space (experimental arena) that is based on LSTM neural network. When compared to previous work done in this field, experiments demonstrate satisfactory results in estimating sizes of bee groups placed in the arena within a larger scope of outputs. Two different approaches were tested: regression and classification, with classification yielding higher accuracy.

It is a challenging task to develop morphologies of structures in response to dynamic environmental factors and constraints. In the context of the EU-funded project flora robotica [1] we are interested in developing selforganized methods that combine local considerations and global requirements and drive the development of structures. Embryogenetic development of biological organisms and cell differentiation are studied for a long time in evolutionary developmental biology (EvoDevo) [2], [3]. Some of the mechanisms from that field are already applied to pattern formation [4] and development of body morphologies [5], [6] and controllers [7] in evolutionary robotics [8] and modular robotics [9]. In this work, vascular system and branching dynamics of plants are used as the source of inspiration for designing a novel algorithm called "Vascular Morphogenesis Controller" (VMC) that is applied to morphological development of modular structures. Plant vessels develop in the stems and roots. They transport water and minerals from the roots to the leaves, and sugars and photosynthates from the leaves to other parts of the plant [10]. There are evidences [11], [12] suggesting that there is a competition between different branches over the vascular growth. The branches that are in better situations (e.g., get more light) produce more photosynthates that flow back from the leaves. The higher flow rate leads to more vascular tissues in the branch and therefore more water and minerals from the roots reach the branch. More water and minerals facilitate the growth of the branch and the branch may end up in an even better situation which in turn reinforces the growth. Different branches with their different local conditions compete over production of new vessels. On the other hand, global resources (i.e., water) are limited and the vessels are subject to degradation as well. Based on the positive and negative feedback loops established by this competition and limitation, a dynamic system of vessels shape the growth of the plants.

Besides the life-as-it-could-be driver of artificial life research there is also the concept of extending natural life by creating hybrids or mixed societies that are built from natural and artificial components. In this paper we motivate and present the research program of the project flora robotic a. Our objective is to develop and to investigate closely linked symbiotic relationships between robots and natural plants and to explore the potentials of a plant-robot society able to produce architectural artifacts and living spaces. These robot-plant bio-hybrids create synergies that allow for new functions of plants and robots. They also create novel design opportunities for an architecture that fuses the design and construction phase. The bio-hybrid is an example of mixed societies between 'hard' artificial and 'wet' natural life, which enables an interaction between natural and artificial ecologies. They form an embodied, self-organizing, and distributed cognitive system which is supposed to grow and develop over long periods of time resulting in the creation of meaningful architectural structures. A key idea is to assign equal roles to robots and plants in order to create a highly integrated, symbiotic system. Besides the gain of knowledge, this project has the objective to create a bio-hybrid system with a defined function and application -- growing architectural artifacts.

Social insects interact extensively with their mates. To reveal the mechanism of their sociality, it is important to observe the behavior of each individual in a colony and to analyze their social contexts. Recently, digital video cameras have become less expensive despite increases in their performance, enabling researchers to record the behavior in various places easily. However, much labor and time is needed to analyze the social behavior manually from the video footage. In this study, we propose a new method to detect and track multiple bees moving on a flat surface. Our proposed method consists of three core processes: (i) detecting the bee candidate regions using a background subtraction method and binarization, (ii) identifying individuals by a combination of overlapping information in temporal changes of position and simple prediction of the candidate regions based on the bee's movement, and (iii) outputting the locations and trajectories of the identified bees. Our system succeeds in processing a video of sixteen bees moving freely on a flat arena for three minutes. More than 95% of the bees' central points were successfully extracted and their trajectories precisely traced.

The EU-funded CoCoRo project studies heterogeneous swarms of AUVs used for the purposes of under water monitoring and search. The CoCoRo underwater swarm system will combine bio-inspired motion principles with biologically-derived collective cognition mechanisms to provide a novel robotic system that is scalable, reliable and flexible with respect its behavioural potential. We will investigate and develop swarm-level emergent self-awareness, taking biological inspiration from fish, honeybees, the immune system and neurons. Low-level, local information processing will give rise to collective-level memory and cognition. CoCoRo will develop a novel bio-inspired operating system whose default behaviour will be to provide AUV shoaling functionality and the maintenance of swarm coherence. Collective discrimination of environmental properties will be processed on an individual-or on a collective-level given the cognitive capabilities of the AUVs. We will investigate collective self-recognition through experiments inspired by ethology and psychology, allowing for the quantification of collective cognition.

One of the prominent challenges in mobile robotics is to develop control methodologies that allow the adaptation to dynamic and unforeseen environments. The classic approach of hand-coded controllers is very efficient for well-defined tasks and specific environments but poor in adapting to changing environmental conditions. One alternative approach is the application of evolutionary algorithms which need, in turn, easily evolvable representations of controllers. In this paper, we investigate one promising approach of an artificial hormone system as a control paradigm which is believed to be easily optimized by evolutionary processes. In a first step of this research, we focus on the simple task of collision avoidance. We present a brief mathematical analysis of this controller approach and an implementation of the controller on a mobile robot to check the feasibility in principle of our approach. The task is successfully accomplished and we conclude with a discussion of the hormone dynamics in the robot.

One of the main aspects of the ‘SYMBRION’ and ‘REPLICATOR’ projects is that the robots can aggregate to form a multi-robot organism. For this reason the control mechanisms have to be able to control a single robot, a swarm of robots or an aggregated collective organism. To break down the complexity of development and to take the interaction with the environment and other robots into account, bio-inspired and evolutionary concepts are applied. In this paper we describe the underlying software architecture for the projects to enable different controller types, evolution and learning.

This article presents a bio-inspired communication strategy for large-scale robot swarms. The strategy is based on robot-to-robot interactions without any central communication unit. Thus, the emerging swarm regulates itself in a purely self-organized way. The strategy is biologically inspired by the trophallactic behavior performed by social insects. The experiments shown in this article are performed with a simulation environment that was developed to model the properties of a specific type of swarm robot. We investigated the suggested communication strategy in several arena scenarios and studied the properties of some emergent collective decisions made by the robots