Evolving Neural Network Controllers for a Team ofSelf-organizing RobotsEvolving Neural Network Controllers for a Team ofSelf-organizing Robots

Abstract—We present and evaluate new ROS packages for coordinated multi-robot exploration, namely communication, global map construction, and exploration. The packages allow completely distributed control and do not rely on (but allow) central controllers. Their integration including application layer protocols allows out of the box installation and execution. The communication package enables reliable ad hoc communication allowing to exchange local maps between robots which are merged to a global map. Exploration uses the global map to spatially spread robots and decrease exploration time. The intention of the implementation is to offer basic functionality for coordinated multi-robot systems and to enable other research groups to experimentally work on multi-robot systems. The packages are tested in real-world experiments using Turtlebot and Pioneer robots. Further, we analyze their performance using simulations and verify their correct working.

Major challenges for the transition of power sys- tems do not only tackle power electronics but also communication technology, power market economy and user acceptance studies. Simulation is an important research method therein, as it helps to avoid costly failures. A common smart grid simulation platform is still missing. We introduce a conceptual model of agents in multiple flow networks. Flow networks extend the depth of established power flow analysis through use of networks of information flow and financial transactions. We use this model as a basis for comparing different power system simulators. Furthermore, a quantitative comparison of simulators is done to facilitate the decision for a suitable tool in comprehensive smart grid simulation.

The idea of changing our energy system from a hierarchical design into a set of nearly inde- pendent microgrids becomes feasible with the availability of small renewable energy generators. The smart microgrid concept comes with several challenges in research and engineering tar- geting load balancing, pricing, consumer integration and home automation. In this paper we first provide an overview on these challenges and present approaches that target the problems identified. While there exist promising algorithms for the particular field, we see a missing integration which specifically targets smart microgrids. Therefore, we propose an architec- ture that integrates the presented approaches and defines interfaces between the identified components such as generators, storage, smart and “dumb” devices.

To improve the energy awareness of consumers, it is necessary to provide them with information about their en- ergy demand, not just on the household level. Non-intrusive load monitoring (NILM) gives the consumer the opportunity to disaggregate their consumed power on the appliance level. The consumer is provided with information about the energy de- mand of each individual appliances. In this paper we present an evolutionary optimization algorithm, applicable to NILM purposes. It can be used to detect appliances with a prob- abilistic power demand model. We show that the detection performance of the evolutionary algorithm can be improved if the single population approach of the evolutionary algorithm is replaced by a parallel population approach with individual exchange and by the introduction of application-oriented pre- processing and mutation methods. The proposed algorithm is tested with Matlab simulations and is evaluated according to the fitness reached and detection probability of the algorithm.

Households account for a significant fraction of overall energy consumption. Energy usage can be reduced by improving the efficiency of devices and optimizing their use as well as by encouraging people to change their behaviour towards a more sustainable lifestyle. In this study, we investigate patterns of domestic energy use in Carinthia (Austria) and Friuli-Venezia Giulia (Italy). In particular, we report the results of an online survey about electrical devices and their use in households. We outline typical scenarios in the two regions and discuss possible strategies to reduce the consumption of energy in these regions.

Purpose: Content delivery in dynamic networks is a challenging task, because paths may change during delivery and content might get lost. Replication is a typical measure to increase robustness and performance. Method: In previous work we proposed a hormone-based algorithm that delivers content, and optimizes the distribution of replicas. Clients express demands by creating hormones that will be released to the network. The corresponding resources are attracted by this hormone and travel towards a higher hormone concentration. This leads to a placement of content close to their most frequent requesters. In addition to that the hormone-based delivery requires an appropriate replication and clean-up strategy to balance the replicas throughout the network without exceeding the nodes’ storage limits or the networks communication capacity.

This paper describes the design and implementation of a distributed self-stabilizing clock synchronization algorithm based on the biological example of Asian Fireflies. Huge swarms of these fireflies use the principle of pulse coupled oscillators in order to synchronously emit light flashes to attract mating partners. When applying this algorithm to real sensor networks, typically, nodes cannot receive messages while transmitting, which prevents the networked nodes from reaching synchronization. In order to counteract this deafness problem, we adopt a variant of the Reachback Firefly Algorithm to distribute the timing of light flashes in a given time window without affecting the quality of the synchronization. A case study implemented on 802.15.4 Zigbee nodes presents the application of this approach for a time-triggered communication scheduling and coordinated duty cycling in order to enhance the battery lifetime of the nodes.

Typically, self-organizing systems comprise of a large number of individual agents whose behavior needs to be controlled by set parameters so that their interactions lead to the creation of the desired system. To be self-organizing, the system must mimic the evolutionary process. One way to do this is by use of an evolutionary algorithm. This mimics naturally-occurring genetic variation (mutation and recombination of genes). To fulfill this purpose, we have created a tool named FREVO (FRamework for EVOlutionary design), which separates the input needed into the following components: target problem evaluation, controller repre- sentation and the optimization method. FREVO provides well-defined interfaces for these components and supports a graphical user interface to simulate the evolutionary process. After obtaining the outcome for a simulation, it is possible to validate and evaluate the results within FREVO. FREVO has been successfully applied to various problems, from cooperative robotics to economics, pattern generation and wireless sensor networks. In this paper, we give an overview of the architecture of FREVO and introduce a case study involving smart grid networks.

Self-organizing systems obtain a global system behavior via typically simple local interactions among a number of components or agents, respectively. The emergent ser- vice often displays properties like adaptability, robust- ness, and scalability, which makes the self-organizing paradigm interesting for technical applications like coop- erative autonomous robots. The behavior for the local in- teractions is usually simple, but it is often difficult to de- fine the right set of interaction rules in order to achieve a desired global behavior. In this paper we describe a novel design approach using an evolutionary algorithm and ar- tificial neural networks to automatize the part of the de- sign process that requires most of the effort. A simulated robot soccer game was implemented to test and evaluate the proposed method. A new approach in evolving com- petitive behavior is also introduced using Swiss System instead of the full tournament to cut down the number of necessary simulations.

Self-organizing systems obtain a global system behavior via typically simple local interactions among a number of components or agents, respectively. The emergent ser- vice often displays properties like adaptability, robust- ness, and scalability, which makes the self-organizing paradigm interesting for technical applications like coop- erative autonomous robots. The behavior for the local in- teractions is usually simple, but it is often difficult to de- fine the right set of interaction rules in order to achieve a desired global behavior. In this paper we describe a novel design approach using an evolutionary algorithm and ar- tificial neural networks to automatize the part of the de- sign process that requires most of the effort. A simulated robot soccer game was implemented to test and evaluate the proposed method. A new approach in evolving com- petitive behavior is also introduced using Swiss System instead of the full tournament to cut down the number of necessary simulations.