Automatic design is a promising approach to realizing robot swarms. Given a mission to be performed by the swarm, an automatic method produces the required control software for the individual robots. Automatic design has concentrated on missions that a swarm can execute independently, interacting only with a static environment and without the involvement of other active entities. In this paper, we investigate the design of robot swarms that perform their mission by interacting with other robots that populate their environment. We frame our research within robot shepherding: the problem of using a small group of robots—the shepherds— to coordinate a relatively larger group—the sheep. In our study, the group of shepherds is the swarm that is automatically designed, and the sheep are pre-programmed robots that populate its environment. We use automatic modular design and neuroevolution to produce the control software for the swarm of shepherds to coordinate the sheep. We show that automatic design can leverage mission-specific interaction strategies to enable an effective coordination between the two groups.

We investigate the automatic design of robot swarms that perform composite missions—that is, missions specified as the composition of consecutive sub-missions. Automatic design through performance optimization has become a viable and appealing approach to designing robot swarms. First, a user defines a mission by specifying a performance measure: a function indicating to what extent the swarm has attained its goal. An optimization process then generates suitable control software for the robots by maximizing the performance measure. The definition of a performance measure is a challenging task that requires expert input, which hinders the automatic nature of the approach. Recently, inverse reinforcement learning was introduced to minimize the need for human intervention in the automatic design of robot swarms. However, this method was only applied to single-objective missions. In this paper, we extend the method to address composite missions, by formulating and solving the design problem as a multi-objective optimization problem. We conduct simulations with a swarm of twenty e-puck robots that perform twelve composite missions. We compare the performance of the swarm when the robots operate with control software produced manually or using inverse reinforcement learning.

Automatic design is a promising approach to generating control software for robot swarms. So far, automatic design has relied on mission-specific objective functions to specify the desired collective behavior. In this paper, we explore the possibility to specify the desired collective behavior via demonstrations. We develop Demo-Cho, an automatic design method that combines inverse reinforcement learning with automatic modular design of control software for robot swarms. We show that, only on the basis of demonstrations and without the need to be provided with an explicit objective function, Demo-Cho successfully generated control software to perform four missions. We present results obtained in simulation and with physical robots.

Due to the lack of systematic empirical analyses and comparisons of ideas and methods, a clearly established state of the art is still missing in the optimization-based design of robot swarms. In this article, we propose an experimental protocol for the comparison of fully automatic design methods. This protocol is characterized by two notable elements: 1) a way to define benchmarks for the evaluation and comparison of design methods and 2) a sampling strategy that minimizes the variance when estimating their expected performance. To define generally applicable benchmarks, we introduce the notion of mission generator: a tool to generate missions that mimic those a design method will eventually have to solve. To minimize the variance of the performance estimation, we show that, under some common assumptions, one should adopt the sampling strategy that maximizes the number of missions considered—a formal proof is provided as the supplementary material. We illustrate the experimental protocol by comparing the performance of two offline fully automatic design methods that were presented in previous publications.

In this article, we present property-driven design, a novel top-down design method for robot swarms based on prescriptive modeling and model checking. Traditionally, robot swarms have been developed using a code-and-fix approach: in a bottom-up iterative process, the developer tests and improves the individual behaviors of the robots until the desired collective behavior is obtained. The code-and-fix approach is unstructured, and the quality of the obtained swarm depends completely on the expertise and ingenuity of the developer who has little scientific or technical support in his activity. Property-driven design aims at providing such scientific and technical support, with many advantages compared to the traditional unstructured approach. Property-driven design is composed of four phases: first, the developer formally specifies the requirements of the robot swarm by stating its desired properties; second, the developer creates a prescriptive model of the swarm and uses model checking to verify that this prescriptive model satisfies the desired properties; third, using the prescriptive model as a blueprint, the developer implements a simulated version of the desired robot swarm and validates the prescriptive model developed in the previous step; fourth, the developer implements the desired robot swarm and validates the previous steps. We demonstrate property-driven design using two case studies: aggregation and foraging.

We study task partitioning in the context of swarm robotics. Task partitioning is the decomposition of a task into subtasks that can be tackled by different workers. We focus on the case in which a task is partitioned into a sequence of subtasks that must be executed in a certain order. This implies that the subtasks must interface with each other, and that the output of a subtask is used as input for the subtask that follows. A distinction can be made between task partitioning with direct transfer and with indirect transfer. We focus our study on the first case: The output of a subtask is directly transferred from an individual working on that subtask to an individual working on the subtask that follows. As a test bed for our study, we use a swarm of robots performing foraging. The robots have to harvest objects from a source, situated in an unknown location, and transport them to a home location. When a robot finds the source, it memorizes its position and uses dead reckoning to return there. Dead reckoning is appealing in robotics, since it is a cheap localization method and it does not require any additional external infrastructure. However, dead reckoning leads to errors that grow in time if not corrected periodically. We compare a foraging strategy that does not make use of task partitioning with one that does. We show that cooperation through task partitioning can be used to limit the effect of dead reckoning errors. This results in improved capability of locating the object source and in increased performance of the swarm. We use the implemented system as a test bed to study benefits and costs of task partitioning with direct transfer. We implement the system with real robots, demonstrating the feasibility of our approach in a foraging scenario.

Automated algorithm configuration methods have proven to be instrumental in deriving high-performing algorithms and such methods are increasingly often used to configure evolutionary algorithms. One major challenge in devising automatic algorithm configuration techniques is to handle the inherent stochasticity in the configuration problems. This article analyses a post-selection mechanism that can also be used for this task. The central idea of the post-selection mechanism is to generate in a first phase a set of high-quality candidate algorithm configurations and then to select in a second phase from this candidate set the (statistically) best configuration. Our analysis of this mechanism indicates its high potential and suggests that it may be helpful to improve automatic algorithm configuration methods.

In swarm robotics, large groups of relatively simple robots cooperate so that they can perform tasks that go beyond their individual capabilities [1], [2]. The interactions among the robots are based on simple behavioral rules that exploit only local information. The robots in a swarm have neither global knowledge, nor a central controller. Therefore, decisions in the swarm have to be taken in a distributed manner based on local interactions. Because of these limitations, the design of collective decision-making methods in swarm robotic systems is a challenging problem. Moreover, the collective decision-making method must be efficient, robust with respect to robot failures, and scale well with the size of the swarm.

During the last decade, many variants of the original particle swarm optimization (PSO) algorithm have been proposed. In many cases, the difference between two variants can be seen as an algorithmic component being present in one variant but not in the other. In the first part of the paper, we present the results and insights obtained from a detailed empirical study of several PSO variants from a component difference point of view. In the second part of the paper, we propose a new PSO algorithm that combines a number of algorithmic components that showed distinct advantages in the experimental study concerning optimization speed and reliability. We call this composite algorithm Frankenstein's PSO in an analogy to the popular character of Mary Shelley's novel. Frankenstein's PSO performance evaluation shows that by integrating components in novel ways effective optimizers can be designed.

In recent years, much attention has been devoted to the development of metaheuristics and local search algorithms for tackling stochastic combinatorial optimization problems. This paper focuses on local search algorithms; their effectiveness is greatly determined by the evaluation procedure that is used to select the best of several solutions in the presence of uncertainty. In this paper, we propose an effective evaluation procedure that makes use of empirical estimation techniques. We illustrate this approach and we assess its performance on the probabilistic traveling salesman problem. Experimental results on a large set of instances show that the proposed approach can lead to a very fast and highly effective local search algorithm.

Ant colony optimization (ACO) is a promising metaheuristic and a great amount of research has been devoted to its empirical and theoretical analysis. Recently, with the introduction of the hypercube framework, Blum and Dorigo have explicitly raised the issue of the invariance of ACO algorithms to transformation of units. They state (Blum and Dorigo, 2004) that the performance of ACO depends on the scale of the problem instance under analysis. In this paper, we show that the ACO internal state—commonly referred to as the pheromone—indeed depends on the scale of the problem at hand. Nonetheless, we formally prove that this does not affect the sequence of solutions produced by the three most widely adopted algorithms belonging to the ACO family: ant system, MAX-MIN ant system, and ant colony system. For these algorithms, the sequence of solutions does not depend on the scale of the problem instance under analysis. Moreover, we introduce three new ACO algorithms, the internal state of which is independent of the scale of the problem instance considered. These algorithms are obtained as minor variations of ant system, MAX-MIN ant system, and ant colony system. We formally show that these algorithms are functionally equivalent to their original counterparts. That is, for any given instance, these algorithms produce the same sequence of solutions as the original ones.

This paper adopts the idea of regularity in the boundaries of financial time series in order to fit forecasting models which are able to outperform random walk predictions. In particular we propose the adoption of a local learning technique, called lazy learning, in order to perform model estimation and prediction in extreme conditions. The lazy learning method is proposed to return predictions in extreme conditions of trends of the Italian stock market index. The experiments show that in boundary conditions the technique is able to outperform a random predictor and to return a significant rate of accuracy.