Search

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. [ABSTRACT FROM AUTHOR]

Quality-diversity (QD) algorithms evolve behaviorally diverse and high-performing solutions. To illuminate the elite solutions for a space of behaviors, QD algorithms require the definition of a suitable behavior space. If the behavior space is high-dimensional, a suitable dimensionality reduction technique is required to maintain a limited number of behavioral niches. While current methodologies for automated behavior spaces focus on changing the geometry of the behavior space or on unsupervised learning of its key features, there remains a need for customizing behavioral diversity to a particular meta-objective specified by the end user. In the newly emerging framework of QD Meta-Evolution, or QD-Meta for short, one evolves a population of QD algorithms, each with different algorithmic and representational characteristics, to optimize the algorithms and their resulting archives to a user-defined meta-objective. Despite promising results compared to traditional QD algorithms, QD-Meta has yet to be compared to state-of-the-art behavior space automation methods, such as centroidal voronoi tessellations multidimensional archive of phenotypic elites (CVT-MAP-Elites) and autonomous robots realizing their abilities (AURORAs). This article performs an empirical study of QD-Meta on function optimization and multilegged robot locomotion benchmarks. Results demonstrate that QD-Meta archives provide improved average performance and faster adaptation to a priori unknown changes to the environment when compared to CVT-MAP-Elites and AURORA. A qualitative analysis shows how the resulting archives are tailored to the meta-objectives provided by the end user. [ABSTRACT FROM AUTHOR]

In quality-diversity algorithms, the behavioral diversity metric is a key design choice that determines the quality of the evolved archives. Although behavioral diversity is traditionally obtained by describing the observed resulting behavior of robot controllers evaluated in a single environment, it is often more easily induced by introducing environmental diversity, i.e., by manipulating the environments in which the controllers are evaluated. This article proposes quality-environment-diversity (QED), an algorithm that repeatedly generates a random environment according to a probability distribution over environmental features (e.g., number of obstacles, arena size and robot sensor and actuator characteristics), evaluates the controller in that environment, and then describes the controller in terms of the features of that environment, the environment descriptor. Our study compares QED to three baseline task-specific and generic behavioral descriptors, in 5 different robot swarm benchmark tasks. For each task, the quality of the evolved archives is assessed by their capability to provide high-performing compensatory behaviors following injection of 250 unique faults to the robots of the swarm. The evolved archives achieve a median 2- to 3-fold reduction in the impact of the faults on the performance of the swarm. A qualitative analysis of evolved archives is done by visualizing the relation between diversity of compensatory behaviors, here called useful behavioral diversity, and fault recovery metrics. The resulting signatures indicate that, due to the diversity of environments inducing useful behavioral diversity, archives evolved by QED provide robot swarm controllers that are capable of recovering from high-impact faults. [ABSTRACT FROM AUTHOR]

We propose Hyb-CCEA, a cooperative coevolutionary algorithm for the evolution of genetically heterogeneous multiagent teams. The proposed approach extends the cooperative coevolution architecture with operators that put the number of coevolving populations under evolutionary control. Populations are dynamically merged based on behavioral similarity, thus decreasing team heterogeneity, and stochastic population splits are used to explore increased team heterogeneity. Hyb-CCEA is capable of converging to suitable team compositions for the given task, be it a completely homogeneous team where all agents share the same control logic, a heterogeneous team where each agent has distinct control logic, or a partially heterogeneous team. By placing both team composition and agent controllers under evolutionary control, Hyb-CCEA can be applied to domains for which the experimenter has limited or no knowledge about possible solutions. We study Hyb-CCEA extensively in an abstract domain, and conduct a series of validation experiments with four simulated multirobot tasks: two multirover foraging tasks and two robotic soccer tasks. The results show that Hyb-CCEA takes advantage of partial heterogeneity and frequently outperforms the standard cooperative coevolution approach, both in terms of fitness scores achieved and number of evaluations needed to evolve solutions. [ABSTRACT FROM AUTHOR]

The evolution of task-oriented control for robots with complex locomotor systems is currently out of reach for traditional evolutionary computation techniques, as the coordination of a high number of locomotion parameters in response to the robot’s sensory inputs is extremely challenging. Evolutionary techniques have therefore mainly been applied to the optimization of specific locomotion patterns, such as forward motion. In this paper, we explore the evolutionary repertoire-based control (EvoRBC) approach, which divides the synthesis of control into two steps: 1) the evolution of a repertoire of locomotion primitives using a quality diversity algorithm and 2) the evolution of a high-level arbitrator that leverages the locomotion primitives in the repertoire to solve a given task. We comprehensively study the main components of the EvoRBC approach using a four-wheel steering robot. We then conduct a set of experiments in simulation using a hexapod robot. Our results show that EvoRBC is robust to parameter variations, and for all the robots tested, it is able to evolve controllers for a maze navigation task and significantly outperforms the traditional evolutionary robotics approach. [ABSTRACT FROM PUBLISHER]

Biological organisms exist within environments in which complex nonlinear dynamics are ubiquitous. They are coupled to these environments via their own complex dynamical networks of enzyme-mediated reactions, known as biochemical networks. These networks, in turn, control the growth and behavior of an organism within its environment. In this paper, we consider computational models whose structure and function are motivated by the organization of biochemical networks. We refer to these as artificial biochemical networks and show how they can evolve to control trajectories within three behaviorally diverse complex dynamical systems: 1) the Lorenz system; 2) Chirikov's standard map; and 3) legged robot locomotion. More generally, we consider the notion of evolving dynamical systems to control dynamical systems, and discuss the advantages and disadvantages of using higher order coupling and configurable dynamical modules (in the form of discrete maps) within artificial biochemical networks (ABNs). We find both approaches to be advantageous in certain situations, though we note that the relative tradeoffs between different models of ABN strongly depend on the type of dynamical systems being controlled. [ABSTRACT FROM PUBLISHER]

The reality gap, which often makes controllers evolved in simulation inefficient once transferred onto the physical robot, remains a critical issue in evolutionary robotics (ER). We hypothesize that this gap highlights a conflict between the efficiency of the solutions in simulation and their transferability from simulation to reality: the most efficient solutions in simulation often exploit badly modeled phenomena to achieve high fitness values with unrealistic behaviors. This hypothesis leads to the transferability approach, a multiobjective formulation of ER in which two main objectives are optimized via a Pareto-based multiobjective evolutionary algorithm: 1) the fitness; and 2) the transferability, estimated by a simulation-to-reality (STR) disparity measure. To evaluate this second objective, a surrogate model of the exact STR disparity is built during the optimization. This transferability approach has been compared to two reality-based optimization methods, a noise-based approach inspired from Jakobi's minimal simulation methodology and a local search approach. It has been validated on two robotic applications: 1) a navigation task with an e-puck robot; and 2) a walking task with a 8-DOF quadrupedal robot. For both experimental setups, our approach successfully finds efficient and well-transferable controllers only with about ten experiments on the physical robot. [ABSTRACT FROM AUTHOR]

In evolutionary algorithms, much time is spent evaluating inferior phenotypes that produce no offspring. A common heuristic to address this inefficiency is to stop evaluations early if they hold little promise of attaining high fitness. However, the form of this heuristic is typically dependent on the fitness function used, and there is a danger of prematurely stopping evaluation of a phenotype that may have recovered in the remainder of the evaluation period. Here a stopping method is introduced that gradually reduces fitness over the phenotype's evaluation, rather than accumulating fitness. This method is independent of the fitness function used, only stops those phenotypes that are guaranteed to become inferior to the current offspring-producing phenotypes, and realizes significant time savings across several evolutionary robotics tasks. It was found that for many tasks, time complexity was reduced from polynomial to sublinear time, and time savings increased with the number of training instances used to evaluate a phenotype as well as with task difficulty. [ABSTRACT FROM AUTHOR]

In cooperative multiagent systems, agents interact to solve tasks. Global dynamics of multiagent teams result from local agent interactions, and are complex and difficult to predict. Evolutionary computation has proven a promising approach to the design of such teams. The majority of current studies use teams composed of agents with identical control rules ("genetically homogeneous teams") and select behavior at the team level ("team-level selection"). Here we extend current approaches to include four combinations of genetic team composition and level of selection. We compare the performance of genetically homogeneous teams evolved with individual-level selection, genetically homogeneous teams evolved with team-level selection, genetically heterogeneous teams evolved with individual-level selection, and genetically heterogeneous teams evolved with team-level selection. We use a simulated foraging task to show that the optimal combination depends on the amount of cooperation required by the task. Accordingly, we distinguish between three types of cooperative tasks and suggest guidelines for the optimal choice of genetic team composition and level of selection. [ABSTRACT FROM AUTHOR]

One of the major obstacles to achieving robots capable of operating in real-world environments is enabling them to cope with a continuous stream of unanticipated situations. In previous work, it was demonstrated that a robot can autonomously generate self-models, and use those self-models to diagnose unanticipated morphological change such as damage. In this paper, it is shown that multiple physical quadrupedal robots with similar morphologies can share self-models in order to accelerate modeling. Further, it is demonstrated that quadrupedal robots which maintain separate self-modeling algorithms but swap self-models perform better than quadrupedal robots that rely on a shared self-modeling algorithm. This finding points the way toward more robust robot teams: a robot can diagnose and recover from unanticipated situations faster by drawing on the previous experiences of the other robots. [ABSTRACT FROM AUTHOR]

We present a coevolutionary algorithm for inferring the topology and parameters of a wide range of hidden nonlinear systems with a minimum of experimentation on the target system. The algorithm synthesizes an explicit model directly from the observed data produced by intelligently generated tests. The algorithm is composed of two coevolving populations. One population evolves candidate models that estimate the structure of the hidden system. The second population evolves informative tests that either extract new information from the hidden system or elicit desirable behavior from it. The fitness of candidate models is their ability to explain behavior of the target system observed in response to all tests carried out so far; the fitness of candidate tests is their ability to make the models disagree in their predictions. We demonstrate the generality of this estimation-exploration algorithm by applying it to four different problems---grammar induction, gene network inference, evolutionary robotics, and robot damage recovery--and discuss how it overcomes several of the pathologies commonly found in other coevolutionary algorithms. We show that the algorithm is able to successfully infer and/or manipulate highly nonlinear hidden systems using very few tests, and that the benefit of this approach increases as the hidden systems possess more degrees of freedom, or become more biased or unobservable. The algorithm provides a systematic method for posing synthesis or analysis tasks to a coevolutionary system. [ABSTRACT FROM AUTHOR]

We propose a novel perspective on the use of evolutionary multiobjective optimization (EMO) as a paradigm for evolving embodied organisms and as a framework for characterizing complexity. The paper demonstrates novel experiments that show the power of EMO in generating robots with different morphologies, yet with very similar locomotion abilities. The proposed framework for comparing the complexity of an object across different complexity measures allowed meaningful and quantifiable comparisons between the evolved organisms. We show empirically that the partial order feature inherited in the Pareto concept exhibits characteristics which are suitable for comparing between the complexities of artificially evolved embodied organisms. [ABSTRACT FROM AUTHOR]