Search

This article addresses the co-evolution of morphology and control in evolutionary robotics, focusing on the challenge of premature convergence and limited morphological diversity. We conduct a comparative analysis of state-of-the-art algorithms, focusing on QD (Quality-Diversity) algorithms, based on a well-defined methodology for benchmarking evolutionary algorithms. We introduce carefully chosen indicators to evaluate their performance in three core aspects: task performance, phenotype diversity, and genotype diversity. Our findings highlight MNSLC (Multi-BC NSLC), with the introduction of aligned novelty to NSLC (Novelty Search with Local Competition), as the most effective algorithm for diversity preservation (genotype and phenotype diversity), while maintaining a competitive level of exploitability (task performance). MAP-Elites, although exhibiting a well-balanced trade-off between exploitation and exploration, fall short in protecting morphological diversity. NSLC, while showing similar performance to MNSLC in terms of exploration, is the least performant in terms of exploitation, contrasting with QN (Fitness-Novelty MOEA), which exhibits much superior exploitation, but inferior exploration, highlighting the effects of local competition in skewing the balance toward exploration. Our study provides valuable insights into the advantages, disadvantages, and trade-offs of different algorithms in co-evolving morphology and control. [ABSTRACT FROM AUTHOR]

We introduce a method that permits to co-evolve the body and the control properties of robots. It can be used to adapt the morphological traits of robots with a hand-designed morphological bauplan or to evolve the morphological bauplan as well. Our results indicate that robots with co-adapted body and control traits outperform robots with fixed hand-designed morphologies. Interestingly, the advantage is not due to the selection of better morphologies but rather to the mutual scaffolding process that results from the possibility to co-adapt the morphological traits to the control traits and vice versa. Our results also demonstrate that morphological variations do not necessarily have destructive effects on robots' skills. [ABSTRACT FROM AUTHOR]

Biological vision incorporates intelligent cooperation between the sensory and the motor systems, which is facilitated by the development of motor skills that help to shape visual information that is relevant to a specific vision task. In this article, we seek to explore an approach to active vision inspired by biological systems, which uses limited constraints for motor strategies through progressive adaptation via an evolutionary method. This type of approach gives freedom to artificial systems in the discovery of eye-movement strategies that may be useful to solve a given vision task but are not known to us. In the experiment sections of this article, we use this type of evolutionary active vision system for more complex natural images in both two-dimensional (2D) and three-dimensional (3D) environments. To further improve the results, we experiment with the use of pre-processing the visual input with both the uniform local binary patterns (ULBP) and the histogram of oriented gradients (HOG) for classification tasks in the 2D and 3D environments. The 3D experiments include application of the active vision system to object categorisation and indoor versus outdoor environment classification. Our experiments are conducted on the iCub humanoid robot simulator platform. [ABSTRACT FROM AUTHOR]

Modularity is a system property of many natural and artificial adaptive systems. Evolutionary algorithms designed to produce modular solutions have increased convergence rates and improved generalization ability; however, their performance can be impacted if the task is inherently nonmodular. Previously, we have shown that some design variables can influence whether the task on the remaining variables is inherently modular. We investigate the possibility of exploiting that dependence to simplify optimization and arrive at a general design pattern that we use to show that evolutionary search can seek such modularity-inducing design variable values, thus easing subsequent search for highly fit, modular organization within the remaining design variables. We investigate this approach with embodied agents in which evolutionary discovery of morphology enables subsequent discovery of highly fit, modular controllers and show that it benefits from biasing search toward modular controllers and setting the mutation rate for control policies higher than that for morphology. This work also reinforces our previous finding that the relationship between modularity and evolvability that is well-studied in nonembodied systems can, under certain conditions, be generalized to include embodied systems as well and provides a practical approach to satisfying the conditions in question. [ABSTRACT FROM AUTHOR]

This paper utilizes evolutionary robotics techniques as a hypothesis generator to explore optical variables and control strategies that could be used to solve a driving-like braking task. Given such a task, humans exhibit two different braking behaviors: continuously regulated braking and impulsive braking. Based on an oft-used experimental task in human perception/action research, a series of evolutionary robotics simulations were developed to explore the space of possible braking strategies by examining how braking strategies change as the optical information is manipulated. Our results can be summarized as follows: (1) behaviors similar to human behavior were observed only when the constraints were selected correctly; (2) the optical variables τ and proportional rate yielded significantly better braking performance; (3) two different classes of impulsive braking behaviors were observed, including one not reported in previous studies: discrete impulsive braking and oscillatory impulsive braking; (4) the optical variable τ is used to initiate and terminate braking; (5) the evolved model agents use a proportional rate control strategy to regulate braking continuously. We argue that combining psychological experiments and evolutionary robotics simulations is a promising research methodology that is useful for testing existing hypotheses and generating new ones. [ABSTRACT FROM PUBLISHER]

The common approach for training robots is to expose them to different environmental scenarios, training their controllers to have the best possible commands when untrained scenarios are encountered. When humans train, they do the same. They try new manipulations by performing within different environments. However, humans training also includes a type of training which, although claimed to improve cognitive capabilities, has not, to date, been adopted for the training of robots. This type of training involves the restriction of manipulation capabilities while performing different tasks, e.g. climbing with just one hand. Recently research that aims at exploring the invigorating idea that such training, termed as mechanical cognitivization (MC), would enhance the robustness of robots, has been published. However, that work has demonstrated the idea by utilizing mathematical functions. In the current paper, the first proof that such training may enhance the robustness of robots is given by using a kinematic system. Specifically, it is shown that such training improves the performances of robots when encountering the need to perform new tasks, when performing within untrained environments and when malfunctions occur. The advantages of the suggested training are highlighted through facilitating a comparison between three learning schemes that include a common neural net and two MC-based nets. These are termed here as amalgamated modes neural net and committee of modes neural net. It is shown that both MC-based schemes are superior when robustness is considered. [ABSTRACT FROM PUBLISHER]

We introduce FARSA, an open-source Framework for Autonomous Robotics Simulation and Analysis, that allows us to easily set up and carry on adaptive experiments involving complex robot/environmental models. Moreover, we show how a simulated iCub robot can be trained, through an evolutionary algorithm, to display reaching and integrated reaching and grasping behaviours. The results demonstrate how the use of an implicit selection criterion, estimating the extent to which the robot is able to produce the expected outcome without specifying the manner through which the action should be realized, is sufficient to develop the required capabilities despite the complexity of the robot and of the task. [ABSTRACT FROM AUTHOR]

The kinematics of human walking are largely driven by passive dynamics, but adaptation to varying terrain conditions and responses to perturbations require some form of active control. The basis for this control is often thought to take the form of entrainment between a neural oscillator (i.e., a central pattern generator and/or distributed counterparts) and the mechanical system. Here we use techniques in evolutionary robotics to explore the potential of a purely reactive, linear controller to control bipedal locomotion over rough terrain. In these simulation studies, joint torques are computed as weighted linear sums of sensor states, and the weights are optimized using an evolutionary algorithm. We show that linear reactive control can enable a seven-link 2D biped and a nine-link 3D biped to walk over rough terrain (steps of ∼5% leg length or more in the 2D case). In other words, the simulated walker gradually learns the appropriate weights to achieve stable locomotion. The results indicate that oscillatory neural structures are not necessarily a requirement for robust bipedal walking. The study of purely reactive control through linear feedback may help to reveal some basic control principles of stable walking. [ABSTRACT FROM AUTHOR]

Walking control of biped robots is a challenging problem, and improving robustness to noise and uncertainty remains difficult. We recently developed a novel control framework for 3D bipedal walking that we call “linear reactive control.” It is linear because control torques are computed as simple weighted sums of sensor states. It is reactive because it depends only on the model’s current state. The present simulation study shows that this controller performs reliably in the presence of realistic models of joint actuation, sensor noise, and uncertainty in model and contact parameters. The controller is able to maintain a stable gait in the presence of noisy sensor inputs and low-impedance actuation. It also performs reliably on models with high uncertainty (up to 20%) in measurements of their dynamic parameters and widely varying ground contact parameters. The robustness of this controller to realistic conditions validates this method as a promising avenue for bipedal control. [ABSTRACT FROM PUBLISHER]

This article describes a robotic system which uses evolution to continuously adapt a group of heterogeneous robots to their current environment while assigning tasks to these robots using an endocrine-based system. The tasks are allocated dependent on the robots’ current ability to perform the task and whether the task is being done by another robot. A series of experiments is presented taking the work from an evolutionary training phase, through simulation trials, to experiments on real robots. The real robot trials show task swapping dependent on the robots’ ability to perform each task. [ABSTRACT FROM PUBLISHER]

In evolutionary robotics, plastic neural network models proved to be promising for evolving adaptive behaviors. In particular, neurocontrollers incorporating hebbian synapses have been shown to be useful for implementing conflicting sub-behaviors. Numerous interesting complex tasks assume such flexibility. However, those evolved controllers often exhibit behavioral instability, as simulation time is extended beyond the short limit used during evolution. In this paper, we propose constrained plastic models inspired by neural homeostasis phenomena, in order to evolve flexible and stable pattern generators for single-legged locomotion. Comparative results show that constrained controllers perform better than unconstrained ones in both terms of evolvability and behavioral stability. Functional analyses of the best evolved controller unveil the adaptivity, robustness and homeostasis arising from the statically constrained plasticity. Interestingly, homeostasis evolved implicitly without relying on any active homeostatic mechanisms and is implemented through hebbian plasticity, usually considered destabilizing. [ABSTRACT FROM PUBLISHER]

In this article we propose a framework for performing embodied evolution with a limited number of robots, by utilizing time-sharing in subpopulations of virtual agents hosted in each robot. Within this framework, we explore the combination of within-generation learning of basic survival behaviors by reinforcement learning, and evolutionary adaptations over the generations of the basic behavior selection policy, the reward functions, and metaparameters for reinforcement learning. We apply a biologically inspired selection scheme, in which there is no explicit communication of the individuals’ fitness information. The individuals can only reproduce offspring by mating—a pair-wise exchange of genotypes—and the probability that an individual reproduces offspring in its own subpopulation is dependent on the individual’s ‘‘health,’’ that is, energy level, at the mating occasion. We validate the proposed method by comparing it with evolution using standard centralized selection, in simulation, and by transferring the obtained solutions to hardware using two real robots. [ABSTRACT FROM PUBLISHER]

This article rigorously characterizes the structure of the entire fitness space of a simple neuromechanical system consisting of a model leg in closed-loop interaction with a neural controller. Using tools from the theory of piecewise-smooth dynamical systems, we derive expressions for the location and layout of the region of high-fitness solutions in parameter space, and we show how both the boundary and the internal structure of this region arise from specific neural, mechanical, and neuromechanical properties of the walking system. In addition, we characterize the structure of the map from neural parameters to gaits to fitness. [ABSTRACT FROM AUTHOR]

This article extends previous work on evolving learning without synaptic plasticity from discrete tasks to continuous tasks. Continuous-time recurrent neural networks without synaptic plasticity are artificially evolved on an associative learning task. The task consists in associating paired stimuli: temperature and food. The temperature to be associated can be either drawn from a discrete set or allowed to range over a continuum of values. We address two questions: Can the learning without synaptic plasticity approach be extended to continuous tasks? And if so, how does learning without synaptic plasticity work in the evolved circuits? Analysis of the most successful circuits to learn discrete stimuli reveal finite state machine (FSM) like internal dynamics. However, when the task is modified to require learning stimuli on the full continuum range, it is not possible to extract a FSM from the internal dynamics. In this case, a continuous state machine is extracted instead. [ABSTRACT FROM AUTHOR]

The interaction of brain, body, and environment can result in complex behavior with rich dynamics, even for relatively simple agents. Such dynamics are, however, often difficult to analyze. In this article, we explore the case of a simple simulated robotic agent, equipped with a reactive neurocontroller and an energy level, which the agent has been evolved to recharge. A dynamical systems analysis shows that a non-neural internal state (energy level), despite its simplicity, dynamically modulates the behavioral attractors of the agent-environment system, such that the robot's behavioral repertoire is continually adapted to its current situation and energy level. What emerges is a dynamic, non-deterministic, and highly self-organized action selection mechanism, originating from the dynamical coupling of four systems (non-neural internal states, neurocontroller, body, and environment) operating at very different timescales. [ABSTRACT FROM AUTHOR]

Enactivism claims that sensory-motor activity and embodiment are crucial in perceiving the environment and that machine vision could be a much simpler business if considered in this context. However, computational models of enactive vision are very rare and often rely on handcrafted control systems. In this article, we argue that the apparent complexity of the environment and of the robot brain can be significantly simplified if perception, behavior, and learning are allowed to co-develop on the same timescale. In doing so, robots become sensitive to, and actively exploit, characteristics of the environment that they can tackle within their own computational and physical constraints. We describe the application of this methodology in three sets of experiments: shape discrimination, car driving, and wheeled robot navigation. A further set of experiments, where the visual system can develop the receptive fields by means of unsupervised Hebbian learning, demonstrates that the receptive fields are consistently and significantly affected by the behavior of the system and differ from those predicted by most computational models of the visual cortex. Finally, we show that our robots can also replicate the performance deficiencies observed in experiments of motor deprivation with kittens. [ABSTRACT FROM AUTHOR]

In this article, we evolve and analyze continuous-time recurrent neural networks capable of associating the smells of different foods with edibility or inedibility in different environments. First, we present an in-depth analysis of this task, highlighting the evolutionary challenges it poses and how these challenges informed our experimental design. Next, we describe the evolution of nonplastic neural circuits that can solve this food edibility learning problem. We then show that the dynamics of the best evolved nonplastic circuits instantiate finite state machines that capture the combinatorial structure of this task. Finally, we demonstrate that successful circuits with Hebbian synaptic plasticity can also be evolved, but that such circuits do not utilize their synaptic plasticity in a traditional way. [ABSTRACT FROM AUTHOR]

A preference is not located anywhere in the agent's cognitive architecture, but it is rather a constraining of behavior which is in turn shaped by behavior. Based on this idea, a minimal model of behavioral preference is proposed. A simulated mobile agent is modeled with a plastic neurocontroller, which holds two separate high dimensional homeostatic boxes in the space of neural dynamics. An evolutionary algorithm is used for creating a link between the boxes and the performance of two different phototactic behaviors. After evolution, the agent's performance exhibits some important aspects of behavioral preferences such as durability and transitions. This article demonstrates (1) the logical consistency of the multi-causal view by producing a case study of its viability and providing insights into its dynamical basis and (2) how durability and transitions arise through the mutual constraining of internal and external dynamics in the flow of alternating high and low susceptibility to environmental variations. Implications for modeling autonomy are discussed. [ABSTRACT FROM AUTHOR]

This paper introduces the evolvable functional circle hypothesis. This hypothesis states that if it is assumed that von Uexküll's concept of functional circles exists in robots and that the models used in evolutionary robotics are altered accordingly, then practitioners of evolutionary robotics will benefit in two ways. The first way is that by promoting the evolution of functional circles rather than sensorimotor loops, it allows evolved robots to select their own stimuli and therefore find their own meaning from environmental information. The second way is that it makes it easier for evolved robots to derive multiple meanings from the signal produced by a sensor. The paper goes on to demonstrate a method to alter our models in evolutionary robotics to promote the evolution of functional circles. [ABSTRACT FROM AUTHOR]

Diverse, complex, and adaptive animal behaviors are achieved by organizing hierarchically structured controllers in motor systems. The levels of control progress from simple spinal reflexes and central pattern generators through to executive cognitive control in the frontal cortex. Various types of hierarchical control structures have been introduced and shown to be effective in past artificial agent models, but few studies have shown how such structures can self-organize. This study describes how such hierarchical control may evolve in a simple recurrent neural network model implemented in a mobile robot. Topological constraints on information flow are found to improve system performance by decreasing interference between different parts of the network. One part becomes responsible for generating lower behavior primitives while another part evolves top-down sequencing of the primitives for achieving global goals. Fast and slow neural response dynamics are automatically generated in specific neurons of the lower and the higher levels, respectively. A hierarchical neural network is shown to outperform a comparable single-level network in controlling a mobile robot. [ABSTRACT FROM AUTHOR]

We discuss the methodological foundations for our work on the development of cognitive architectures, or control systems, for situated autonomous agents. Our focus is the problems of developing sensorimotor control systems for mobile robots, but we also discuss the applicability of our approach to the study of biological systems. We argue that, for agents required to exhibit sophisticated interactions with their environments, complex sensorimotor processing is necessary, and the design, by hand, of control systems capable of such processing is likely to become prohibitively difficult as complexity increases. We propose an automatic design process involving artificial evolution, wherein the basic building blocks used for evolving cognitive architectures are noise-tolerant dynamical neural networks. These networks may be recurrent and should operate in real time. The evolution should be incremental, using an extended and modified version of a genetic algorithm.Practical constraints suggest that initial architecture evaluations should be done largely in simulation. To support our claims and proposals, we summarize results from some preliminary simulation experiments in which visually guided robots are evolved to operate in simple environments. Significantly, our results demonstrate that robust visually guided control systems evolve from evaluation functions that do not explicitly require monitoring visual input. We outline the difficulties involved in continuing with simulations and conclude by describing specialized visuorobotic equipment, designed to eliminate the need for simulated sensors and actuators. [ABSTRACT FROM PUBLISHER]

The author presents robotic engineer Takashi Gomi as a visionary and gives him the credit for introducing Artificial Intelligence( AI) and evolutionary robotics in Japan.Topics include the author's research work on embodied AI at Ecole Polytechnique Federale de Lausanne (EPFL) in Switzerland; Khepera, a research robot which could contribute to embodied artificial intelligence; and the concept of soft robots.

The article presents information about the Department of Human and Artificial Intelligence Systems (HART) at the University of Fukui in Fukui city, Japan. It mentions that Dr.Takashi Gomi, a robotic researcher helped the university in starting this department by arranging symposiums on evolutionary robotics (ER). A message to Dr.Gomi from Professor Rodney A. Brooks of Massachusetts Institute of Technology, Massachusetts for the HART 2002 symposium is included.