GECCO '18- Proceedings of the Genetic and Evolutionary Computation ConferenceFull Citation in the ACM Digital Library
SESSION: Complex systems (artificial life/artificial immune systems/generative and developmental systems/evolutionary robotics/evolvable hardware)
This research considers the task of evolving the physical structure of a robot to enhance its performance in various environments, which is a significant problem in the field of Evolutionary Robotics. Inspired by the fields of evolutionary art and sculpture, we evolve only targeted parts of a robot, which simplifies the optimisation problem compared to traditional approaches that must simultaneously evolve both (actuated) body and brain. Exploration fidelity is emphasised in areas of the robot most likely to benefit from shape optimisation, whilst exploiting existing robot structure and control. Our approach uses a Genetic Algorithm to optimise collections of Bezier splines that together define the shape of a legged robot's tibia, and leg performance is evaluated in parallel in a high-fidelity simulator. The leg is represented in the simulator as 3D-printable file, and as such can be readily instantiated in reality. Provisional experiments in three distinct environments show the evolution of environment-specific leg structures that are both high-performing and notably different to those evolved in the other environments. This proof-of-concept represents an important step towards the environment-dependent optimisation of performance-critical components for a range of ubiquitous, standard, and already-capable robots that can carry out a wide variety of tasks.
Enabling artificial agents to automatically learn complex, versatile and high-performing behaviors is a long-lasting challenge. This paper presents a step in this direction with hierarchical behavioral repertoires that stack several behavioral repertoires to generate sophisticated behaviors. Each repertoire of this architecture uses the lower repertoires to create complex behaviors as sequences of simpler ones, while only the lowest repertoire directly controls the agent's movements. This paper also introduces a novel approach to automatically define behavioral descriptors thanks to an unsupervised neural network that organizes the produced high-level behaviors. The experiments show that the proposed architecture enables a robot to learn how to draw digits in an unsupervised manner after having learned to draw lines and arcs. Compared to traditional behavioral repertoires, the proposed architecture reduces the dimensionality of the optimization problems by orders of magnitude and provides behaviors with a twice better fitness. More importantly, it enables the transfer of knowledge between robots: a hierarchical repertoire evolved for a robotic arm to draw digits can be transferred to a humanoid robot by simply changing the lowest layer of the hierarchy. This enables the humanoid to draw digits although it has never been trained for this task.
How swarm size during evolution impacts the behavior, generalizability, and brain complexity of animats performing a spatial navigation task
While it is relatively easy to imitate and evolve natural swarm behavior in simulations, less is known about the social characteristics of simulated, evolved swarms, such as the optimal (evolutionary) group size, why individuals in a swarm perform certain actions, and how behavior would change in swarms of different sizes. To address these questions, we used a genetic algorithm to evolve animats equipped with Markov Brains in a spatial navigation task that facilitates swarm behavior. The animats' goal was to frequently cross between two rooms without colliding with other animats. Animats were evolved in swarms of various sizes. We then evaluated the task performance and social behavior of the final generation from each evolution when placed with swarms of different sizes in order to evaluate their generalizability across conditions. According to our experiments, we find that swarm size during evolution matters: animats evolved in a balanced swarm developed more flexible behavior, higher fitness across conditions, and, in addition, higher brain complexity.
Surrogate-assistance approaches have long been used in computationally expensive domains to improve the data-efficiency of optimization algorithms. Neuroevolution, however, has so far resisted the application of these techniques because it requires the surrogate model to make fitness predictions based on variable topologies, instead of a vector of parameters. Our main insight is that we can sidestep this problem by using kernel-based surrogate models, which require only the definition of a distance measure between individuals. Our second insight is that the well-established Neuroevolution of Augmenting Topologies (NEAT) algorithm provides a computationally efficient distance measure between dissimilar networks in the form of "compatibility distance", initially designed to maintain topological diversity. Combining these two ideas, we introduce a surrogate-assisted neuroevolution algorithm that combines NEAT and a surrogate model built using a compatibility distance kernel. We demonstrate the data-efficiency of this new algorithm on the low dimensional cart-pole swing-up problem, as well as the higher dimensional half-cheetah running task. In both tasks the surrogate-assisted variant achieves the same or better results with several times fewer function evaluations as the original NEAT.
Traditional evolutionary algorithms tend to converge to a single good solution, which can limit their chance of discovering more diverse and creative outcomes. Divergent search, on the other hand, aims to counter convergence to local optima by avoiding selection pressure towards the objective. Forms of divergent search such as novelty or surprise search have proven to be beneficial for both the efficiency and the variety of the solutions obtained in deceptive tasks. Importantly for this paper, early results in maze navigation have shown that combining novelty and surprise search yields an even more effective search strategy due to their orthogonal nature. Motivated by the largely unexplored potential of coupling novelty and surprise as a search strategy, in this paper we investigate how fusing the two can affect the evolution of soft robot morphologies. We test the capacity of the combined search strategy against objective, novelty, and surprise search, by comparing their efficiency and robustness, and the variety of robots they evolve. Our key results demonstrate that novelty-surprise search is generally more efficient and robust across eight different resolutions. Further, surprise search explores the space of robot morphologies more broadly than any other algorithm examined.
The presence of functional diversity within a group has been demonstrated to lead to greater robustness, higher performance and increased problem-solving ability in a broad range of studies that includes insect groups, human groups and swarm robotics. Evolving group diversity however has proved challenging within Evolutionary Robotics, requiring reproductive isolation and careful attention to population size and selection mechanisms. To tackle this issue, we introduce a novel, decentralised, variant of the MAP-Elites illumination algorithm which is hybridised with a well-known distributed evolutionary algorithm (mEDEA). The algorithm simultaneously evolves multiple diverse behaviours for multiple robots, with respect to a simple token-gathering task. Each robot in the swarm maintains a local archive defined by two pre-specified functional traits which is shared with robots it come into contact with. We investigate four different strategies for sharing, exploiting and combining local archives and compare results to mEDEA. Experimental results show that in contrast to previous claims, it is possible to evolve a functionally diverse swarm without geographical isolation, and that the new method outperforms mEDEA in terms of the diversity, coverage and precision of the evolved swarm.
Typically, AI researchers and roboticists try to realize intelligent behavior in machines by tuning parameters of a predefined structure (body plan and/or neural network architecture) using evolutionary or learning algorithms. Another but not unrelated longstanding property of these systems is their brittleness to slight aberrations, as highlighted by the growing deep learning literature on adversarial examples. Here we show robustness can be achieved by evolving the geometry of soft robots, their control systems, and how their material properties develop in response to one particular interoceptive stimulus (engineering stress) during their lifetimes. By doing so we realized robots that were equally fit but more robust to extreme material defects (such as might occur during fabrication or by damage thereafter) than robots that did not develop during their lifetimes, or developed in response to a different interoceptive stimulus (pressure). This suggests that the interplay between changes in the containing systems of agents (body plan and/or neural architecture) at different temporal scales (evolutionary and developmental) along different modalities (geometry, material properties, synaptic weights) and in response to different signals (interoceptive and external perception) all dictate those agents' abilities to evolve or learn capable and robust strategies.
While neuroevolution (evolving neural networks) has been successful across a variety of domains from reinforcement learning, to artificial life, to evolutionary robotics, it is rarely applied to large, deep neural networks. A central reason is that while random mutation generally works in low dimensions, a random perturbation of thousands or millions of weights will likely break existing functionality. This paper proposes a solution: a family of safe mutation (SM) operators that facilitate exploration without dramatically altering network behavior or requiring additional interaction with the environment. The most effective SM variant scales the degree of mutation of each individual weight according to the sensitivity of the network's outputs to that weight, which requires computing the gradient of outputs with respect to the weights (instead of the gradient of error, as in conventional deep learning). This safe mutation through gradients (SM-G) operator dramatically increases the ability of a simple genetic algorithm-based neuroevolution method to find solutions in high-dimensional domains that require deep and/or recurrent neural networks, including domains that require processing raw pixels. By improving our ability to evolve deep neural networks, this new safer approach to mutation expands the scope of domains amenable to neuroevolution.
For robots to handle the numerous factors that can affect them in the real world, they must adapt to changes and unexpected events. Evolutionary robotics tries to solve some of these issues by automatically optimizing a robot for a specific environment. Most of the research in this field, however, uses simplified representations of the robotic system in software simulations. The large gap between performance in simulation and the real world makes it challenging to transfer the resulting robots to the real world. In this paper, we apply real world multi-objective evolutionary optimization to optimize both control and morphology of a four-legged mammal-inspired robot. We change the supply voltage of the system, reducing the available torque and speed of all joints, and study how this affects both the fitness, as well as the morphology and control of the solutions. In addition to demonstrating that this real-world evolutionary scheme for morphology and control is indeed feasible with relatively few evaluations, we show that evolution under the different hardware limitations results in comparable performance for low and moderate speeds, and that the search achieves this by adapting both the control and the morphology of the robot.
This paper presents an evolutionary computing based approach to automatically synthesise swarm behavioural rules from their atomic components, thus making a step forward in trying to mitigate human bias from the rule generation process, and leverage the full potential of swarm systems in the real world by modelling more complex behaviours. We identify four components that make-up the structure of a rule: control structures, parameters, logical/relational connectives and preliminary actions, which form the rule space for the proposed approach. A boids simulation system is employed to evaluate the approach with grammatical evolution and genetic programming techniques using the rule space determined. While statistical analysis of the results demonstrates that both methods successfully evolve desired complex behaviours from their atomic components, the grammatical evolution model shows more potential in generating complex behaviours in a modularised approach. Furthermore, an analysis of the structure of the evolved rules implies that the genetic programming approach only derives non-reusable rules composed of a group of actions that is combined to result in emergent behaviour. In contrast, the grammatical evolution approach synthesises sound and stable behavioural rules which can be extracted and reused, hence making it applicable in complex application domains where manual design is infeasible.
Traditional Evolutionary Robotics (ER) employs evolutionary techniques to search for a single monolithic controller which can aid a robot to learn a desired task. These techniques suffer from boot-strap and deception issues when the tasks are complex for a single controller to learn. Behaviour-decomposition techniques have been used to divide a task into multiple subtasks and evolve separate subcontrollers for each subtask. However, these subcontrollers and the associated subcontroller arbitrator(s) are all evolved off-line. A distributed, fully embodied and evolutionary version of such approaches will greatly aid online learning and help reduce the reality gap. In this paper, we propose an immunology-inspired embodied action-evolution cum selection algorithm that can cater to distributed ER. This algorithm evolves different subcontrollers for different portions of the search space in a distributed manner just as antibodies are evolved and primed for different antigens in the antigenic space. Experimentation on a collective of real robots embodied with the algorithm showed that a repertoire of antibody-like subcontrollers was created, evolved and shared on-the-fly to cope up with different environmental conditions. In addition, instead of the conventionally used approach of broadcasting for sharing, we present an Intelligent Packet Migration scheme that reduces energy consumption.
Evolution has produced an astonishing diversity of species, each filling a different niche. Algorithms like MAP-Elites mimic this divergent evolutionary process to find a set of behaviorally diverse but high-performing solutions, called the elites. Our key insight is that species in nature often share a surprisingly large part of their genome, in spite of occupying very different niches; similarly the elites are likely to be concentrated in a specific "elite hypervolume" whose shape is defined by their common features. In this paper, we first introduce the elite hypervolume concept and propose two metrics to characterize it: the genotypic spread and the genotypic similarity. We then introduce a new variation operator, called "directional variation", that exploits interspecies (or inter-elites) correlations to accelerate the MAP-Elites algorithm. We demonstrate the effectiveness of this operator in three problems (a toy function, a redundant robotic arm, and a hexapod robot).
Soft robotics is a growing field of research and one of its challenges is how to efficiently design a controller for a soft morphology. This paper presents a marine soft robot inspired by the ghost knifefish that swims on the water surface by using an undulating fin underneath its body. We investigate how propagating wave functions can be evolved and how these affect the swimming performance of the robot. The fin and body of the robot are constructed from silicone and six wooden fin rays actuated by servo motors. In order to bypass the reality gap, which would necessitate a complex simulation of the fish, we implemented a Covariance Matrix Adaptation Evolution Strategy (CMA-ES) directly on the physical robot to optimize its controller for travel speed. Our results show that evolving a simple sine wave or a Fourier series can generate controllers that outperform a hand programmed controller. The results additionally demonstrate that the best evolved controllers share similarities with the undulation patterns of actual knifefish. Based on these results we suggest that evolution on physical robots is promising for future application in optimizing behaviors of soft robots.
A robot to shape your natural plant: the machine learning approach to model and control bio-hybrid systems
Bio-hybrid systems-close couplings of natural organisms with technology-are high potential and still underexplored. In existing work, robots have mostly influenced group behaviors of animals. We explore the possibilities of mixing robots with natural plants, merging useful attributes. Significant synergies arise by combining the plants' ability to efficiently produce shaped material and the robots' ability to extend sensing and decision-making behaviors. However, programming robots to control plant motion and shape requires good knowledge of complex plant behaviors. Therefore, we use machine learning to create a holistic plant model and evolve robot controllers. As a benchmark task we choose obstacle avoidance. We use computer vision to construct a model of plant stem stiffening and motion dynamics by training an LSTM network. The LSTM network acts as a forward model predicting change in the plant, driving the evolution of neural network robot controllers. The evolved controllers augment the plants' natural light-finding and tissue-stiffening behaviors to avoid obstacles and grow desired shapes. We successfully verify the robot controllers and bio-hybrid behavior in reality, with a physical setup and actual plants.
The Evolutionary Robotics process requires the evaluation of large numbers of robot controllers in order to determine their relative fitnesses. The evaluation of many controllers is typically performed in simulation instead of real-world hardware in order to speed up the evolutionary process and avoid damage to robot hardware. Physics-based simulators are traditionally used in robotics for evaluating controllers. Effective traditional simulators may require a high level of accuracy and their creation requires specialised knowledge of the dynamics of the robotic system. Alternatively Artificial Neural Network based simulators are relatively simple to construct, are highly accurate, efficient and assume little specialised knowledge of the dynamics involved.
This paper proposes a novel controller evolution method that integrates uncertainty information from an ensemble of Artificial Neural Network based simulators. By estimating the accuracy of simulator predictions during the controller evolution process, robot behaviours that are not accurately simulated are likely to be avoided. A performance comparison of a single versus ensembles configuration of Artificial Neural Network based simulators is investigated. The ensemble approach developed in this work has so far outperformed all prior controller evolution methods investigated for the presented problem.