Josh Bongard

Caitlin Grasso, Josh Bongard

Information-theoretic fitness functions are becoming increasingly popular to produce generally useful, task-independent behaviors. One such universal function, dubbed empowerment, measures the amount of control an agent exerts on its environment via its sensorimotor system. Specifically, empowerment attempts to maximize the mutual information between an agent's actions and its received sensor states at a later point in time. Traditionally, empowerment has been applied to a conventional sensorimotor apparatus, such as a robot. Here, we expand the approach to a distributed, multi-agent sensorimotor system embodied by a neural cellular automaton (NCA). We show that the addition of empowerment as a secondary objective in the evolution of NCA to perform the task of morphogenesis, growing and maintaining a pre-specified shape, results in higher fitness compared to evolving for morphogenesis alone. Results suggest there may be a synergistic relationship between morphogenesis and empowerment. That is, indirectly selecting for coordination between neighboring cells over the duration of development is beneficial to the developmental process itself. Such a finding may have applications in developmental biology by providing potential mechanisms of communication between cells during growth from a single cell to a multicellular, target morphology. Source code for the experiments in this paper can be found at: \url{https://github.com/caitlingrasso/empowered-nca}.

Q. Tyrell Davis, Josh Bongard

Substantial efforts have been applied to engineer CA with desired emergent properties, such as supporting gliders. Recent work in continuous CA has generated a wide variety of compelling bioreminiscent patterns, and the expansion of CA research into continuously-valued domains, multiple channels, and higher dimensions complicates their study. In this work we devise a strategy for evolving CA and CA patterns in two steps, based on the simple idea that CA are likely to be complex and computationally capable if they support patterns that grow indefinitely as well as patterns that vanish completely, and are difficult to predict the difference in advance. The second part of our strategy evolves patterns by selecting for mobility and conservation of mean cell value. We validate our pattern evolution method by re-discovering gliders in 17 of 17 Lenia CA, and also report 4 new evolved CA and 1 randomly evolved CA that support novel evolved glider patterns. The CA reported here share neighborhood kernels with previously described Lenia CA, but exhibit a wider range of typical dynamics than their Lenia counterparts. Code for evolving continuous CA is made available under an MIT License (https://github.com/rivesunder/yuca).

David Matthews, Andrew Spielberg, Daniela Rus et al.

Robots are notoriously difficult to design because of complex interdependencies between their physical structure, sensory and motor layouts, and behavior. Despite this, almost every detail of every robot built to date has been manually determined by a human designer after several months or years of iterative ideation, prototyping, and testing. Inspired by evolutionary design in nature, the automated design of robots using evolutionary algorithms has been attempted for two decades, but it too remains inefficient: days of supercomputing are required to design robots in simulation that, when manufactured, exhibit desired behavior. Here we show for the first time de-novo optimization of a robot's structure to exhibit a desired behavior, within seconds on a single consumer-grade computer, and the manufactured robot's retention of that behavior. Unlike other gradient-based robot design methods, this algorithm does not presuppose any particular anatomical form; starting instead from a randomly-generated apodous body plan, it consistently discovers legged locomotion, the most efficient known form of terrestrial movement. If combined with automated fabrication and scaled up to more challenging tasks, this advance promises near instantaneous design, manufacture, and deployment of unique and useful machines for medical, environmental, vehicular, and space-based tasks.

Thomas F. Varley, Josh Bongard

The study of complex systems has produced a huge library of different descriptive statistics that scientists can use to describe the various emergent patterns that characterize complex systems. The problem of engineering systems to display those patterns from first principles is a much harder one, however, as a hallmark of complexity is that macro-scale emergent properties are often difficult to predict from micro-scale features. Here, we propose a general optimization-based pipeline to automate the difficult problem of engineering emergent features by re-purposing descriptive statistics as loss functions, and letting a gradient descent optimizer do the hard work of designing the relevant micro-scale features and interactions. Using Kuramoto systems of coupled oscillators as a test bed, we show that our approach can reliably produce systems with non-trivial global properties, including higher-order synergistic information, multi-attractor metastability, and meso-scale structures such as modules and integrated information. We further show that this pipeline can also account for and accommodate constraints on the system properties, such as the costs of connections, or topological restrictions. This work is a step forward on the path moving complex systems science from a field predicated largely on description and post-hoc storytelling towards one capable of engineering real-world systems with desirable emergent meso-scale and macro-scale properties.

Caitlin Grasso, Josh Bongard

Empowerment -- a domain independent, information-theoretic metric -- has previously been shown to assist in the evolutionary search for neural cellular automata (NCA) capable of homeostasis when employed as a fitness function. In our previous study, we successfully extended empowerment, defined as maximum time-lagged mutual information between agents' actions and future sensations, to a distributed sensorimotor system embodied as an NCA. However, the time-delay between actions and their corresponding sensations was arbitrarily chosen. Here, we expand upon previous work by exploring how the time scale at which empowerment operates impacts its efficacy as an auxiliary objective to accelerate the discovery of homeostatic NCAs. We show that shorter time delays result in marked improvements over empowerment with longer delays, when compared to evolutionary selection only for homeostasis. Moreover, we evaluate stability and adaptability of evolved NCAs, both hallmarks of living systems that are of interest to replicate in artificial ones. We find that short-term empowered NCA are more stable and are capable of generalizing better to unseen homeostatic challenges. Taken together, these findings motivate the use of empowerment during the evolution of other artifacts, and suggest how it should be incorporated to accelerate evolution of desired behaviors for them. Source code for the experiments in this paper can be found at: https://github.com/caitlingrasso/empowered-nca-II.

Sam Kriegman, Amir Mohammadi Nasab, Dylan Shah et al.

The manual design of soft robots and their controllers is notoriously challenging, but it could be augmented---or, in some cases, entirely replaced---by automated design tools. Machine learning algorithms can automatically propose, test, and refine designs in simulation, and the most promising ones can then be manufactured in reality (sim2real). However, it is currently not known how to guarantee that behavior generated in simulation can be preserved when deployed in reality. Although many previous studies have devised training protocols that facilitate sim2real transfer of control polices, little to no work has investigated the simulation-reality gap as a function of morphology. This is due in part to an overall lack of tools capable of systematically designing and rapidly manufacturing robots. Here we introduce a low cost, open source, and modular soft robot design and construction kit, and use it to simulate, fabricate, and measure the simulation-reality gap of minimally complex yet soft, locomoting machines. We prove the scalability of this approach by transferring an order of magnitude more robot designs from simulation to reality than any other method. The kit and its instructions can be found here: https://github.com/skriegman/sim2real4designs

Nam H. Le, Patrick Erikson, Yanbo Zhang et al.

Guiding biological systems toward desired states, such as morphogenetic outcomes, remains a fundamental challenge with far-reaching implications for medicine and synthetic biology. While large language models (LLMs) have enabled natural language as an interface for interpretable control in AI systems, their use as mediators for steering biological or cellular dynamics remains largely unexplored. In this work, we present a functional pipeline that translates natural language prompts into spatial vector fields capable of directing simulated cellular collectives. Our approach combines a large language model with an evolvable neural controller (Prompt-to-Intervention, or P2I), optimized via evolutionary strategies to generate behaviors such as clustering or scattering in a simulated 2D environment. We demonstrate that even with constrained vocabulary and simplified cell models, evolved P2I networks can successfully align cellular dynamics with user-defined goals expressed in plain language. This work offers a complete loop from language input to simulated bioelectric-like intervention to behavioral output, providing a foundation for future systems capable of natural language-driven cellular control.

Atoosa Parsa, Corey S. O'Hern, Rebecca Kramer-Bottiglio et al.

There is growing interest in engineering unconventional computing devices that leverage the intrinsic dynamics of physical substrates to perform fast and energy-efficient computations. Granular metamaterials are one such substrate that has emerged as a promising platform for building wave-based information processing devices with the potential to integrate sensing, actuation, and computation. Their high-dimensional and nonlinear dynamics result in nontrivial and sometimes counter-intuitive wave responses that can be shaped by the material properties, geometry, and configuration of individual grains. Such highly tunable rich dynamics can be utilized for mechanical computing in special-purpose applications. However, there are currently no general frameworks for the inverse design of large-scale granular materials. Here, we build upon the similarity between the spatiotemporal dynamics of wave propagation in material and the computational dynamics of Recurrent Neural Networks to develop a gradient-based optimization framework for harmonically driven granular crystals. We showcase how our framework can be utilized to design basic logic gates where mechanical vibrations carry the information at predetermined frequencies. We compare our design methodology with classic gradient-free methods and find that our approach discovers higher-performing configurations with less computational effort. Our findings show that a gradient-based optimization method can greatly expand the design space of metamaterials and provide the opportunity to systematically traverse the parameter space to find materials with the desired functionalities.

James Bagrow, Josh Bongard

Efforts to improve Kolmogorov-Arnold networks (KANs) with architectural enhancements have been stymied by the complexity those enhancements bring, undermining the interpretability that makes KANs attractive in the first place. Here we study overprovisioned architectures combined with sparsification to learn compact, interpretable KANs without sacrificing accuracy. Crucially, we focus on differentiable sparsification, turning architecture search into an end-to-end optimization problem. Across function approximation benchmarks, dynamical systems forecasting, and real-world prediction tasks, we demonstrate competitive or superior accuracy while discovering substantially smaller models. Overprovisioning and sparsification are synergistic, with the combination outperforming either alone. The result is a principled path toward models that are both more expressive and more interpretable, addressing a key tension in scientific machine learning.

James Bagrow, Josh Bongard

Kolmogorov-Arnold Networks (KANs) offer a promising path toward interpretable machine learning: their learnable activations can be studied individually, while collectively fitting complex data accurately. In practice, however, trained activations often lack symbolic fidelity, learning pathological decompositions with no meaningful correspondence to interpretable forms. We propose Softly Symbolified Kolmogorov-Arnold Networks (S2KAN), which integrate symbolic primitives directly into training. Each activation draws from a dictionary of symbolic and dense terms, with learnable gates that sparsify the representation. Crucially, this sparsification is differentiable, enabling end-to-end optimization, and is guided by a principled Minimum Description Length objective. When symbolic terms suffice, S2KAN discovers interpretable forms; when they do not, it gracefully degrades to dense splines. We demonstrate competitive or superior accuracy with substantially smaller models across symbolic benchmarks, dynamical systems forecasting, and real-world prediction tasks, and observe evidence of emergent self-sparsification even without regularization pressure.

Nam H. Le, Patrick Erickson, Yanbo Zhang et al.

Human language is one of the most expressive tools for conveying intent, yet most artificial or biological systems lack mechanisms to interpret or respond meaningfully to it. Bridging this gap could enable more natural forms of control over complex, decentralized systems. In AI and artificial life, recent work explores how language can specify high-level goals, but most systems still depend on engineered rewards, task-specific supervision, or rigid command sets, limiting generalization to novel instructions. Similar constraints apply in synthetic biology and bioengineering, where the locus of control is often genomic rather than environmental perturbation. A key open question is whether artificial or biological collectives can be guided by free-form natural language alone, without task-specific tuning or carefully designed evaluation metrics. We provide one possible answer here by showing, for the first time, that simple agents' collective behavior can be guided by free-form language prompts: one AI model transforms an imperative prompt into an intervention that is applied to simulated cells; a second AI model scores how well the prompt describes the resulting cellular dynamics; and the former AI model is evolved to improve the scores generated by the latter. Unlike previous work, our method does not require engineered fitness functions or domain-specific prompt design. We show that the evolved system generalizes to unseen prompts without retraining. By treating natural language as a control layer, the system suggests a future in which spoken or written prompts could direct computational, robotic, or biological systems to desired behaviors. This work provides a concrete step toward this vision of AI-biology partnerships, in which language replaces mathematical objective functions, fixed rules, and domain-specific programming.

James Bagrow, Josh Bongard

Kolmogorov-Arnold Networks (KANs) uniquely combine high accuracy with interpretability, making them valuable for scientific modeling. However, it is unclear a priori how deep a network needs to be for any given task, and deeper KANs can be difficult to optimize and interpret. Here we introduce multi-exit KANs, where each layer includes its own prediction branch, enabling the network to make accurate predictions at multiple depths simultaneously. This architecture provides deep supervision that improves training while discovering the right level of model complexity for each task. Multi-exit KANs consistently outperform standard, single-exit versions on synthetic functions, dynamical systems, and real-world datasets. Remarkably, the best predictions often come from earlier, simpler exits, revealing that these networks naturally identify smaller, more parsimonious and interpretable models without sacrificing accuracy. To automate this discovery, we develop a differentiable "learning-to-exit" algorithm that balances contributions from exits during training. Our approach offers scientists a practical way to achieve both high performance and interpretability, addressing a fundamental challenge in machine learning for scientific discovery.

Nam H. Le, Josh Bongard

Genetic Programming (GP) has traditionally entangled the evolution of symbolic representations with their performance-based evaluation, often relying solely on raw fitness scores. This tight coupling makes GP solutions more fragile and prone to overfitting, reducing their ability to generalize. In this work, we propose LaSER (Latent Semantic Representation Regression)} -- a general framework that decouples representation evolution from lifetime learning. At each generation, candidate programs produce features which are passed to an external learner to model the target task. This approach enables any function approximator, from linear models to neural networks, to serve as a lifetime learner, allowing expressive modeling beyond conventional symbolic forms. Here we show for the first time that LaSER can outcompete standard GP and GP followed by linear regression when it employs non-linear methods to fit coefficients to GP-generated equations against complex data sets. Further, we explore how LaSER enables the emergence of innate representations, supporting long-standing hypotheses in evolutionary learning such as the Baldwin Effect. By separating the roles of representation and adaptation, LaSER offers a principled and extensible framework for symbolic regression and classification.

Atoosa Parsa, James Bagrow, Corey S. O'Hern et al.

Externally driven dense packings of particles can exhibit nonlinear wave phenomena that are not described by effective medium theory or linearized approximate models. Such nontrivial wave responses can be exploited to design sound-focusing/scrambling devices, acoustic filters, and analog computational units. At high amplitude vibrations or low confinement pressures, the effect of nonlinear particle contacts becomes increasingly noticeable, and the interplay of nonlinearity, disorder, and discreteness in the system gives rise to remarkable properties, particularly useful in designing structures with exotic properties. In this paper, we build upon the data-driven methods in dynamical system analysis and show that the Koopman spectral theory can be applied to granular crystals, enabling their phase space analysis beyond the linearizable regime and without recourse to any approximations considered in the previous works. We show that a deep neural network can map the dynamics to a latent space where the essential nonlinearity of the granular system unfolds into a high-dimensional linear space. As a proof of concept, we use data from numerical simulations of a two-particle system and evaluate the accuracy of the trajectory predictions under various initial conditions. By incorporating data from experimental measurements, our proposed framework can directly capture the underlying dynamics without imposing any assumptions about the physics model. Spectral analysis of the trained surrogate system can help bridge the gap between the simulation results and the physical realization of granular crystals and facilitate the inverse design of materials with desired behaviors.

Atoosa Parsa, Sven Witthaus, Nidhi Pashine et al.

Unconventional computing devices are increasingly of interest as they can operate in environments hostile to silicon-based electronics, or compute in ways that traditional electronics cannot. Mechanical computers, wherein information processing is a material property emerging from the interaction of components with the environment, are one such class of devices. This information processing can be manifested in various physical substrates, one of which is granular matter. In a granular assembly, vibration can be treated as the information-bearing mode. This can be exploited to realize "polycomputing": materials can be evolved such that a single grain within them can report the result of multiple logical operations simultaneously at different frequencies, without recourse to quantum effects. Here, we demonstrate the evolution of a material in which one grain acts simultaneously as two different NAND gates at two different frequencies. NAND gates are of interest as any logical operations can be built from them. Moreover, they are nonlinear thus demonstrating a step toward general-purpose, computationally dense mechanical computers. Polycomputation was found to be distributed across each evolved material, suggesting the material's robustness. With recent advances in material sciences, hardware realization of these materials may eventually provide devices that challenge the computational density of traditional computers.

Joshua Powers, Ryan Grindle, Lapo Frati et al.

In robotics, catastrophic interference continues to restrain policy training across environments. Efforts to combat catastrophic interference to date focus on novel neural architectures or training methods, with a recent emphasis on policies with good initial settings that facilitate training in new environments. However, none of these methods to date have taken into account how the physical architecture of the robot can obstruct or facilitate catastrophic interference, just as the choice of neural architecture can. In previous work we have shown how aspects of a robot's physical structure (specifically, sensor placement) can facilitate policy learning by increasing the fraction of optimal policies for a given physical structure. Here we show for the first time that this proxy measure of catastrophic interference correlates with sample efficiency across several search methods, proving that favorable loss landscapes can be induced by the correct choice of physical structure. We show that such structures can be found via co-optimization -- optimization of a robot's structure and control policy simultaneously -- yielding catastrophic interference resistant robot structures and policies, and that this is more efficient than control policy optimization alone. Finally, we show that such structures exhibit sensor homeostasis across environments and introduce this as the mechanism by which certain robots overcome catastrophic interference.

Sam Kriegman, Amir Mohammadi Nasab, Douglas Blackiston et al.

Robots deployed at orders of magnitude different size scales, and that retain the same desired behavior at any of those scales, would greatly expand the environments in which the robots could operate. However it is currently not known whether such robots exist, and, if they do, how to design them. Since self similar structures in nature often exhibit self similar behavior at different scales, we hypothesize that there may exist robot designs that have the same property. Here we demonstrate that this is indeed the case for some, but not all, modular soft robots: there are robot designs that exhibit a desired behavior at a small size scale, and if copies of that robot are attached together to realize the same design at higher scales, those larger robots exhibit similar behavior. We show how to find such designs in simulation using an evolutionary algorithm. Further, when fractal attachment is not assumed and attachment geometries must thus be evolved along with the design of the base robot unit, scale invariant behavior is not achieved, demonstrating that structural self similarity, when combined with appropriate designs, is a useful path to realizing scale invariant robot behavior. We validate our findings by demonstrating successful transferal of self similar structure and behavior to pneumatically-controlled soft robots. Finally, we show that biobots can spontaneously exhibit self similar attachment geometries, thereby suggesting that self similar behavior via self similar structure may be realizable across a wide range of robot platforms in future.

Dylan S. Shah, Joshua P. Powers, Liana G. Tilton et al.

Many organisms, including various species of spiders and caterpillars, change their shape to switch gaits and adapt to different environments. Recent technological advances, ranging from stretchable circuits to highly deformable soft robots, have begun to make shape-changing robots a possibility. However, it is currently unclear how and when shape change should occur, and what capabilities could be gained, leading to a wide range of unsolved design and control problems. To begin addressing these questions, here we simulate, design, and build a soft robot that utilizes shape change to achieve locomotion over both a flat and inclined surface. Modeling this robot in simulation, we explore its capabilities in two environments and demonstrate the existence of environment-specific shapes and gaits that successfully transfer to the physical hardware. We found that the shape-changing robot traverses these environments better than an equivalent but non-morphing robot, in simulation and reality.

Kent Rosser, Jia Kok, Javaan Chahl et al.

Morphology of a robot design is important to its ability to achieve a stated goal and therefore applying machine learning approaches that incorporate morphology in the design space can provide scope for significant advantage. Our study is set in a domain known to be reliant on morphology: flapping wing flight. We developed a parameterised morphology design space that draws features from biological exemplars and apply automated design to produce a set of high performance robot morphologies in simulation. By performing sim2real transfer on a selection, for the first time we measure the shape of the reality gap for variations in design complexity. We found for the flapping wing that the reality gap changes non-monotonically with complexity, suggesting that certain morphology details narrow the gap more than others, and that such details could be identified and further optimised in a future end-to-end automated morphology design process.

Joshua Powers, Ryan Grindle, Sam Kriegman et al.

Catastrophic forgetting continues to severely restrict the learnability of controllers suitable for multiple task environments. Efforts to combat catastrophic forgetting reported in the literature to date have focused on how control systems can be updated more rapidly, hastening their adjustment from good initial settings to new environments, or more circumspectly, suppressing their ability to overfit to any one environment. When using robots, the environment includes the robot's own body, its shape and material properties, and how its actuators and sensors are distributed along its mechanical structure. Here we demonstrate for the first time how one such design decision (sensor placement) can alter the landscape of the loss function itself, either expanding or shrinking the weight manifolds containing suitable controllers for each individual task, thus increasing or decreasing their probability of overlap across tasks, and thus reducing or inducing the potential for catastrophic forgetting.

David Matthews, Sam Kriegman, Collin Cappelle et al.

Enabling machines to respond appropriately to natural language commands could greatly expand the number of people to whom they could be of service. Recently, advances in neural network-trained word embeddings have empowered non-embodied text-processing algorithms, and suggest they could be of similar utility for embodied machines. Here we introduce a method that does so by training robots to act similarly to semantically-similar word2vec encoded commands. We show that this enables them to act appropriately, after training, to previously-unheard commands. Finally, we show that inducing such an alignment between motoric and linguistic similarities can be facilitated or hindered by the mechanical structure of the robot. This points to future, large scale methods that find and exploit relationships between action, language, and robot structure.

Sam Kriegman, Stephanie Walker, Dylan Shah et al.

A robot's mechanical parts routinely wear out from normal functioning and can be lost to injury. For autonomous robots operating in isolated or hostile environments, repair from a human operator is often not possible. Thus, much work has sought to automate damage recovery in robots. However, every case reported in the literature to date has accepted the damaged mechanical structure as fixed, and focused on learning new ways to control it. Here we show for the first time a robot that automatically recovers from unexpected damage by deforming its resting mechanical structure without changing its control policy. We found that, especially in the case of "deep insult", such as removal of all four of the robot's legs, the damaged machine evolves shape changes that not only recover the original level of function (locomotion) as before, but can in fact surpass the original level of performance (speed). This suggests that shape change, instead of control readaptation, may be a better method to recover function after damage in some cases.

Zahra Mahoor, Jack Felag, Josh Bongard

As more robots act in physical proximity to people, it is essential to ensure they make decisions and execute actions that align with human values. To do so, robots need to understand the true intentions behind human-issued commands. In this paper, we define a safe robot as one that receives a natural-language command from humans, considers an action in response to that command, and accurately predicts how humans will judge that action if is executed in reality. Our contribution is two-fold: First, we introduce a web platform for human users to propose commands to simulated robots. The robots receive commands and act based on those proposed commands, and then the users provide positive and/or negative reinforcement. Next, we train a critic for each robot to predict the crowd's responses to one of the crowd-proposed commands. Second, we show that the morphology of a robot plays a role in the way it grounds language: The critics show that two of the robots used in the experiment achieve a lower prediction error than the others. Thus, those two robots are safer, according to our definition, since they ground the proposed command more accurately.

Sam Kriegman, Nick Cheney, Josh Bongard

Organisms result from adaptive processes interacting across different time scales. One such interaction is that between development and evolution. Models have shown that development sweeps over several traits in a single agent, sometimes exposing promising static traits. Subsequent evolution can then canalize these rare traits. Thus, development can, under the right conditions, increase evolvability. Here, we report on a previously unknown phenomenon when embodied agents are allowed to develop and evolve: Evolution discovers body plans robust to control changes, these body plans become genetically assimilated, yet controllers for these agents are not assimilated. This allows evolution to continue climbing fitness gradients by tinkering with the developmental programs for controllers within these permissive body plans. This exposes a previously unknown detail about the Baldwin effect: instead of all useful traits becoming genetically assimilated, only traits that render the agent robust to changes in other traits become assimilated. We refer to this as differential canalization. This finding also has implications for the evolutionary design of artificial and embodied agents such as robots: robots robust to internal changes in their controllers may also be robust to external changes in their environment, such as transferal from simulation to reality or deployment in novel environments.

Francesco Corucci, Nick Cheney, Francesco Giorgio-Serchi et al.

Designing soft robots poses considerable challenges: automated design approaches may be particularly appealing in this field, as they promise to optimize complex multi-material machines with very little or no human intervention. Evolutionary soft robotics is concerned with the application of optimization algorithms inspired by natural evolution in order to let soft robots (both morphologies and controllers) spontaneously evolve within physically-realistic simulated environments, figuring out how to satisfy a set of objectives defined by human designers. In this paper a powerful evolutionary system is put in place in order to perform a broad investigation on the free-form evolution of walking and swimming soft robots in different environments. Three sets of experiments are reported, tackling different aspects of the evolution of soft locomotion. The first two sets explore the effects of different material properties on the evolution of terrestrial and aquatic soft locomotion: particularly, we show how different materials lead to the evolution of different morphologies, behaviors, and energy-performance tradeoffs. It is found that within our simplified physics world stiffer robots evolve more sophisticated and effective gaits and morphologies on land, while softer ones tend to perform better in water. The third set of experiments starts investigating the effect and potential benefits of major environmental transitions (land - water) during evolution. Results provide interesting morphological exaptation phenomena, and point out a potential asymmetry between land-water and water-land transitions: while the first type of transition appears to be detrimental, the second one seems to have some beneficial effects.

Nick Cheney, Josh Bongard, Vytas SunSpiral et al.

Evolution sculpts both the body plans and nervous systems of agents together over time. In contrast, in AI and robotics, a robot's body plan is usually designed by hand, and control policies are then optimized for that fixed design. The task of simultaneously co-optimizing the morphology and controller of an embodied robot has remained a challenge. In psychology, the theory of embodied cognition posits that behavior arises from a close coupling between body plan and sensorimotor control, which suggests why co-optimizing these two subsystems is so difficult: most evolutionary changes to morphology tend to adversely impact sensorimotor control, leading to an overall decrease in behavioral performance. Here, we further examine this hypothesis and demonstrate a technique for "morphological innovation protection", which temporarily reduces selection pressure on recently morphologically-changed individuals, thus enabling evolution some time to "readapt" to the new morphology with subsequent control policy mutations. We show the potential for this method to avoid local optima and converge to similar highly fit morphologies across widely varying initial conditions, while sustaining fitness improvements further into optimization. While this technique is admittedly only the first of many steps that must be taken to achieve scalable optimization of embodied machines, we hope that theoretical insight into the cause of evolutionary stagnation in current methods will help to enable the automation of robot design and behavioral training -- while simultaneously providing a testbed to investigate the theory of embodied cognition.