Particle robotics based on statistical mechanics of loosely coupled components.

Biological organisms achieve robust high-level behaviours by combining and coordinating stochastic low-level components1-3. By contrast, most current robotic systems comprise either monolithic mechanisms4,5 or modular units with coordinated motions6,7. Such robots require explicit control of individual components to perform specific functions, and the failure of one component typically renders the entire robot inoperable. Here we demonstrate a robotic system whose overall behaviour can be successfully controlled by exploiting statistical mechanics phenomena. We achieve this by incorporating many loosely coupled 'particles', which are incapable of independent locomotion and do not possess individual identity or addressable position. In the proposed system, each particle is permitted to perform only uniform volumetric oscillations that are phase-modulated by a global signal. Despite the stochastic motion of the robot and lack of direct control of its individual components, we demonstrate physical robots composed of up to two dozen particles and simulated robots with up to 100,000 particles capable of robust locomotion, object transport and phototaxis (movement towards a light stimulus). Locomotion is maintained even when 20 per cent of the particles malfunction. These findings indicate that stochastic systems may offer an alternative approach to more complex and exacting robots via large-scale robust amorphous robotic systems that exhibit deterministic behaviour.

The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communities.

Evolution provides a creative fount of complex and subtle adaptations that often surprise the scientists who discover them. However, the creativity of evolution is not limited to the natural world: Artificial organisms evolving in computational environments have also elicited surprise and wonder from the researchers studying them. The process of evolution is an algorithmic process that transcends the substrate in which it occurs. Indeed, many researchers in the field of digital evolution can provide examples of how their evolving algorithms and organisms have creatively subverted their expectations or intentions, exposed unrecognized bugs in their code, produced unexpectedly adaptations, or engaged in behaviors and outcomes, uncannily convergent with ones found in nature. Such stories routinely reveal surprise and creativity by evolution in these digital worlds, but they rarely fit into the standard scientific narrative. Instead they are often treated as mere obstacles to be overcome, rather than results that warrant study in their own right. Bugs are fixed, experiments are refocused, and one-off surprises are collapsed into a single data point. The stories themselves are traded among researchers through oral tradition, but that mode of information transmission is inefficient and prone to error and outright loss. Moreover, the fact that these stories tend to be shared only among practitioners means that many natural scientists do not realize how interesting and lifelike digital organisms are and how natural their evolution can be. To our knowledge, no collection of such anecdotes has been published before. This article is the crowd-sourced product of researchers in the fields of artificial life and evolutionary computation who have provided first-hand accounts of such cases. It thus serves as a written, fact-checked collection of scientifically important and even entertaining stories. In doing so we also present here substantial evidence that the existence and importance of evolutionary surprises extends beyond the natural world, and may indeed be a universal property of all complex evolving systems.

Automated Identification of Northern Leaf Blight-Infected Maize Plants from Field Imagery Using Deep Learning.

Northern leaf blight (NLB) can cause severe yield loss in maize; however, scouting large areas to accurately diagnose the disease is time consuming and difficult. We demonstrate a system capable of automatically identifying NLB lesions in field-acquired images of maize plants with high reliability. This approach uses a computational pipeline of convolutional neural networks (CNNs) that addresses the challenges of limited data and the myriad irregularities that appear in images of field-grown plants. Several CNNs were trained to classify small regions of images as containing NLB lesions or not; their predictions were combined into separate heat maps, then fed into a final CNN trained to classify the entire image as containing diseased plants or not. The system achieved 96.7% accuracy on test set images not used in training. We suggest that such systems mounted on aerial- or ground-based vehicles can help in automated high-throughput plant phenotyping, precision breeding for disease resistance, and reduced pesticide use through targeted application across a variety of plant and disease categories.

Inverse Gillespie for inferring stochastic reaction mechanisms from intermittent samples.

Gillespie stochastic simulation is used extensively to investigate stochastic phenomena in many fields, ranging from chemistry to biology to ecology. The inverse problem, however, has remained largely unsolved: How to reconstruct the underlying reactions de novo from sparse observations. A key challenge is that often only aggregate concentrations, proportional to the population numbers, are observable intermittently. We discovered that under specific assumptions, the set of relative population updates in phase space forms a convex polytope whose vertices are indicative of the dominant underlying reactions. We demonstrate the validity of this simple principle by reconstructing stochastic models (reaction structure plus propensities) from a variety of simulated and experimental systems, where hundreds and even thousands of reactions may be occurring in between observations. In some cases, the inferred models provide mechanistic insight. This principle can lead to the understanding of a broad range of phenomena, from molecular biology to population ecology.

Particle shape effects on the stress response of granular packings.

We present measurements of the stress response of packings formed from a wide range of particle shapes. Besides spheres these include convex shapes such as the Platonic solids, truncated tetrahedra, and triangular bipyramids, as well as more complex, non-convex geometries such as hexapods with various arm lengths, dolos, and tetrahedral frames. All particles were 3D-printed in hard resin. Well-defined initial packing states were established through preconditioning by cyclic loading under given confinement pressure. Starting from such initial states, stress-strain relationships for axial compression were obtained at four different confining pressures for each particle type. While confining pressure has the largest overall effect on the mechanical response, we find that particle shape controls the details of the stress-strain curves and can be used to tune packing stiffness and yielding. By correlating the experimentally measured values for the effective Young's modulus under compression, yield stress and energy loss during cyclic loading, we identify trends among the various shapes that allow for designing a packing's aggregate behavior.

Unveiling intra-person fingerprint similarity via deep contrastive learning.

Fingerprint biometrics are integral to digital authentication and forensic science. However, they are based on the unproven assumption that no two fingerprints, even from different fingers of the same person, are alike. This renders them useless in scenarios where the presented fingerprints are from different fingers than those on record. Contrary to this prevailing assumption, we show above 99.99% confidence that fingerprints from different fingers of the same person share very strong similarities. Using deep twin neural networks to extract fingerprint representation vectors, we find that these similarities hold across all pairs of fingers within the same person, even when controlling for spurious factors like sensor modality. We also find evidence that ridge orientation, especially near the fingerprint center, explains a substantial part of this similarity, whereas minutiae used in traditional methods are almost nonpredictive. Our experiments suggest that, in some situations, this relationship can increase forensic investigation efficiency by almost two orders of magnitude.

Human-robot facial coexpression.

Large language models are enabling rapid progress in robotic verbal communication, but nonverbal communication is not keeping pace. Physical humanoid robots struggle to express and communicate using facial movement, relying primarily on voice. The challenge is twofold: First, the actuation of an expressively versatile robotic face is mechanically challenging. A second challenge is knowing what expression to generate so that the robot appears natural, timely, and genuine. Here, we propose that both barriers can be alleviated by training a robot to anticipate future facial expressions and execute them simultaneously with a human. Whereas delayed facial mimicry looks disingenuous, facial coexpression feels more genuine because it requires correct inference of the human's emotional state for timely execution. We found that a robot can learn to predict a forthcoming smile about 839 milliseconds before the human smiles and, using a learned inverse kinematic facial self-model, coexpress the smile simultaneously with the human. We demonstrated this ability using a robot face comprising 26 degrees of freedom. We believe that the ability to coexpress simultaneous facial expressions could improve human-robot interaction.

The evolutionary origins of modularity.

A central biological question is how natural organisms are so evolvable (capable of quickly adapting to new environments). A key driver of evolvability is the widespread modularity of biological networks--their organization as functional, sparsely connected subunits--but there is no consensus regarding why modularity itself evolved. Although most hypotheses assume indirect selection for evolvability, here we demonstrate that the ubiquitous, direct selection pressure to reduce the cost of connections between network nodes causes the emergence of modular networks. Computational evolution experiments with selection pressures to maximize network performance and minimize connection costs yield networks that are significantly more modular and more evolvable than control experiments that only select for performance. These results will catalyse research in numerous disciplines, such as neuroscience and genetics, and enhance our ability to harness evolution for engineering purposes.

Functional inference of complex anatomical tendinous networks at a macroscopic scale via sparse experimentation.

In systems and computational biology, much effort is devoted to functional identification of systems and networks at the molecular-or cellular scale. However, similarly important networks exist at anatomical scales such as the tendon network of human fingers: the complex array of collagen fibers that transmits and distributes muscle forces to finger joints. This network is critical to the versatility of the human hand, and its function has been debated since at least the 16(th) century. Here, we experimentally infer the structure (both topology and parameter values) of this network through sparse interrogation with force inputs. A population of models representing this structure co-evolves in simulation with a population of informative future force inputs via the predator-prey estimation-exploration algorithm. Model fitness depends on their ability to explain experimental data, while the fitness of future force inputs depends on causing maximal functional discrepancy among current models. We validate our approach by inferring two known synthetic Latex networks, and one anatomical tendon network harvested from a cadaver's middle finger. We find that functionally similar but structurally diverse models can exist within a narrow range of the training set and cross-validation errors. For the Latex networks, models with low training set error [<4%] and resembling the known network have the smallest cross-validation errors [∼5%]. The low training set [<4%] and cross validation [<7.2%] errors for models for the cadaveric specimen demonstrate what, to our knowledge, is the first experimental inference of the functional structure of complex anatomical networks. This work expands current bioinformatics inference approaches by demonstrating that sparse, yet informative interrogation of biological specimens holds significant computational advantages in accurate and efficient inference over random testing, or assuming model topology and only inferring parameters values. These findings also hold clues to both our evolutionary history and the development of versatile machines.

Inverted laser sintering of metal powders.

We demonstrate the ability of the inverted laser sintering process to manufacture parts composed of metal powder. We fabricate a 10-layer part by depositing a layer of copper powder onto a sapphire plate, then pressing the plate against the part being built and sintering the powder onto the part by shining a 14W 445 nm laser through the glass. The process was then repeated multiple times, each time adding a new layer to the component being printed until completion. We discuss the potential applications and impacts of this process, including the ability to directly fabricate multi-material metallic parts without the use of a powder bed.

Design and fabrication of carbon fiber lattices using 3D weaving.

We present a method of designing and fabricating 3D carbon fiber lattices. The lattice design and fabrication is based on crocheting and sewing techniques, where carbon fiber tow is woven through two parallel carbon fiber grids and reinforced with vertical carbon fiber tubes. Compression testing is then performed on three different designs, and these results are compared to other similar lattice structures, finding that the lattices perform similarly to comparable lattices. Finite element analysis is also performed to validate the experimental findings, and provides some insight into the experimental results. The process presented here allows for more design flexibility than other current methods. For example, within a single lattice, different density weave patterns can be used to address specific load requirements. Though fabricated manually here, this process can also be automated for large scale production. With this design flexibility, simplified fabrication, and high strength, the lattices proposed here offer an advantage as compared to similar existing structures.

Regenerative Topology Optimization of Fine Lattice Structures.

We present a generative approach for creating three-dimensional lattice structures optimized for mass and deflection composed of thousands of one-dimensional strut primitives. Our approach draws inspiration from topology optimization principles. The proposed method iteratively determines unnecessary lattice struts through stress analysis, erodes those struts, and then randomly generates new struts across the entire structure. The objects resulting from this distributed optimization technique demonstrate high strength-to-weight ratios that are at par with state-of-the-art topology optimization approaches, but are qualitatively very different. We use a dynamics simulator that allows optimization of structures subject to dynamic load cases, such as vibrating structures and robotic components. Because optimization is performed simultaneously with simulation, the process scales efficiently on massively parallel graphics processing units. The intricate nature of the output lattices contributes to a new class of objects intended specifically for additive manufacturing. Our work contributes a highly parallel simulation method and simultaneous algorithm for analyzing and optimizing lattices with thousands of struts. In this study, we validate multiple versions of our algorithm across sample load cases, to show its potential for creating high-resolution objects with implicit optimized microstructural patterns.

The future of software-controlled cooking.

To date, analog methods of cooking such as by grills, cooktops, stoves and microwaves have remained the world's predominant cooking modalities. With the continual evolution of digital technologies, however, laser cooking and 3D food printing may present nutritious, convenient and cost-effective cooking opportunities. Food printing is an application of additive manufacturing that utilizes user-generated models to construct 3D shapes from edible food inks and laser cooking uses high-energy targeted light for high-resolution tailored heating. Using software to combine and cook ingredients allows a chef to more easily control the nutrient content of a meal, which could lead to healthier and more customized meals. With more emphasis on food safety following COVID-19, food prepared with less human handling may lower the risk of foodborne illness and disease transmission. Digital cooking technologies allow an end consumer to take more control of the macro and micro nutrients that they consume on a per meal basis and due to the rapid growth and potential benefits of 3D technology advancements, a 3D printer may become a staple home and industrial cooking device.

Fully body visual self-modeling of robot morphologies.

Internal computational models of physical bodies are fundamental to the ability of robots and animals alike to plan and control their actions. These "self-models" allow robots to consider outcomes of multiple possible future actions without trying them out in physical reality. Recent progress in fully data-driven self-modeling has enabled machines to learn their own forward kinematics directly from task-agnostic interaction data. However, forward kinematic models can only predict limited aspects of the morphology, such as the position of end effectors or velocity of joints and masses. A key challenge is to model the entire morphology and kinematics without prior knowledge of what aspects of the morphology will be relevant to future tasks. Here, we propose that instead of directly modeling forward kinematics, a more useful form of self-modeling is one that could answer space occupancy queries, conditioned on the robot's state. Such query-driven self-models are continuous in the spatial domain, memory efficient, fully differentiable, and kinematic aware and can be used across a broader range of tasks. In physical experiments, we demonstrate how a visual self-model is accurate to about 1% of the workspace, enabling the robot to perform various motion planning and control tasks. Visual self-modeling can also allow the robot to detect, localize, and recover from real-world damage, leading to improved machine resiliency.

Electro-Osmotic Gripper Characterization for Layered Assembly.

Layered assembly is a voxel-based additive manufacturing process that relies on parallel grasping of voxels to produce multi-material parts. Although there exists substantial diversity in mechanisms of gripping, there still exists a lack of consistency, accuracy, and efficacy in positioning very large numbers of milli-, micro-, and nano-scale objects. We demonstrate the use of parallel electro-osmotic grippers to selectively transport multiple millimeter-sized voxels simultaneously. In contrast to previous research focused on using arrays of droplets to grab a single substrate, each element in the array is individually controlled via capillary effects, which are, in turn, controlled by an electric field to create predetermined patterns of droplets to pick and place selected objects. The demonstrated fluidic pick-and-place method has two key advantages: It is suitable for transport of fragile and complex objects due to the lack of mechanical contact, and it easily parallelizes to arbitrary array sizes for massively parallel pick-and-place. This work demonstrates a 25-element parallel assembly of 1.5-mm spheres with 95-98% grasping reliability, in a variety of geometric patterns. Experimental performance was validated against both analytical and computational models. The results suggest that electro-osmotic droplet arrays may enable the additive manufacturing of multi-material objects containing millions of components in the same print bed.

Automated discovery of fundamental variables hidden in experimental data.

All physical laws are described as mathematical relationships between state variables. These variables give a complete and non-redundant description of the relevant system. However, despite the prevalence of computing power and artificial intelligence, the process of identifying the hidden state variables themselves has resisted automation. Most data-driven methods for modelling physical phenomena still rely on the assumption that the relevant state variables are already known. A longstanding question is whether it is possible to identify state variables from only high-dimensional observational data. Here we propose a principle for determining how many state variables an observed system is likely to have, and what these variables might be. We demonstrate the effectiveness of this approach using video recordings of a variety of physical dynamical systems, ranging from elastic double pendulums to fire flames. Without any prior knowledge of the underlying physics, our algorithm discovers the intrinsic dimension of the observed dynamics and identifies candidate sets of state variables.

What we look for at Science Robotics.

Science Robotics welcomes papers demonstrating technical and scientific advances, with potential for influence beyond robotics.

Automated refinement and inference of analytical models for metabolic networks.

The reverse engineering of metabolic networks from experimental data is traditionally a labor-intensive task requiring a priori systems knowledge. Using a proven model as a test system, we demonstrate an automated method to simplify this process by modifying an existing or related model--suggesting nonlinear terms and structural modifications--or even constructing a new model that agrees with the system's time series observations. In certain cases, this method can identify the full dynamical model from scratch without prior knowledge or structural assumptions. The algorithm selects between multiple candidate models by designing experiments to make their predictions disagree. We performed computational experiments to analyze a nonlinear seven-dimensional model of yeast glycolytic oscillations. This approach corrected mistakes reliably in both approximated and overspecified models. The method performed well to high levels of noise for most states, could identify the correct model de novo, and make better predictions than ordinary parametric regression and neural network models. We identified an invariant quantity in the model, which accurately derived kinetics and the numerical sensitivity coefficients of the system. Finally, we compared the system to dynamic flux estimation and discussed the scaling and application of this methodology to automated experiment design and control in biological systems in real time.

Robotics: self-reproducing machines.

Self-reproduction is central to biological life for long-term sustainability and evolutionary adaptation. Although these traits would also be desirable in many engineered systems, the principles of self-reproduction have not been exploited in machine design. Here we create simple machines that act as autonomous modular robots and are capable of physical self-reproduction using a set of cubes.