A Modular Soft Sensing Skin for Fast Measurement of Wing Deformation in Small Unmanned Aerial Vehicles.

Insects, bats, and small birds show outstanding flight performance even under complex atmospheric conditions, which is partially due to the ability of these natural fliers to sense and react to disturbances quickly. These biological systems often use large numbers of sensors arrayed across their bodies to detect disturbances, but previous efforts to use large arrays of sensors in engineered fliers have typically resulted in slow responses due to the need to scan and process data from the large number of sensors. To address the challenges of capturing disturbances in a large sensing array with low latency, this work proposes and demonstrates a modular soft sensing system to quickly detect disturbances in small unmanned aerial vehicles. A large array of soft strain sensors with high sensing resolution covers the entire wingspan, providing rich information on wing deformation. Owing to the modular design, decentralized computation enables the sensing system to efficiently manage sensor data, resulting in sufficiently fast sampling to capture wing dynamics while all 32 sensors embedded in the modular soft sensing skin are used. This hardware architecture also results in significantly reduced noise in the sensing system, leading to a high signal-to-noise ratio. These methods can ultimately enable fast and reliable control of both soft and rigid robotic systems using large arrays of soft sensors.

Controlling jumps through latches in small jumping robots.

Small jumping robots can use springs to maximize jump performance, but they are typically not able to control the height of each jump owing to design constraints. This study explores the use of the jumper's latch, the component that mediates the release of energy stored in the spring, as a tool for controlling jumps. A reduced-order model that considers the dynamics of the actuator pulling the latch and the effect of spring force on the latch is presented. This model is then validated using high speed video and ground reaction force measurements from a 4gjumper. Both the model and experimental results demonstrate that jump performance in small insect-inspired resource-constrained robots can be tuned to a range of outputs using latch mediation, despite starting with a fixed spring potential energy. For a fixed set of input voltages to the latch actuator, the results also show that a jumper with a larger latch radius has greater tunability. However, this greater tunability comes with a trade-off in maximum performance. Finally, we define a new metric, 'Tunability Range,' to capture the range of controllable jump behaviors that a jumper with a fixed spring compression can attain given a set of control inputs (i.e. latch actuation voltage) to choose from.

Buoyant magnetic milliswimmers reveal design rules for optimizing microswimmer performance.

Magnetically-actuated swimming microrobots are an emerging tool for navigating and manipulating materials in confined spaces. Recent work has demonstrated that it is possible to build such systems at the micro and nanoscales using polymer microspheres, magnetic particles and DNA nanotechnology. However, while these materials enable an unprecedented ability to build at small scales, such systems often demonstrate significant polydispersity resulting from both the material variations and the assembly process itself. This variability makes it difficult to predict, let alone optimize, the direction or magnitude of microswimmer velocity from design parameters such as link shape or aspect ratio. To isolate questions of a swimmer's design from variations in its physical dimensions, we present a novel experimental platform using two-photon polymerization to build a two-link, buoyant milliswimmer with a fully customizable shape and integrated flexible linker (the swimmer is underactuated, enabling asymmetric cyclic motion and net translation). Our approach enables us to control both swimming direction and repeatability of swimmer performance. These studies provide ground truth data revealing that neither the first order nor second order models currently capture the key features of milliswimmer performance. We therefore use our experimental platform to develop design guidelines for tuning the swimming speeds, and we identify the following three approaches for increasing speed: (1) tuning the actuation frequency for a fixed aspect ratio, (2) adjusting the aspect ratio given a desired range of operating frequencies, and (3) using the weaker value of linker stiffness from among the values that we tested, while still maintaining a robust connection between the links. We also find experimentally that spherical two-link swimmers with dissimilar link diameters achieve net velocities comparable to swimmers with cylindrical links, but that two-link spherical swimmers of equal diameter do not.

Fusion of video and inertial sensing data via dynamic optimization of a biomechanical model.

Inertial sensing and computer vision are promising alternatives to traditional optical motion tracking, but until now these data sources have been explored either in isolation or fused via unconstrained optimization, which may not take full advantage of their complementary strengths. By adding physiological plausibility and dynamical robustness to a proposed solution, biomechanical modeling may enable better fusion than unconstrained optimization. To test this hypothesis, we fused video and inertial sensing data via dynamic optimization with a nine degree-of-freedom model and investigated when this approach outperforms video-only, inertial-sensing-only, and unconstrained-fusion methods. We used both experimental and synthetic data that mimicked different ranges of video and inertial measurement unit (IMU) data noise. Fusion with a dynamically constrained model significantly improved estimation of lower-extremity kinematics over the video-only approach and estimation of joint centers over the IMU-only approach. It consistently outperformed single-modality approaches across different noise profiles. When the quality of video data was high and that of inertial data was low, dynamically constrained fusion improved estimation of joint kinematics and joint centers over unconstrained fusion, while unconstrained fusion was advantageous in the opposite scenario. These findings indicate that complementary modalities and techniques can improve motion tracking by clinically meaningful margins and that data quality and computational complexity must be considered when selecting the most appropriate method for a particular application.

Nonuniform structural properties of wings confer sensing advantages.

Sensory feedback is essential to both animals and robotic systems for achieving coordinated, precise movements. Mechanosensory feedback, which provides information about body deformation, depends not only on the properties of sensors but also on the structure in which they are embedded. In insects, wing structure plays a particularly important role in flapping flight: in addition to generating aerodynamic forces, wings provide mechanosensory feedback necessary for guiding flight while undergoing dramatic deformations during each wingbeat. However, the role that wing structure plays in determining mechanosensory information is relatively unexplored. Insect wings exhibit characteristic stiffness gradients and are subject to both aerodynamic and structural damping. Here we examine how both of these properties impact sensory performance, using finite element analysis combined with sensor placement optimization approaches. We show that wings with nonuniform stiffness exhibit several advantages over uniform stiffness wings, resulting in higher accuracy of rotation detection and lower sensitivity to the placement of sensors on the wing. Moreover, we show that higher damping generally improves the accuracy with which body rotations can be detected. These results contribute to our understanding of the evolution of the nonuniform stiffness patterns in insect wings, as well as suggest design principles for robotic systems.

Adapting small jumping robots to compliant environments.

Jumping animals launch themselves from surfaces that vary widely in compliance from grasses and shrubs to tree branches. However, studies of robotic jumpers have been largely limited to those jumping from rigid substrates. In this paper, we leverage recent work describing how latches in jumping systems can mediate the transition from stored potential energy to kinetic energy. By including a description of the latch in our system model of both the jumper and compliant substrate, we can describe conditions in which a jumper can either lose energy to the substrate or recover energy from the substrate resulting in an improved jump performance. Using our mathematical model, we illustrate how the latch plays a role in the ability of a system to adapt its jump performance to a wide range of substrates that vary in their compliance. Our modelling results are validated using a 4 g jumper with a range of latch designs jumping from substrates with varying mass and compliance. Finally, we demonstrate the jumper recovering energy from a tree branch during take-off, extending these mechanistic findings to robots interacting with a more natural environment.

Geometric latches enable tuning of ultrafast, spring-propelled movements.

The smallest, fastest, repeated-use movements are propelled by power-dense elastic mechanisms, yet the key to their energetic control may be found in the latch-like mechanisms that mediate transformation from elastic potential energy to kinetic energy. Here, we tested how geometric latches enable consistent or variable outputs in ultrafast, spring-propelled systems. We constructed a reduced-order mathematical model of a spring-propelled system that uses a torque reversal (over-center) geometric latch. The model was parameterized to match the scales and mechanisms of ultrafast systems, specifically snapping shrimp. We simulated geometric and energetic configurations that enabled or reduced variation of strike durations and dactyl rotations given variation of stored elastic energy and latch mediation. Then, we collected an experimental dataset of the energy storage mechanism and ultrafast snaps of live snapping shrimp (Alpheus heterochaelis) and compared our simulations with their configuration. We discovered that snapping shrimp deform the propodus exoskeleton prior to the strike, which may contribute to elastic energy storage. Regardless of the amount of variation in spring loading duration, strike durations were far less variable than spring loading durations. When we simulated this species' morphological configuration in our mathematical model, we found that the low variability of strike duration is consistent with their torque reversal geometry. Even so, our simulations indicate that torque reversal systems can achieve either variable or invariant outputs through small adjustments to geometry. Our combined experiments and mathematical simulations reveal the capacity of geometric latches to enable, reduce or enhance variation of ultrafast movements in biological and synthetic systems.

A Method for 3D Printing and Rapid Prototyping of Fieldable Untethered Soft Robots.

Because they are made of elastically deformable and compliant materials, soft robots can passively change shape and conform to their environment, providing potential advantages over traditional robotics approaches. However, existing manufacturing workflows are often labor intensive and limited in their ability to create highly integrated three-dimensional (3D) heterogeneous material systems. In this study, we address this with a streamlined workflow to produce field-deployable soft robots based on 3D printing with digital light processing (DLP) of silicone-like soft materials. DLP-based 3D printing is used to create soft actuators (2.2 g) capable of exerting up to 0.5 Newtons of force that are integrated into a bioinspired untethered soft robot. The robot walks underwater at speeds comparable with its biological analog, the brittle star. Using a model-free planning algorithm and feedback, the robot follows remote commands to move to desired positions. Moreover, we show that the robot is able to perform untethered locomotion outside of a laboratory and in a natural aquatic environment. Our results represent progress in soft robot manufacturing autonomy for a 3D printed untethered soft robot.

Can You Do That Again? Time Series Consolidation as a Robust Method of Tailoring Gesture Recognition to Individual Users.

Robust inter-session modeling of gestures is still an open learning challenge. A sleeve equipped with capacitive strap sensors was used to capture two gesture data sets from a convenience sample of eight subjects. Two pipelines were explored. In FILT a novel two-stage algorithm was introduced which uses an unsupervised learning algorithm to find samples representing gesture transitions and discards them prior to training and validating conventional models. In TSC a confusion matrix was used to automatically consolidate commonly confused class labels, resulting in a set of gestures tailored to an individual subject's abilities. The inter-session testing accuracy using the Time Series Consolidation (TSC) method increased from a baseline inter-session average of 42.47 ± 3.83% to 93.02% ± 4.97% while retaining an average of 5.29 ± 0.46 out of the 11 possible gesture categories. These pipelines used classic machine learning algorithms which require relatively small amounts of data and computational power compared to deep learning solutions. These methods may also offer more flexibility in interface design for users suffering from handicaps limiting their manual dexterity or ability to reliably make gestures, and be possible to implement on edge devices with low computational power.

Dual spring force couples yield multifunctionality and ultrafast, precision rotation in tiny biomechanical systems.

Small organisms use propulsive springs rather than muscles to repeatedly actuate high acceleration movements, even when constrained to tiny displacements and limited by inertial forces. Through integration of a large kinematic dataset, measurements of elastic recoil, energetic math modeling and dynamic math modeling, we tested how trap-jaw ants (Odontomachus brunneus) utilize multiple elastic structures to develop ultrafast and precise mandible rotations at small scales. We found that O. brunneus develops torque on each mandible using an intriguing configuration of two springs: their elastic head capsule recoils to push and the recoiling muscle-apodeme unit tugs on each mandible. Mandibles achieved precise, planar, circular trajectories up to 49,100 rad s-1 (470,000 rpm) when powered by spring propulsion. Once spring propulsion ended, the mandibles moved with unconstrained and oscillatory rotation. We term this mechanism a 'dual spring force couple', meaning that two springs deliver energy at two locations to develop torque. Dynamic modeling revealed that dual spring force couples reduce the need for joint constraints and thereby reduce dissipative joint losses, which is essential to the repeated use of ultrafast, small systems. Dual spring force couples enable multifunctionality: trap-jaw ants use the same mechanical system to produce ultrafast, planar strikes driven by propulsive springs and for generating slow, multi-degrees of freedom mandible manipulations using muscles, rather than springs, to directly actuate the movement. Dual spring force couples are found in other systems and are likely widespread in biology. These principles can be incorporated into microrobotics to improve multifunctionality, precision and longevity of ultrafast systems.

Flow separation sensing on airfoil using a 3D printed biomimetic artificial hair sensor.

Small-scale unmanned air vehicles require lightweight, compact, and low-power sensors that encompass a variety of sensing modalities to enable flight control and navigation in challenging environments. Flow sensing is one such modality that has attracted much interest in recent years. In this paper, a micro-scale artificial hair sensor is developed to resolve both the direction and magnitude of airflow. The sensor structure employs a high-aspect ratio hair structure and a thin flexible membrane to facilitate the transduction of directional airflow to membrane deflection. The sensor readout is based on capacitive sensing and two pairs of electrodes orthogonal to each other are used to obtain airflow directional information. The sensor structure was fabricated using two-photon polymerization and integration onto a miniature printed circuit board to enable simple measurement. The sensor's responses to static displacement loading from different directions were characterized. The experimental results are in good agreement with the simulation results. Furthermore, the sensor's capability to measure the direction and magnitude of flow was demonstrated. Finally, the sensor was mounted on an airfoil and its ability to detect flow separation was verified.

Pendulum-based measurements reveal impact dynamics at the scale of a trap-jaw ant.

Small organisms can produce powerful, sub-millisecond impacts by moving tiny structures at high accelerations. We developed and validated a pendulum device to measure the impact energetics of microgram-sized trap-jaw ant mandibles accelerated against targets at 105 m s-2 Trap-jaw ants (Odontomachus brunneus; 19 individuals, 212 strikes) were suspended on one pendulum and struck swappable targets that were either attached to an opposing pendulum or fixed in place. Mean post-impact kinetic energy (energy from a strike converted to pendulum motion) was higher with a stiff target (21.0-21.5 µJ) than with a compliant target (6.4-6.5 µJ). Target mobility had relatively little influence on energy transfer. Mean contact duration of strikes against stiff targets was shorter (3.9-4.5 ms) than against compliant targets (6.2-7.9 ms). Shorter contact duration was correlated with higher post-impact kinetic energy. These findings contextualize and provide an energetic explanation for the diverse, natural uses of trap-jaw ant strikes such as impaling prey, launching away threats and performing mandible-powered jumps. The strong effect of target material on energetic exchange suggests material interactions as an avenue for tuning performance of small, high acceleration impacts. Our device offers a foundation for novel research into the ecomechanics and evolution of tiny biological impacts and their application in synthetic systems.

Controlled Assembly of Liquid Metal Inclusions as a General Approach for Multifunctional Composites.

Soft composites that use droplets of gallium-based liquid metal (LM) as the dispersion phase have the potential for transformative impact in multifunctional material engineering. However, it is unclear whether percolation pathways of LM can support high electrical conductivity in a wide range of matrix materials. This issue is addressed through an approach to LM composite synthesis that focuses on the interrelated effects of matrix curing/solidification and droplet formation. The combined influence of LM concentration, particle size, and sedimentation is explored. By developing this approach, the functionalities that have been demonstrated with LM composites can be generalized to other matrix materials that impart additional functionality. Specifically, composites are synthesized using a biodegradable/reprocessable plastic (polycaprolactone), a hydrogel (poly(vinyl alcohol)), and a processable rubber (a styrene-ethylene-butylene-styrene derivative) to demonstrate wide applicability. This method enables synthesis of composites: i) with high stretchability and negligible electromechanical coupling (>600% strain); ii) with Joule-heated healing and reprocessability; iii) with electrical and mechanical self-healing; and iv) that can be printed. This approach to controlled assembly represents a widely applicable technique for creating new classes of LM composites with unprecedented multifunctionality.

Latch-based control of energy output in spring actuated systems.

The inherent force-velocity trade-off of muscles and motors can be overcome by instead loading and releasing energy in springs to power extreme movements. A key component of this paradigm is the latch that mediates the release of spring energy to power the motion. Latches have traditionally been considered as switches; they maintain spring compression in one state and allow the spring to release energy without constraint in the other. Using a mathematical model of a simplified contact latch, we reproduce this instantaneous release behaviour and also demonstrate that changing latch parameters (latch release velocity and radius) can reduce and delay the energy released by the spring. We identify a critical threshold between instantaneous and delayed release that depends on the latch, spring, and mass of the system. Systems with stiff springs and small mass can attain a wide range of output performance, including instantaneous behaviour, by changing latch release velocity. We validate this model in both a physical experiment as well as with data from the Dracula ant, Mystrium camillae, and propose that latch release velocity can be used in both engineering and biological systems to control energy output.

Viscoelastic legs for open-loop control of gram-scale robots.

Gram-scale insects, such as cockroaches, take advantage of the mechanical properties of the musculoskeletal system to enable rapid and robust running. Engineering gram-scale robots, much like their biological counterparts, comes with inherent constraints on resources due to their small sizes. Resource-constrained robots are generally limited in their computational complexity, making controlled, high-speed locomotion a challenge, especially in unstructured environments. In this paper we show that embedding control into the leg mechanics of robots, similarly to cockroaches, results in predictable dynamics from an open-loop control strategy that can be modified through material choice. Tuning the mechanical properties of gram-scale robot legs promotes high-speed, stable running, reducing the need for active control. We utilize a torque-driven damped spring-loaded inverted pendulum model to explore the behavior and the design space of a spring-damper leg at this scale. The resulting design maps show the trade-offs in performance goals, such as speed and efficiency, with stability, as well as the sensitivity in performance to the leg properties and the control input. Finally, we demonstrate experimental results with magnetically actuated quadrupedal gram-scale robots, incorporating viscoelastic legs and demonstrating speeds up to 11.7 body lengths per second.

Multimaterial 3D Printing for Microrobotic Mechanisms.

Multimaterial mechanisms are seen throughout natural organisms across all length scales. The different materials in their bodies, from rigid, structural materials to soft, elastic materials, enable mobility in complex environments. As robots leave the lab and begin to move in real environments, including a range of materials in 3D robotics mechanisms can help robots handle uncertainty and lessen control requirements. For the smallest robots, soft materials combined with rigid materials can facilitate large motions in compact spaces due to the increased compliance. However, integrating various material components in 3D at the microscale is a challenge. We present an approach for 3D microscale multimaterial fabrication using two-photon polymerization. Two materials with three orders of magnitude difference in Young's moduli are printed in consecutive cycles. Integrating a soft elastic material that is capable of more than 200% strain along with a rigid material has enabled the formation of hybrid elements, strongly adhered together, with layer accuracy below 3-µm resolution. We demonstrate a multilink multimaterial mechanism showing large deformation, and a 3D-printed 2-mm wingspan flapping wing mechanism, showing rapid prototyping of complex designs. This fabrication strategy can be extended to other materials, thus enhancing the functionality and complexity of small-scale robots.

The principles of cascading power limits in small, fast biological and engineered systems.

Mechanical power limitations emerge from the physical trade-off between force and velocity. Many biological systems incorporate power-enhancing mechanisms enabling extraordinary accelerations at small sizes. We establish how power enhancement emerges through the dynamic coupling of motors, springs, and latches and reveal how each displays its own force-velocity behavior. We mathematically demonstrate a tunable performance space for spring-actuated movement that is applicable to biological and synthetic systems. Incorporating nonideal spring behavior and parameterizing latch dynamics allows the identification of critical transitions in mass and trade-offs in spring scaling, both of which offer explanations for long-observed scaling patterns in biological systems. This analysis defines the cascading challenges of power enhancement, explores their emergent effects in biological and engineered systems, and charts a pathway for higher-level analysis and synthesis of power-amplified systems.

Rapid Three-Dimensional Printing in Water Using Semiconductor-Metal Hybrid Nanoparticles as Photoinitiators.

Additive manufacturing processes enable fabrication of complex and functional three-dimensional (3D) objects ranging from engine parts to artificial organs. Photopolymerization, which is the most versatile technology enabling such processes through 3D printing, utilizes photoinitiators that break into radicals upon light absorption. We report on a new family of photoinitiators for 3D printing based on hybrid semiconductor-metal nanoparticles. Unlike conventional photoinitiators that are consumed upon irradiation, these particles form radicals through a photocatalytic process. Light absorption by the semiconductor nanorod is followed by charge separation and electron transfer to the metal tip, enabling redox reactions to form radicals in aerobic conditions. In particular, we demonstrate their use in 3D printing in water, where they simultaneously form hydroxyl radicals for the polymerization and consume dissolved oxygen that is a known inhibitor. We also demonstrate their potential for two-photon polymerization due to their giant two-photon absorption cross section.