Bioinspiration & Biomimetics

Aquatic organisms often employ maneuverable and agile swimming behavior to escape from predators, find prey, or navigate through complex environments. Many of these organisms use metachronally coordinated appendages to execute complex maneuvers. However, though metachrony is used across body sizes ranging from microns to tens of centimeters, it is understudied compared to the swimming of fish, cetaceans, and other groups. In particular, metachronal coordination and control of multiple appendages for three-dimensional maneuvering is not fully understood. To explore the maneuvering capabilities of metachronal swimming, we combine 3D high-speed videography of freely swimming ctenophores (Bolinopsis vitrea) with reduced-order mathematical modeling. Experimental results show that ctenophores can quickly reorient, and perform tight turns while maintaining forward swimming speeds close to 70% of their observed maximum -- performance comparable to or exceeding that of many vertebrates with more complex locomotor systems. We use a reduced-order model to investigate turning performance across a range of beat frequencies and appendage control strategies, and reveal that ctenophores are capable of near-omnidirectional turning. Based on both recorded and modeled swimming trajectories, we conclude that the ctenophore body plan enables a high degree of maneuverability and agility, and may be a useful starting point for future bioinspired aquatic vehicles. Author summaryMetachronal swimming--the sequential, coordinated beating of appendages arranged in a row-- exists across a wide range of sizes, from unicellular organisms (micrometers) to marine crustaceans (tens of centimeters). While metachronal swimming is known to be scalable and efficient, the level of maneuverability and agility afforded by this strategy is not well understood. This study explores the remarkable 3D maneuverability of ctenophores (comb jellies), and the appendage control strategies they use to achieve it. Ctenophores have eight rows of appendages (instead of the one or two found in crustaceans and other organisms). This higher number of appendages, their distribution along the body, and the independent frequency control between paired rows enables near-omnidirectional swimming and turning performance, placing ctenophores among the most maneuverable swimmers. We use experiments and mathematical modeling to explore both the real and theoretical performance landscape of the ctenophore body plan, and show that ctenophores are capable of executing tight turns at high speeds in nearly any plane. This omnidirectional swimming capability gives insight into the ecology and behavior of an important taxonomic group, and shows the potential of metachronal swimming as a source of design inspiration for robotic vehicles (particularly those that must navigate complex environments).

Humans ability to grasp and dynamically manipulate objects with their hands is unmatched by current robots. To better understand human dynamic manipulation, we studied dice stacking, a task in which humans form a vertical stack of dice from a set of initially unstacked playing dice using an overturned cup and the surface of a table. This task is high dimensional and under-actuated, so it may superficially seem an incredible feat of state estimation and feedback control, but we show that this task is amenable to open-loop strategies. We simulated a cup with dice oscillated by fixed arm movement patterns using two different computer simulation frameworks with different contact models. These simulations showed that, for a range of arm and wrist movements, the dice naturally stack without any dice state feedback. We verified the predictions of these simulations with a physical robot. Thus, we have added dice stacking to the small list of dynamic manipulation tasks that can be robustly performed open-loop. We speculate that, for highly under-actuated tasks, humans may be biased to learn open-loop strategies over state feedback strategies. Future work could investigate the presence of such a bias in humans and its potential value for reinforcement learning algorithms.

Flying insects routinely demonstrate coordinated flight in groups. How they achieve this with very limited communication, vision, and neural systems remains an open question. We measured the visual reaction time in flying honeybees while they chased a moving target, and compared in-flight reaction times for solo animals with those flying in groups. Across 425 insects, the solo honeybees show diverse reaction times (an average of 30ms and a standard deviation of 50ms). The reaction times in groups are significantly more uniform (an average of 15ms and a standard deviation of only 7ms), indicating that honeybees in group flight adjust their reaction times to match their neighbors. To investigate the role of this adjustment, we curve fit the reaction time distributions and analyzed them in a mathematical model of swarming, finding that the reaction time increases the stable region of a cohesive swarm. To verify the stabilizing effect was not an artifact of curve fitting, we then inserted the measured delays in a swarm simulation, which breaks apart under the solo reaction times and achieves stable formations for the group reaction times. Together, our findings highlight how flying animals can synchronize their reaction times in group flights to improve group cohesion.

In this paper,we consider the feasibility of mimicking the sprawling gait of a live varanid (Varanus salvator) using a necrobot (named: Pak Biawak), a robot constructed using the skeletal parts of a deceased varanid of the same species. Pak Biawak is manufactured using simple joints and components, and limb motion is coupled to passive spine bending to enable the sprawling gait. Here, we assess both the lateral and dorsal kinematics of Pak Biawak at different speeds, and compare the metrics from each to those of a similarly sized live varanid. When assessing lateral view shape metrics (stride aspect ratio, stride circularity, normalised stride swept area, normalised stride swept area perimeter), we find that Pak Biawaks gait is consistent across all speeds and the majority of Pak Biawaks lateral shape metrics are kinematically aligned with those of the live varanid. This also proves true when comparing Pak Biawaks lateral trajectory metrics (radial distance of swept area, normallised root mean squared error) against those of the live varanid, and at different speeds of sprawling. Pak Biawaks dorsal metrics include the spine bending amplitude and period, and these are not found to be significantly different to those of the live varanid, however, Pak Biawaks amplitude is affected by sprawling speed. We use three metrics to compare forward and reverse limb sweeps including, angular curvature, differential curvature, and a normalised arc length. Of these, a preponderance of highly significant differences (p[≤] 0.001) are observed on comparing the forward sweep arc length of Pak Biawak at every sprawling speed against the forward sweep arc length of the live lizard. All other kinematic metrics in the necrobot are nevertheless very close to those of the live lizard. Finally, when comparing the trackway width of Pak Biawak against the live lizard, we again find there is very close kinematic compatibility between the two, and conclude that our necrobot can be designed and manufactured to mimic the sprawling gait of a real varanid, even when using simple kinematic linkages in unison with a passive spine bending differential applied at only one central location in the necrobot spine.

Insect flight is a complex interdisciplinary phenomenon. Understanding its multiple aspects, such as flight control, sensory integration and genetics, often requires the analysis of large amounts of free flight kinematic data. Yet, one of the main bottlenecks in this field is automatically and accurately extracting such data from multi-view videos. Here, we present a model-based method for pose-estimation of free-flying fruit flies from multi-view high-speed videos. To obtain a faithful representation of the fly with minimum free parameters, our method uses a 3D model that mimics two new aspects of wing deformation: a non-fixed wing hinge and a twisting wing surface. The method is demonstrated for free and perturbed flight. Our method does not use prior assumptions on the kinematics apart from the continuity of one wing angle. Hence, this method can be readily adjusted for other insect species.

Bipedal locomotion has several key challenges, such as balancing, foot placement, and gait optimization. We reach optimality from a very early age by using natural supports, such as our parents hands, chairs, and training wheels, and bootstrap a new knowledge from the recently acquired one. In this paper, we propose a scaffolded learning method from an evolutionary robotics perspective, where a biped creature achieves stable and independent bipedal walking while exploiting the natural scaffold of its changing morphology to create a third limb. Hence, we compare three conditions of scaffolded learning to reach bipedalism, and we prove that a performance-based scaffold is the most conducive to accelerate the learning of ontogenetic bipedal walking. Beyond a pedagogical experiment, this work presents a powerful tool to accelerate learning on robots.

Individuals who suffer from severe paralysis often lose the capacity to perform fundamental body movements and everyday activities. Empowering these individuals with the ability to operate robotic arms, in high-dimensions, helps to maximize both functional utility and human agency. However, high-dimensional robot teleoperation currently lacks accessibility due to the challenge in capturing high-dimensional control signals from the human, especially in the face of motor impairments. Body-machine interfacing is a viable option that offers the necessary high-dimensional motion capture, and it moreover is noninvasive, affordable, and promotes movement and motor recovery. Nevertheless, to what extent body-machine interfacing is able to scale to high-dimensional robot control, and whether it is feasible for humans to learn, remains an open question. In this exploratory multi-session study, we demonstrate the feasibility of human learning to operate a body-machine interface to control a complex, assistive robotic arm in reaching and Activities of Daily Living tasks. Our results suggest the manner of control space mapping, from interface to robot, to play a critical role in the evolution of human learning.

Systematic descriptions of the underlying interaction rules that insects use to support group and swarm flight has the potential to contribute to mathematics, biology, and robotics, including aerial swarming under sensory and computational limitations. This study analyzes 1,000 trajectories of flying honeybees in crowded conditions approaching a moving stimulus and finds how during this stimulus, honeybees coordinate flight through pairwise interactions involving a novel three-zone decision-making process. The experimental setup consists of 3-D position reconstructions via a high speed camera system recording honeybee foragers returning to a hive entrance actuated to move robotically. The analysis consists of neighborhood identification through three methods (cross-correlation, distance threshold, and average distance threshold), which reveals the dominant interaction is pairwise. The individual leader-follower pair interactions are then tested against three regulation candidates: optic flow, relative velocity, and optical expansion rate, based on minimizing root mean square error. The results show that each follower demonstrates a three stage process involving a feedback rule change, linked by an intermediate observation/decision phase. During the initial "lock" phase, an insect maintains a consistent optical expansion rate until inter-agent distance closes to 10 cm. The regulation candidates then undergo large variations during a relatively long observation/decision zone, with 1.04 seconds being the average time in the decision zone. 79% of the paired insect entries into the decision zone result in subsequent re-engagement to track the same initial leader, while 21% result in disengagement from the group behavior. Visual regulation candidate comparison in the third stage indicates that upon re-engagement, the follower relative velocity is regulated to provide consistent velocity matching between agents. The third stages velocity tracking is consistent with a closed-loop feedback proportional-integral (PI) controller regulating velocity tracking error. Across the insect population studied, the proportional gain remained showed minimal variability over individuals, a derivative gain was considered and found negligible, and the integral gain varied by individual. Collectively, these findings underscore the existence of an alternative swarm architecture, highlighting individual decision-making capabilities, feedback regulation target changes, and the presence of reactive, deliberative, and moderate (PI control) timescale interaction rules contained within aerial groups.

Upper body function is lost when injuries are sustained to the cervical spinal cord. Assistive machines can support the loss in upper body motor function. To regain functionality at the level of performing activities of daily living (e.g., self-feeding), though, assistive machines need to be able to operate in high dimensions. This means there is a need for interfaces with the capability to match high-dimensional operation. The body-machine interface provides this capability and has shown to be a suitable interface even for individuals with limited mobility. This is because it can take advantage of peoples available residual body movements. Previous studies using this interface have only shown that the interface can control low-dimensional assistive machines. In this pilot study, we demonstrate the interface can scale to high-dimensional robots, can be learned to control a 7-dimensional assistive robotic arm, to perform complex reaching and functional tasks, by an uninjured population. We also share results from various analyses that hint at learning, even when performance is extremely low. Decoupling intrinsic correlations between robot control dimensions seem to be a factor in learning--that is, proficiency in activating each control dimension independently may contribute to learning and skill acquisition of high-dimensional robot control. In addition, we show that learning to control the robot and learning to perform complex movement tasks can occur simultaneously.

Intermittent locomotion composed of periods of active flapping/stroking followed by inactive gliding has been observed with species that inhabit both aerial and marine environments. However, studies on the energetic benefits of a fluke-and-glide (FG) gait during horizontal locomotion are limited for dolphins. This work presents a physics-based model of FG gait and analysis of the associated costs of transport of bottlenose dolphins (Tursiops truncatus). New estimates of gliding drag coefficients for the model were estimated experimentally from free-swimming bottlenose dolphins. The data-driven approach used kinematic measurement from 84 hours of biologging tag data collected from 3 animals to estimate the coefficients. A set of 532 qualified gliding events were automatically extracted for gliding drag coefficient estimation, and an additional 783 FG bouts were parameterized and used to inform the model-based dynamic analysis. Experimental results indicate that FG gait was preferred at speeds around 2.2 - 2.7 m/s. Observed FG bouts had an average duty factor of 0.45 and gliding duration of 5 sec. The average associated metabolic cost of transport (COT) and mechanical cost of transport (MECOT) of FG gait are 2.53 and 0.35 J {middle dot} m-1 {middle dot} kg-1 at the preferred speeds. This corresponded to an 18.9% and 27.1% reduction in cost when compared to modeled continuous fluking gait at the same reference speed. Average thrust was positively correlated with fluking frequency and amplitude as animals accelerated during the FG bouts. While fluking frequency and amplitude were negatively correlated for a given thrust range. These results support the supposition that FG gait enhances the horizontal swimming efficiency of bottlenose dolphins and provides new dynamical insights into the gait of these animals.

Bloodworms, Glycera dibranchiata, possess an eversible proboscis that normally remains concealed within their bodies but explosively everts if the worm attacks or burrows. How does the bloodworm evert quickly and reliably? In a series of experiments, we characterize bloodworm kinematics, pressure, and material properties to estimate the criteria for eversion safely without rupture of the proboscis. We predict the proboscis can withstand pressures 50 times higher and bending strains up to three times higher than the respective values observed. We also present a dimensional analysis of eversion, finding that everting animals, from frogs to snails to sharks, do not satisfy Froudes law for equivalence of velocities. Our findings may help inspire the development of pressure-driven soft robots with efficient retraction capabilities.

The study of animal locomotion and neuromechanical control offers valuable insights for advancing research in neuroscience, biomechanics, and robotics. We have developed FARMS (Framework for Animal and Robot Modeling and Simulation), an open-source, interdisciplinary framework, designed to facilitate access to modeling, simulation, and analysis of animal locomotion and bio-inspired robotic systems. By providing an accessible and user-friendly platform, FARMS aims to lower the barriers for researchers to explore the complex interactions between the nervous system, musculoskeletal structures, and their environment. Integrating the MuJoCo physics engine in a modular manner, FARMS enables realistic simulations and fosters collaboration among neuroscientists, biologists, and roboticists. FARMS has already been extensively used to study locomotion in animals such as mice, drosophila, fish, salamanders, and centipedes, serving as a platform to investigate the role of central pattern generators and sensory feedback. This article provides an overview of the FARMS framework, discusses its interdisciplinary approach, showcases its versatility through specific case studies, and highlights its effectiveness in advancing our understanding of locomotion. Overall, the goal of FARMS is to contribute to a deeper understanding of animal locomotion, the development of innovative bio-inspired robotic systems, and promote accessibility in neuromechanical research.

In this study, we utilize a high-speed camera array to meticulously capture the intricate wing kinematics of honeybees during free voluntary takeoffs. This allows us to investigate the variation in kinematic parameters over time. According to the variation of Euler angles, the takeoff process can be categorized into three stages: initial, adjusting, and stable. Our analysis reveals that honeybees typically execute at least 15 wingbeats before taking off, with wing stroke amplitudes exceeding 100 degrees and wingbeat frequencies ranging from 180 to 260 Hz. Significantly, the hinged wing mechanism, connecting the forewing and hindwing via hamuli, undergoes considerable chordwise deformation during this process, with the angle between the fore- and hind-wings surpassing 50 degrees. Notably, the forewing and hindwing maintain a positive camber throughout the wingbeat cycle during takeoff, contributing to the enhanced thrust generation instead of lift, comparing to the flat wings. The positive camber can be passively formed at beginning and ending of each half stroke, but should be actively maintained around middle half-stroke.This study provides valuable insights for aeronautical engineers in the design of flapping wing micro air vehicles, specifically in the effective implementation of hinged wings inspired by the wing motion of honeybees.

Animals and living organisms are continuously adapting to changes in their environment. How do animals, especially those that are critical to their ecosystem, respond to rapidly changing conditions in their environment? Here, we report on the three-dimensional trajectories of flying honeybees under calm and windy conditions in front of the hive entrance. We also investigate the pitch and yaw in our experiments. We find that the mean velocities, accelerations and angular velocities of honeybees increase with increasing wind speeds. We observed that pair separation between honeybees is highly controlled and independent of wind speeds. Our results on the coordination used by honeybees may have potential applications for coordinated flight of unmanned aerial vehicles.

Insects show diverse flight kinematics and morphologies reflecting their evolutionary histories and ecological adaptations. Many silkmoths utilizing low wingbeat frequencies and large wings to fly display body oscillations: Their bodies pitch and bob periodically - synchronized with their wing flapping cycle. Similar oscillations in butterflies augment weight support and thrust and reduce flight power requirements. However, how the instantaneous body and wing kinematics interact for these beneficial aerodynamic and power consequences is not well understood. We hypothesized that the body oscillations affect aerodynamic power requirements by influencing the wing rotation relative to the airflow. Using three-dimensional forward flight video recordings of four silkmoth species and a quasi-steady blade-element aerodynamic method, we found that the body pitch angle and the wing sweep angle maintain a narrow range of phase differences to enhance the angle of attack variation between each half-stroke due to enhanced wing rotation relative to the airflow. This redirects the aerodynamic force to increase upward and forward force during downstroke and upstroke respectively thus lowering the overall drag without compromising weight support and forward thrust. A reduction in energy expenditure is beneficial because adult silkmoths do not feed and rely on limited energy budgets.

To better understand the turning flight characteristics of the bees, we developed a procedure for estimating the instantaneous total force generated by the bee, along with its centrifugal force component, at each instant of time. We calculated the magnitude and direction of the total force vector (TFV) with respect to the three body axes of the bee and examined its variation with time during turning flights. The results of this study revealed that turns in the cloud are executed by (a) holding the magnitude of the TFV vector constant and (b) by redirecting the body (and therefore the turning force) appropriately to execute a coordinated turn. We also calculated and analysed the TFVs of bees flying in a curved tunnel. The characteristics of these TFVs (magnitude and direction) are very similar to those for bees flying in the cloud. This is a novel finding as there are no studies which have estimated the instantaneous flight force magnitude and direction of a turning bee (these are not saccadic or evasive turns) in an outdoor or indoor setup. However, similar results have been reported for pigeons and cockatiels while executing smooth turns in the horizontal plane.

Task space control, also known as operational space control, is a useful paradigm for investigating neural control of human movement using predictive simulations. While some efforts have been made to implement task space control in the widely used open-source platform OpenSim, existing implementations do not support floating base kinematics, which is necessary for simulating gait and other types of human movement. Our aim in this work is to fill that gap. In this paper, we describe the theory and implementation of a floating base kinematics task space framework for torque- and muscle-driven simulations in OpenSim. Our framework builds on previous work that was limited to models with a base (i.e., root) segment fixed to ground. In addition, we integrate various algorithms from robotics in order to handle dynamically changing contacts and task prioritization. The framework can be used to generate realistic walking gaits by prescribing a small set of controller gains and gait parameters such as step length, step width and center of mass velocity. Task can be specified as desired positions, rotations, or higher-order feature such as base of support and whole-body angular momentum. We provide several examples to demonstrate how framework is successful in orchestrating a complex hierarchy of tasks that work in concert to perform both balance control and gait generation. The implementation is freely available for roboticists and biomechanists to use with OpenSim. Author summaryRecent advances in computational biomechanics have provided researchers with tools capable of predicting human movement. Previous approaches to simulating human movement required experimental data as input and the simulation would replicate the experimental motion. This conventional approach limited the scientific insights to the specific movement recorded in the laboratory. With predictive approaches, researchers can investigate how a person might respond to various factors, such as reduced muscle strength or an assistive device such as a robotic exoskeleton. Existing approaches for generating predictive simulations utilize optimization-based approaches, which can be time-consuming and difficult to troubleshoot. Task space control is an alternative approach which is widely used in robotics. Conceptually, task space control aims to generate a simulation by specifying "tasks", such as moving a hand or foot to a desired position, and computing the joint angles required to achieve the task. Here we aim to outline the mathematics behind task space control and demonstrate how task space control can be used to generate simulations of movements such as walking.

Insects exhibit remarkable adaptability in their locomotive strategies across diverse environments, a crucial trait for foraging, survival, and predator avoidance. Microvelia, tiny 2-3 mm insects that adeptly walk on water surfaces, exemplify this adaptability by using the alternating tripod gait in both aquatic and terrestrial terrains. These insects commonly inhabit low-flow ponds and streams cluttered with natural debris like leaves, twigs, and duckweed. Using high-speed imaging and pose-estimation software, we analyze Microvelia spp.s movement across water, sandpaper (simulating land), and varying duckweed densities (10%, 25%, and 50% coverage). Our results reveal Microvelia maintain consistent joint angles and strides of their upper and hind legs across all duckweed coverages, mirroring those seen on sandpaper. Microvelia adjust the stride length of their middle legs based on the amount of duckweed present, decreasing with increased duckweed coverage and at 50% duckweed coverage, their middle legs strides closely mimic their strides on sandpaper. Notably, Microvelia achieve speeds up to 56 body lengths per second on water, nearly double those observed on sandpaper and duckweed (both rough, frictional surfaces), highlighting their higher speeds on low friction surfaces such as the waters surface. This study highlights Microvelias ecological adaptability, setting the stage for advancements in amphibious robotics that emulate their unique tripod gait for navigating complex terrains.

Contact robots are increasingly used to assist humans in physical training and manufacturing tasks. However, the effectiveness of current systems is limited as their control focuses on the system performance without explicitly considering the upcoming human users control. Here we present a differential game-based controller for contact robots that ensures optimal interaction with the human user by predicting their control during movement while considering their inherently finite planning horizon. Using this model-predictive game (MPG) controller, we systematically investigated human-robot co-adaptation in experiments, demonstrating that: (a) interaction with MPG remains stable for all participants while effectively reducing human effort; (b) the robot adapts to human behavior, identifying and characterizing individual motor control strategies that remain consistent over time and may be used as control biomarkers; (c) the human adapts to the robots behavior, and their interaction behavior can be modulated through an assistance meta-parameter. These findings indicate that humans can understand and adapt to a partners control strategy during movement, thereby exhibiting behavior consistent with game theory principles. Furthermore, the ability of the assistance meta-parameter to guide human users toward specific interaction behaviors can be used to develop versatile robot-assisted learning systems for physical training and rehabilitation.

Experimental observations of the 3-dimensional wake of swimmers are scarce. This study provides the first experimental measurements of the three-dimensional vortical structures of an anguilliform swimmer. A volumetric velocimetry (DDPTV) setup was used to quantify the wake of freely swimming dice snakes (Natrix tessellata) in a tank. Multiple swimming trials were recorded with three snakes swimming at a forward swimming speed ranging from 0.2 to 0.8L.s-1. The volumetric measurements show that during swimming, Natrix tessellata sheds vortex tubes from different parts of its body with oscillating lateral undulations. The vortex tubes are then linked to form a hairpin-like structure as predicted by computational fluid dynamic analyses. Quantitative measurements show that the vortex size gradually increases over time. The vortex circulation decreases after attaining a maximum. These results provide an experimental demonstration that the 3D wake structure of an anguilliform swimmer takes well organized dynamic shapes as predicted and provide baseline data for future studies comparing the wake structure of snakes with different locomotor ecologies.