Bridging two insect flight modes in evolution, physiology and robophysics.

Since taking flight, insects have undergone repeated evolutionary transitions between two seemingly distinct flight modes1-3. Some insects neurally activate their muscles synchronously with each wingstroke. However, many insects have achieved wingbeat frequencies beyond the speed limit of typical neuromuscular systems by evolving flight muscles that are asynchronous with neural activation and activate in response to mechanical stretch2-8. These modes reflect the two fundamental ways of generating rhythmic movement: time-periodic forcing versus emergent oscillations from self-excitation8-10. How repeated evolutionary transitions have occurred and what governs the switching between these distinct modes remain unknown. Here we find that, despite widespread asynchronous actuation in insects across the phylogeny3,6, asynchrony probably evolved only once at the order level, with many reversions to the ancestral, synchronous mode. A synchronous moth species, evolved from an asynchronous ancestor, still preserves the stretch-activated muscle physiology. Numerical and robophysical analyses of a unified biophysical framework reveal that rather than a dichotomy, these two modes are two regimes of the same dynamics. Insects can transition between flight modes across a bridge in physiological parameter space. Finally, we integrate these two actuation modes into an insect-scale robot11-13 that enables transitions between modes and unlocks a new self-excited wingstroke strategy for engineered flight. Together, this framework accounts for repeated transitions in insect flight evolution and shows how flight modes can flip with changes in physiological parameters.

The origin of blinking in both mudskippers and tetrapods is linked to life on land.

Blinking, the transient occlusion of the eye by one or more membranes, serves several functions including wetting, protecting, and cleaning the eye. This behavior is seen in nearly all living tetrapods and absent in other extant sarcopterygian lineages suggesting that it might have arisen during the water-to-land transition. Unfortunately, our understanding of the origin of blinking has been limited by a lack of known anatomical correlates of the behavior in the fossil record and a paucity of comparative functional studies. To understand how and why blinking originates, we leverage mudskippers (Oxudercinae), a clade of amphibious fishes that have convergently evolved blinking. Using microcomputed tomography and histology, we analyzed two mudskipper species, Periophthalmus barbarus and Periophthalmodon septemradiatus, and compared them to the fully aquatic round goby, Neogobius melanostomus. Study of gross anatomy and epithelial microstructure shows that mudskippers have not evolved novel musculature or glands to blink. Behavioral analyses show the blinks of mudskippers are functionally convergent with those of tetrapods: P. barbarus blinks more often under high-evaporation conditions to wet the eye, a blink reflex protects the eye from physical insult, and a single blink can fully clean the cornea of particulates. Thus, eye retraction in concert with a passive occlusal membrane can achieve functions associated with life on land. Osteological correlates of eye retraction are present in the earliest limbed vertebrates, suggesting blinking capability. In both mudskippers and tetrapods, therefore, the origin of this multifunctional innovation is likely explained by selection for increasingly terrestrial lifestyles.

Moth resonant mechanics are tuned to wingbeat frequency and energetic demands.

An insect's wingbeat frequency is a critical determinant of its flight performance and varies by multiple orders of magnitude across Insecta. Despite potential energetic benefits for an insect that matches its wingbeat frequency to its resonant frequency, recent work has shown that moths may operate off their resonant peak. We hypothesized that across species, wingbeat frequency scales with resonance frequency to maintain favourable energetics, but with an offset in species that use frequency modulation as a means of flight control. The moth superfamily Bombycoidea is ideal for testing this hypothesis because their wingbeat frequencies vary across species by an order of magnitude, despite similar morphology and actuation. We used materials testing, high-speed videography and a model of resonant aerodynamics to determine how components of an insect's flight apparatus (stiffness, wing inertia, muscle strain and aerodynamics) vary with wingbeat frequency. We find that the resonant frequency of a moth correlates with wingbeat frequency, but resonance curve shape (described by the Weis-Fogh number) and peak location vary within the clade in a way that corresponds to frequency-dependent biomechanical demands. Our results demonstrate that a suite of adaptations in muscle, exoskeleton and wing drive variation in resonant mechanics, reflecting potential constraints on matching wingbeat and resonant frequencies.

An information theoretic method to resolve millisecond-scale spike timing precision in a comprehensive motor program.

Sensory inputs in nervous systems are often encoded at the millisecond scale in a precise spike timing code. There is now growing evidence in behaviors ranging from slow breathing to rapid flight for the prevalence of precise timing encoding in motor systems. Despite this, we largely do not know at what scale timing matters in these circuits due to the difficulty of recording a complete set of spike-resolved motor signals and assessing spike timing precision for encoding continuous motor signals. We also do not know if the precision scale varies depending on the functional role of different motor units. We introduce a method to estimate spike timing precision in motor circuits using continuous MI estimation at increasing levels of added uniform noise. This method can assess spike timing precision at fine scales for encoding rich motor output variation. We demonstrate the advantages of this approach compared to a previously established discrete information theoretic method of assessing spike timing precision. We use this method to analyze the precision in a nearly complete, spike resolved recording of the 10 primary wing muscles control flight in an agile hawk moth, Manduca sexta. Tethered moths visually tracked a robotic flower producing a range of turning (yaw) torques. We know that all 10 muscles in this motor program encode the majority of information about yaw torque in spike timings, but we do not know whether individual muscles encode motor information at different levels of precision. We demonstrate that the scale of temporal precision in all motor units in this insect flight circuit is at the sub-millisecond or millisecond-scale, with variation in precision scale present between muscle types. This method can be applied broadly to estimate spike timing precision in sensory and motor circuits in both invertebrates and vertebrates.

Perturbing the muscle work loop paradigm to unravel the neuromechanics of unsteady locomotion.

Muscle function during movement is more than a simple, linear transformation of neural activity into force. The classic work loop technique has pioneered our understanding of muscle, but typically only characterizes function during unperturbed movement cycles, such as those experienced during steady walking, running, swimming and flying. Yet perturbations away from steady movement often place greater demands on muscle structure and function and offer a unique window into muscle's broader capacity. Recently, studies in diverse organisms from cockroaches to humans have started to grapple with muscle function in unsteady (perturbed, transient and fluctuating) conditions, but the vast range of possible parameters and the challenge of connecting in vitro to in vivo experiments are daunting. Here, we review and organize these studies into two broad approaches that extend the classic work loop paradigm. First, in the top-down approach, researchers record length and activation patterns of natural locomotion under perturbed conditions, replay these conditions in isolated muscle work loop experiments to reveal the mechanism by which muscle mediates a change in body dynamics and, finally, generalize across conditions and scale. Second, in the bottom-up approach, researchers start with an isolated muscle work loop and then add structural complexity, simulated loads and neural feedback to ultimately emulate the muscle's neuromechanical context during perturbed movement. In isolation, each of these approaches has several limitations, but new models and experimental methods coupled with the formal language of control theory give several avenues for synthesizing an understanding of muscle function under unsteady conditions.

The hawkmoth wingbeat is not at resonance.

Flying insects have elastic materials within their exoskeletons that could reduce the energetic cost of flight if their wingbeat frequency is matched to a mechanical resonance frequency. Flapping at resonance may be essential across flying insects because of the power demands of small-scale flapping flight. However, building up large-amplitude resonant wingbeats over many wingstrokes may be detrimental for control if the total mechanical energy in the spring-wing system exceeds the per-cycle work capacity of the flight musculature. While the mechanics of the insect flight apparatus can behave as a resonant system, the question of whether insects flap their wings at their resonant frequency remains unanswered. Using previous measurements of body stiffness in the hawkmoth, Manduca sexta, we develop a mechanical model of spring-wing resonance with aerodynamic damping and characterize the hawkmoth's resonant frequency. We find that the hawkmoth's wingbeat frequency is approximately 80% above resonance and remains so when accounting for uncertainty in model parameters. In this regime, hawkmoths may still benefit from elastic energy exchange while enabling control of aerodynamic forces via frequency modulation. We conclude that, while insects use resonant mechanics, tuning wingbeats to a simple resonance peak is not a necessary feature for all centimetre-scale flapping flyers.

Precise timing is ubiquitous, consistent, and coordinated across a comprehensive, spike-resolved flight motor program.

Sequences of action potentials, or spikes, carry information in the number of spikes and their timing. Spike timing codes are critical in many sensory systems, but there is now growing evidence that millisecond-scale changes in timing also carry information in motor brain regions, descending decision-making circuits, and individual motor units. Across all of the many signals that control a behavior, how ubiquitous, consistent, and coordinated are spike timing codes? Assessing these open questions ideally involves recording across the whole motor program with spike-level resolution. To do this, we took advantage of the relatively few motor units controlling the wings of a hawk moth, Manduca sexta. We simultaneously recorded nearly every action potential from all major wing muscles and the resulting forces in tethered flight. We found that timing encodes more information about turning behavior than spike count in every motor unit, even though there is sufficient variation in count alone. Flight muscles vary broadly in function as well as in the number and timing of spikes. Nonetheless, each muscle with multiple spikes consistently blends spike timing and count information in a 3:1 ratio. Coding strategies are consistent. Finally, we assess the coordination of muscles using pairwise redundancy measured through interaction information. Surprisingly, not only are all muscle pairs coordinated, but all coordination is accomplished almost exclusively through spike timing, not spike count. Spike timing codes are ubiquitous, consistent, and essential for coordination.

A mechanism for sarcomere breathing: volume change and advective flow within the myofilament lattice.

During muscle contraction, myosin motors anchored to thick filaments bind to and slide actin thin filaments. These motors rely on energy derived from ATP, supplied, in part, by diffusion from the sarcoplasm to the interior of the lattice of actin and myosin filaments. The radial spacing of filaments in this lattice may change or remain constant during contraction. If the lattice is isovolumetric, it must expand when the muscle shortens. If, however, the spacing is constant or has a different pattern of axial and radial motion, then the lattice changes volume during contraction, driving fluid motion and assisting in the transport of molecules between the contractile lattice and the surrounding intracellular space. We first create an advective-diffusive-reaction flow model and show that the flow into and out of the sarcomere lattice would be significant in the absence of lattice expansion. Advective transport coupled to diffusion has the potential to substantially enhance metabolite exchange within the crowded sarcomere. Using time-resolved x-ray diffraction of contracting muscle, we next show that the contractile lattice is neither isovolumetric nor constant in spacing. Instead, lattice spacing is time varying, depends on activation, and can manifest as an effective time-varying Poisson ratio. The resulting fluid flow in the sarcomere lattice of synchronous insect flight muscles is even greater than expected for constant lattice spacing conditions. Lattice spacing depends on a variety of factors that produce radial force, including cross-bridges, titin-like molecules, and other structural proteins. Volume change and advective transport varies with the phase of muscle stimulation during periodic contraction but remains significant at all conditions. Although varying in magnitude, advective transport will occur in all cases in which the sarcomere is not isovolumetric. Akin to "breathing," advective-diffusive transport in sarcomeres is sufficient to promote metabolite exchange and may play a role in the regulation of contraction itself.

Rapid frequency modulation in a resonant system: aerial perturbation recovery in hawkmoths.

Centimetre-scale fliers must contend with the high power requirements of flapping flight. Insects have elastic elements in their thoraxes which may reduce the inertial costs of their flapping wings. Matching wingbeat frequency to a mechanical resonance can be energetically favourable, but also poses control challenges. Many insects use frequency modulation on long timescales, but wingstroke-to-wingstroke modulation of wingbeat frequencies in a resonant spring-wing system is potentially costly because muscles must work against the elastic flight system. Nonetheless, rapid frequency and amplitude modulation may be a useful control modality. The hawkmoth Manduca sexta has an elastic thorax capable of storing and returning significant energy. However, its nervous system also has the potential to modulate the driving frequency of flapping because its flight muscles are synchronous. We tested whether hovering hawkmoths rapidly alter frequency during perturbations with vortex rings. We observed both frequency modulation (32% around mean) and amplitude modulation (37%) occurring over several wingstrokes. Instantaneous phase analysis of wing kinematics revealed that more than 85% of perturbation responses required active changes in neurogenic driving frequency. Unlike their robotic counterparts that abdicate frequency modulation for energy efficiency, synchronous insects use wingstroke-to-wingstroke frequency modulation despite the power demands required for deviating from resonance.

Adaptive shifts underlie the divergence in wing morphology in bombycoid moths.

The evolution of flapping flight is linked to the prolific success of insects. Across Insecta, wing morphology diversified, strongly impacting aerodynamic performance. In the presence of ecological opportunity, discrete adaptive shifts and early bursts are two processes hypothesized to give rise to exceptional morphological diversification. Here, we use the sister-families Sphingidae and Saturniidae to answer how the evolution of aerodynamically important traits is linked to clade divergence and through what process(es) these traits evolve. Many agile Sphingidae evolved hover feeding behaviours, while adult Saturniidae lack functional mouth parts and rely on a fixed energy budget as adults. We find that Sphingidae underwent an adaptive shift in wing morphology coincident with life history and behaviour divergence, evolving small high aspect ratio wings advantageous for power reduction that can be moved at high frequencies, beneficial for flight control. By contrast, Saturniidae, which do not feed as adults, evolved large wings and morphology which surprisingly does not reduce aerodynamic power, but could contribute to their erratic flight behaviour, aiding in predator avoidance. We suggest that after the evolution of flapping flight, diversification of wing morphology can be potentiated by adaptative shifts, shaping the diversity of wing morphology across insects.

Wing damage affects flight kinematics but not flower tracking performance in hummingbird hawkmoths.

Wing integrity is crucial to the many insect species that spend distinct portions of their life in flight. How insects cope with the consequences of wing damage is therefore a central question when studying how robust flight performance is possible with such fragile chitinous wings. It has been shown in a variety of insect species that the loss in lift-force production resulting from wing damage is generally compensated by an increase in wing beat frequency rather than amplitude. The consequences of wing damage for flight performance, however, are less well understood, and vary considerably between species and behavioural tasks. One hypothesis reconciling the varying results is that wing damage might affect fast flight manoeuvres with high acceleration, but not slower ones. To test this hypothesis, we investigated the effect of wing damage on the manoeuvrability of hummingbird hawkmoths (Macroglossum stellatarum) tracking a motorised flower. This assay allowed us to sample a range of movements at different temporal frequencies, and thus assess whether wing damage affected faster or slower flight manoeuvres. We show that hummingbird hawkmoths compensate for the loss in lift force mainly by increasing wing beat amplitude, yet with a significant contribution of wing beat frequency. We did not observe any effects of wing damage on flight manoeuvrability at either high or low temporal frequencies.

History-dependent perturbation response in limb muscle.

Muscle mediates movement but movement is typically unsteady and perturbed. Muscle is known to behave non-linearly and with history-dependent properties during steady locomotion, but the importance of history dependence in mediating muscle function during perturbations remains less clear. To explore the capacity of muscles to mitigate perturbations during locomotion, we constructed a series of perturbations that varied only in kinematic history, keeping instantaneous position, velocity and time from stimulation constant. We found that the response of muscle to a perturbation is profoundly history dependent, varying 4-fold as baseline frequency changes, and dissipating energy equivalent to ∼6 times the kinetic energy of all the limbs in 5 ms (nearly 2400 W kg-1). Muscle energy dissipation during a perturbation is predicted primarily by the force at the onset of the perturbation. This relationship holds across different frequencies and timings of stimulation. This history dependence behaves like a viscoelastic memory producing perturbation responses that vary with the frequency of the underlying movement.

Nanometer-scale structure differences in the myofilament lattice spacing of two cockroach leg muscles correspond to their different functions.

Muscle is highly organized across multiple length scales. Consequently, small changes in the arrangement of myofilaments can influence macroscopic mechanical function. Two leg muscles of a cockroach have identical innervation, mass, twitch responses, length-tension curves and force-velocity relationships. However, during running, one muscle is dissipative (a 'brake'), while the other dissipates and produces significant positive mechanical work (bifunctional). Using time-resolved X-ray diffraction in intact, contracting muscle, we simultaneously measured the myofilament lattice spacing, packing structure and macroscopic force production of these muscles to test whether structural differences in the myofilament lattice might correspond to the muscles' different mechanical functions. While the packing patterns are the same, one muscle has 1 nm smaller lattice spacing at rest. Under isometric stimulation, the difference in lattice spacing disappeared, consistent with the two muscles' identical steady-state behavior. During periodic contractions, one muscle undergoes a 1 nm greater change in lattice spacing, which correlates with force. This is the first identified structural feature in the myofilament lattice of these two muscles that shares their whole-muscle dynamic differences and quasi-static similarities.

Integration of parallel mechanosensory and visual pathways resolved through sensory conflict.

The acquisition of information from parallel sensory pathways is a hallmark of coordinated movement in animals. Insect flight, for example, relies on both mechanosensory and visual pathways. Our challenge is to disentangle the relative contribution of each modality to the control of behavior. Toward this end, we show an experimental and analytical framework leveraging sensory conflict, a means for independently exciting and modeling separate sensory pathways within a multisensory behavior. As a model, we examine the hovering flower-feeding behavior in the hawkmoth Manduca sexta In the laboratory, moths feed from a robotically actuated two-part artificial flower that allows independent presentation of visual and mechanosensory cues. Freely flying moths track lateral flower motion stimuli in an assay spanning both coupled motion, in which visual and mechanosensory cues follow the same motion trajectory, and sensory conflict, in which the two sensory modalities encode different motion stimuli. Applying a frequency-domain system identification analysis, we find that the tracking behavior is, in fact, multisensory and arises from a linear summation of visual and mechanosensory pathways. The response dynamics are highly preserved across individuals, providing a model for predicting the response to novel multimodal stimuli. Surprisingly, we find that each pathway in and of itself is sufficient for driving tracking behavior. When multiple sensory pathways elicit strong behavioral responses, this parallel architecture furnishes robustness via redundancy.

Hawkmoth flight in the unsteady wakes of flowers.

Flying animals maneuver and hover through environments where wind gusts and flower wakes produce unsteady flow. Although both flight maneuvers and aerodynamic mechanisms have been studied independently, little is known about how these interact in an environment where flow is already unsteady. Moths forage from flowers by hovering in the flower's wake. We investigated hawkmoths tracking a 3D-printed robotic flower in a wind tunnel. We visualized the flow in the wake and around the wings and compared tracking performance with previous experiments in a still-air flight chamber. As in still air, moths flying in the flower wake exhibit near-perfect tracking at the low frequencies at which natural flowers move. However, tracking in the flower wake results in a larger overshoot between 2 and 5 Hz. System identification of flower tracking reveals that moths also display reduced-order dynamics in wind compared with still air. Smoke visualization of the flower wake shows that the dominant vortex shedding corresponds to the same frequency band as the increased overshoot. Despite these large effects on tracking dynamics in wind, the leading edge vortex (LEV) remains bound to the wing throughout the wingstroke and does not burst. The LEV also maintains the same qualitative structure seen in steady air. Persistence of a stable LEV during decreased flower tracking demonstrates the interplay between hovering and maneuvering.

Dual dimensionality reduction reveals independent encoding of motor features in a muscle synergy for insect flight control.

What are the features of movement encoded by changing motor commands? Do motor commands encode movement independently or can they be represented in a reduced set of signals (i.e. synergies)? Motor encoding poses a computational and practical challenge because many muscles typically drive movement, and simultaneous electrophysiology recordings of all motor commands are typically not available. Moreover, during a single locomotor period (a stride or wingstroke) the variation in movement may have high dimensionality, even if only a few discrete signals activate the muscles. Here, we apply the method of partial least squares (PLS) to extract the encoded features of movement based on the cross-covariance of motor signals and movement. PLS simultaneously decomposes both datasets and identifies only the variation in movement that relates to the specific muscles of interest. We use this approach to explore how the main downstroke flight muscles of an insect, the hawkmoth Manduca sexta, encode torque during yaw turns. We simultaneously record muscle activity and turning torque in tethered flying moths experiencing wide-field visual stimuli. We ask whether this pair of muscles acts as a muscle synergy (a single linear combination of activity) consistent with their hypothesized function of producing a left-right power differential. Alternatively, each muscle might individually encode variation in movement. We show that PLS feature analysis produces an efficient reduction of dimensionality in torque variation within a wingstroke. At first, the two muscles appear to behave as a synergy when we consider only their wingstroke-averaged torque. However, when we consider the PLS features, the muscles reveal independent encoding of torque. Using these features we can predictably reconstruct the variation in torque corresponding to changes in muscle activation. PLS-based feature analysis provides a general two-sided dimensionality reduction that reveals encoding in high dimensional sensory or motor transformations.

Sensory processing within cockroach antenna enables rapid implementation of feedback control for high-speed running maneuvers.

Animals are remarkably stable during high-speed maneuvers. As the speed of locomotion increases, neural bandwidth and processing delays can limit the ability to achieve and maintain stable control. Processing the information of sensory stimuli into a control signal within the sensor itself could enable rapid implementation of whole-body feedback control during high-speed locomotion. Here, we show that processing in antennal afferents is sufficient to act as the control signal for a fast sensorimotor loop. American cockroaches Periplaneta americana use their antennae to mediate escape running by tracking vertical surfaces such as walls. A control theoretic model of wall following predicts that stable control is possible if the animal can compute wall position (P) and velocity, its derivative (D). Previous whole-nerve recordings from the antenna during simulated turning experiments demonstrated a population response consistent with P and D encoding, and suggested that the response was synchronized with the timing of a turn executed while wall following. Here, we record extracellularly from individual mechanoreceptors distributed along the antenna and show that these receptors encode D and have distinct latencies and filtering properties. The summed output of these receptors can be used as a control signal for rapid steering maneuvers. The D encoding within the antenna in addition to the temporal filtering properties and P dependence of the population of afferents support a sensory-encoding notion from control theory. Our findings support the notion that peripheral sensory processing can enable rapid implementation of whole-body feedback control during rapid running maneuvers.

Nanometer scale difference in myofilament lattice structure of muscle alter muscle function in a spatially explicit model.

Crossbridge binding, state transitions, and force in active muscle is dependent on the radial spacing between the myosin-containing thick filament and the actin-containing thin filament in the filament lattice. This radial lattice spacing has been previously shown through spatially explicit modeling and experimental efforts to greatly affect quasi-static, isometric, force production in muscle. It has recently been suggested that this radial spacing might also be able to drive differences in mechanical function, or net work, under dynamic oscillations like those which occur in muscles in vivo. However, previous spatially explicit models either had no radial spacing dependence, meaning the lattice spacing could not be investigated, or did include radial spacing dependence but could not reproduce in vivo net work during dynamic oscillations and only investigated isometric contractions. Here we show the first spatially explicit model to include radial crossbridge dependence which can produce mechanical function similar to real muscle. Using this spatially explicit model of a half sarcomere, we show that when oscillated at strain amplitudes and frequencies like those in the hawk moth Manduca sexta, mechanical function (net work) does depend on the lattice spacing. In addition, since the trajectory of lattice spacing changes during dynamic oscillation can vary from organism to organism, we can prescribe a trajectory of lattice spacing changes in the spatially explicit half sarcomere model and investigate the extent to which the time course of lattice spacing changes can affect mechanical function. We simulated a half sarcomere undergoing dynamic oscillations and prescribed the Poisson's ratio of the lattice to be either 0 (constant lattice spacing) or 0.5 (isovolumetric lattice spacing changes). We also simulated net work using lattice spacing data taken from Manduca sexta which has a variable Poisson's ratio. Our simulation results indicate that the lattice spacing can change the mechanical function of muscle, and that in some cases a 1 nm difference can switch the net work of the half sarcomere model from positive (motor-like) to negative (brake-like).

The Weis-Fogh number describes resonant performance tradeoffs in flapping insects.

Dimensionless numbers have long been used in comparative biomechanics to quantify competing scaling relationships and connect morphology to animal performance. While common in aerodynamics, few relate the biomechanics of the organism to the forces produced on the environment during flight. We discuss the Weis-Fogh number, N, as a dimensionless number specific to flapping flight, which describes the resonant properties of an insect and resulting tradeoffs between energetics and control. Originally defined by Torkel Weis-Fogh in his seminal 1973 paper, N measures the ratio of peak inertial to aerodynamic torque generated by an insect over a wingbeat. In this perspectives piece, we define N for comparative biologists and describe its interpretations as a ratio of torques, and as the width of an insect's resonance curve. We then discuss the range of N realized by insects and explain the fundamental tradeoffs between an insect's aerodynamic efficiency, stability, and responsiveness that arise as a consequence of variation in N, both across and within species. N is therefore an especially useful quantity for comparative approaches to the role of mechanics and aerodynamics in insect flight.

Why animals can outrun robots.

Animals are much better at running than robots. The difference in performance arises in the important dimensions of agility, range, and robustness. To understand the underlying causes for this performance gap, we compare natural and artificial technologies in the five subsystems critical for running: power, frame, actuation, sensing, and control. With few exceptions, engineering technologies meet or exceed the performance of their biological counterparts. We conclude that biology's advantage over engineering arises from better integration of subsystems, and we identify four fundamental obstacles that roboticists must overcome. Toward this goal, we highlight promising research directions that have outsized potential to help future running robots achieve animal-level performance.