AirPulse: Ultrasound & Flapping-Wing Systems

• AirPulse is a dual-technology platform that integrates an ultrasound-driven air jet with a butterfly-inspired flapping-wing MAV to manipulate airflow with high precision.
• The ultrasound system uses a phased-array of 249 transducers to synthesize a steerable Bessel beam, achieving sub-centimeter resolution and near-instantaneous (<1 ms) response.
• The flapping-wing MAV employs compliant wings with STAR modulation for dynamic, tailless flight, offering stable maneuvers and improved power loading.

AirPulse denotes two distinct state-of-the-art technological platforms: (1) an electronically steerable, ultrasound-driven air jet based on acoustic streaming in a Bessel-beam field, and (2) a lightweight, butterfly-inspired, flapping-wing micro air vehicle (FWMAV) that demonstrates autonomous tailless flight. Each implementation advances the manipulation and control of airflow through either non-contact ultrasound actuation or embodied flapping biomechanics, with deep relevance to fluid dynamics, robotics, and biologically inspired design.

1. Ultrasound-Driven AirPulse: Physical Principles

The ultrasound-based AirPulse system generates a “stretching air flow” by producing a Bessel beam in air using an active phased array of acoustic transducers, enabling remote, non-contact flow generation (Hasegawa et al., 2017). The theoretical foundation is steady nonlinear acoustic streaming: a unidirectional mean flow induced by viscous absorption of high-intensity ultrasound. The pressure field of a lossless Bessel beam is expressed as

$p(r,z) = p_0 J_0(k_t r) e^{i k_z z}$

where $p_0$ is the amplitude, $J_0$ the zeroth-order Bessel function, $k_t$ and $k_z$ the transverse and axial wave-numbers with $k_t^2 + k_z^2 = k^2 = (\omega/c)^2$, and $(r, z)$ the radial and axial coordinates. The axial streaming velocity $v_s$ is given by

$v_s = \frac{2 \beta I}{\rho c^2}$

with $\beta$ the absorption coefficient, $p_0$0 the acoustic intensity, $p_0$1 ambient density, and $p_0$2 the speed of sound. This formulation quantifies the relationship between ultrasound excitation and resultant airflow.

2. Ultrasound Phased-Array Design and Beam Steering

The array consists of four identical modules, each containing $p_0$3 disc-type transducers resonant at $p_0$4 kHz, arranged on a sub-wavelength grid (8.5 mm pitch) over a $p_0$5 planar aperture (Hasegawa et al., 2017). Individual continuous-wave phase control enables electronic synthesis of a conical (axicon) wavefront, producing a collimated Bessel beam whose diameter is determined by the virtual cone’s half-angle $p_0$6. Steering is accomplished by digitally controlling the phase delay $p_0$7 at each element, with the signed delay $p_0$8. The beam’s axis can be tilted to any unit vector $p_0$9 via a Rodrigues’ rotation. Real-time re-targeting of both beam position and direction is possible within ±30° cone angle, at kHz rates; sub-centimeter scale targeting is enabled by the high element density.

3. Ultrasound AirPulse: Experimental Characterization and Metrics

A 3-axis robot arm actuates a hot-wire anemometer through three spatial planes, mapping the generated airflow. For a non-tilted beam at $J_0$0, the system achieves peak streaming velocity $J_0$1 m/s at $J_0$2 mm, with the highest velocity region (“fastest spot”) lying apart from the array surface. Radial confinement at this point yields half-maximal velocity within 3 mm, demonstrating high spatial resolution. The flow response is near-instantaneous (microsecond–millisecond), determined by electronic signal timing. Steering at $J_0$3 preserves performance while redirecting the flow. The useful streaming range spans 200–600 mm, with measurable flows to 1.5 m. Core limitations include the need for high ultrasonic power and the risk of grating lobes if element spacing exceeds $J_0$4 or with excessive steering (Hasegawa et al., 2017).

Metric	Ultrasound AirPulse Value	Notes
Peak axial velocity	0.5–0.6 m/s @ 400 mm	0.55 m/s typical
Radial resolution	≈6 mm FWHM	At peak plane ($J_0$5 mm)
Steering agility	±30° cone, kHz update rates	Arbitrary, sub-ms redirection
Response time	≲1 ms	Electronically determined
Required power	200 W (4 × 50 W units)	High-power ultrasound driving

4. Ultrasound AirPulse: Applications, Advantages, and Limitations

Remote, electronically steerable flow generation allows for localized cooling, particle or aerosol delivery/extraction, microfluidic manipulation, VR/AR haptic feedback, and acoustic cleaning or drying. Advantages include the lack of moving parts, real-time agility, deep sub-centimeter resolution, and ultra-fast response. Limitations involve hardware power requirements, the exclusive measurement of axial flow magnitude (lacking full vector mapping), and peak velocities that are lower than those of conventional fans. These characteristics open new directions in non-contact fluid transport, targeted aerodynamic manipulation, and controlled atmospheric delivery technologies (Hasegawa et al., 2017).

5. Butterfly-Inspired AirPulse Robot: Morphology and Mechanisms

The AirPulse FWMAV is a 26-gram, tailless butterfly-inspired robot employing two compliant wings with carbon-fiber reinforcement and a 3D-printed PETG fuselage (Gu et al., 6 Feb 2026). The wingspan is $J_0$6 mm, mean chord $J_0$7 mm, with AR $J_0$8, and very low wing loading $J_0$9 N/m². This matches biological reference values for Papilio species. Dihedral is fixed at 50° to promote inter-wing coupling and body undulation—a crucial mechanism for dynamic pitch stabilization. The spatial gradient in wing membrane stiffness (rigid costal/basal, compliant distal) enables both torque transfer and passive feathering under cyclic loads.

6. Flapping-Wing Kinematics, STAR, and Control Architecture

The system’s stroke dynamics are governed by

$k_t$0

with $k_t$1 (amplitude), flapping frequency $k_t$2, and $k_t$3 (stroke-plane tilt, for pitch) being independently controllable. The Stroke Timing Asymmetry Rhythm (STAR) generator produces time-asymmetric wingstrokes through a phase-domain modulation $k_t$4, with $k_t$5 ensuring continuous, bounded trajectories. STAR maintains mean flapping frequency while enabling direct, linear control over down/upstroke durations. The wings’ inertial dynamics produce moment of inertia $k_t$6 oscillations by up to 2.5× per beat, leading to pronounced ±70° body pitch undulations.

Onboard sensing employs an ICM-42688-P IMU (1 kHz), magnetometer BMM350 (400 Hz), and BMP390L barometer (200 Hz), all co-located with the CG for minimal lever arms. State estimation uses a Madgwick gradient-descent filter for Euler angle tracking ($k_t$7 drift), and a recursive least squares (RLS) model decouples undulatory pitch from mean pitch, feeding the latter to a PID controller. Pitch and yaw are regulated via symmetric and antisymmetric modifications of offset and STAR parameters.

7. AirPulse Robot: Flight Performance and Applications

Free-flight experiments demonstrate stable climbing and turning via either angle-offset or stroke-timing modulation, with average power consumption of ≈5.9 W and a power loading of 4.4 g/W (significantly lower than ~10 g/W for micro-quadrotors) (Gu et al., 6 Feb 2026). In climbing, $k_t$8 achieves a ~15° climb angle; $k_t$9 achieves ~30°, with tracking errors below 2°. STAR-based pitch control yields similar rates and power. Yaw turns (e.g., $k_z$0 or corresponding STAR timings) produce ~50°/s turn rates and ~0.4 m radius, with roll excursions under 5°, confirming stability-through–dynamic-coupling.

Applications include confined-space inspection, ecological monitoring, and experimental platforms for modeling butterfly-like aerial mechanics. The aerodynamic regime—$k_z$1, reduced frequency $k_z$2—supports highly unsteady, vortex-mediated lift generation and efficient maneuvering through complex, cluttered environments.

Metric	Butterfly-Inspired AirPulse	Notes
Mass	26 g	Including all avionics
Wingspan	60 mm
Power loading	4.4 g/W	Micro-quadrotors: ~10 g/W
Max pitch undulation	±70° per beat	Measured
Roll excursion	<5° during turning
Min turn radius	~0.4 m

8. Biomechanical and Fluid-Mechanics Implications

The FWMAV platform demonstrates the efficacy of inertial–aerodynamic coupling for flight stability without tail surfaces. The mapping of flapping modulation to resultant force–torque vector supports closed-loop, untethered flight, and provides a physical model for research into butterfly flight regimes. The spatial compliance tuning, low wing loading, and STAR-based kinematic asymmetry collectively enable robust, quasi-silent, biologically relevant flight in indoor and natural environments. This suggests that dynamic coupling between structure, kinematics, and feedback—a hallmark of biological flyers—can be systematically engineered and generalized across micro air vehicle scales (Gu et al., 6 Feb 2026).