Ultrasound Midair Haptics: A Technology Overview

• Ultrasound Midair Haptics is a contactless haptic technology that creates perceivable forces on the skin using focused ultrasound beams.
• It employs phased-array transducers and electronic beamforming to dynamically modulate spatiotemporal tactons with high precision.
• UMH enables real-time, adaptive haptic feedback in XR, HCI, and robotics while integrating closed-loop control and advanced modulation techniques.

Ultrasound Midair Haptics (UMH) refers to the generation of tactile sensations in free space by focusing and modulating ultrasound fields to exert perceivable forces on bare skin, enabling contactless haptic feedback. Utilizing electronic beamforming of phased-array transducers, UMH produces dynamic, spatiotemporally controlled “tactons” (tactile icons), rendering a wide gamut of vibrotactile, kinesthetic, and thermal cues for applications across Extended Reality (XR), Human-Computer Interaction (HCI), robotic manipulation, and multisensory communication. Technical frameworks now support real-time, adaptive, closed-loop integration with XR environments and user interfaces, providing scalable avenues for expressive, programmable haptic interactions.

1. Physical Principles and System Architectures

At its core, UMH exploits acoustic radiation pressure arising from focused ultrasound beams in air, delivered by arrays of piezoelectric transducers typically resonant at 40 kHz. The net time-averaged force per unit area on the skin is governed by

$$P_{\rm rad} = \frac{2\langle I\rangle}{c} = \frac{\langle p^2\rangle}{\rho c}$$

where $\langle I\rangle$ denotes acoustic intensity, $c$ sound speed, and $\rho$ air density (Alakhawand et al., 2022, Fleig, 8 Dec 2025). Individual transducer elements, arranged as dense phased arrays (e.g., 16×16×1.5 cm, 324 channels (Vaquero-Melchor et al., 13 Jan 2025)), are phase-shifted to synchronize ultrasonic wavefronts at arbitrary spatial foci. The focal spot diameter is diffraction-limited, typically 5–10 mm, with intensity decaying rapidly away from focus. Hardware configurations scale from desktop arrays for localized interaction to robotically actuated or flexurally vibrated plates for large-area and room-scale rendering (Faridan et al., 2022, Hasegawa et al., 2024).

2. Beamforming, Modulation, and Rendering Algorithms

UMH haptic field synthesis employs real-time phased-array beamforming. Element drive phases $\phi_i$ (for element at position $r_i$, focus at $r_f$) are calculated as

$\phi_i = k \| r_i - r_f \| + \phi_0$

with $k=2\pi/\lambda$ acoustic wavenumber (John et al., 2024, Morisaki et al., 2023). Real-time control supports rapid (<1 ms) updates for dynamic focal-point steering, enabling both static (“fixed”) and adaptive (“runtime-modifiable”) tactons. Tactons are specified as spatiotemporal trajectories, $x(t)$, in 3D space mapped per interaction logic (Lim et al., 2024). Vibration is created on skin by amplitude modulation (AM) of the ultrasound envelope (typically 30–200 Hz), exciting FA-I/FA-II mechanoreceptors, while static pressure cues utilize low-frequency focus rotation to target SA-I afferents (Morisaki et al., 23 Jan 2026).

Spatiotemporal modulation (STM) can be used to “draw” shapes, textures, and trajectories, with update rates reaching 500–1,200 kHz in advanced engines (John et al., 2024). Dynamic adaptation is achieved using closed-loop input (e.g., hand-tracking, environmental sensing), runtime parameter mapping, and conditional logic for closed-loop feedback.

3. Adaptive Design Spaces and Software Toolkits

Contemporary UMH toolkits (e.g., AdapTics (John et al., 2024)) encode a five-dimensional adaptation space:

• Granularity: Tacton/global transformation or per-keyframe adaptation.
• Timing: Playback speed and sequence control.
• Spatial Configuration: Position, scale, and rotation.
• Feel: Amplitude envelope and modulation frequency.
• Transformation Type: Continuous parameter mapping or conditional triggers.

Authoring is supported by web GUIs for keyframe and parameter assignment; real-time engines (Rust/C++/Unity) take external variables (e.g., hand pose, game state) via API callbacks, continuously evaluating and pushing updated phase/amplitude batches at each array-refresh tick (20–40 kHz) (John et al., 2024).

For high-dimensional parameter tuning, Sequential Line Search (SLS) methods reduce multi-dimensional spaces to intuitive 1-D slider controls, while real-time visualization (e.g., “growing grass” metaphors) synchronize haptic state with visual feedback in Unity (Zhang et al., 2024).

Modulation Technique	Primary Sensation	Mechanoreceptor Targeted
AM @ 150 Hz	Vibratory	FA-II
AM @ 30 Hz	Flutter	FA-I
STM (rotation @ 5 Hz)	Static Pressure	SA-I
Combined AM/STM	Texture synthesis	Multiple

4. Quantitative Psychophysics and Evaluation

Empirical studies reveal threshold discrimination for focal forces at ≈4 mN (200 Hz modulation) with lateral resolution ≈6–13 mm (Hashizume et al., 2019, Alakhawand et al., 2022). Adaptive, multiband haptic stimuli achieve six discriminable “textures”, spanning smooth/slippery (pressure-only; indistinguishable from glass marble) to rough/grippy (150 Hz vibration; matches 100-grit sandpaper) (Morisaki et al., 23 Jan 2026). User studies (Creativity Support Index, N=12) demonstrate that adaptive tactons significantly increase expressiveness and exploration scores over fixed patterns (e.g., CSI: adaptive = 7.79, non-adaptive = 6.76, p = .015) (John et al., 2024). Latencies in closed-loop systems are maintained under 2–5 ms, well within perceptual thresholds for real-time interactivity.

Simulators and mapping robots automate tactile intensity field measurement: biomimetic sensors and laser vibrometry enable rapid mapping and validation, with cost and speed advantages over optical setups (Alakhawand et al., 2022, Lim et al., 2024).

5. Integration with XR, AR, and Multimodal Systems

UMH system architectures are built for seamless integration in XR/AR frameworks (Unity/Unreal/HoloLens), supporting synchronized rendering of haptic fields, volumetric meshes, and gesture-based logic (Vaquero-Melchor et al., 13 Jan 2025, Romanus et al., 2020). Focal-point alignment between virtual and real workspace is achieved by spatial registration (manual marker-based or automatic SLAM) and real-time hand tracking (Leap Motion, Intel RealSense).

Applications include shape perception, button pressing, resizing, and push interactions. Quantitative user studies indicate mid-air haptics improve form identification and depth estimation in resizing tasks, while noting challenges in user interface metaphors and the persistent reliance on visual anchors (Vaquero-Melchor et al., 13 Jan 2025).

Thermal and vibrotactile cues, extended to “hot warning” and “cold” feedback, are realized via deliberate control of ultrasonic irradiation and sound-absorbing gloves (Kamigaki et al., 2020).

6. Scalability, Alternative Focusing, and Limitations

Phased-array devices are currently limited by physical aperture (view range ~30–50 cm), hardware cost, and per-focus intensity (max ≈0.5 N, subject to safety standards) (Fleig, 8 Dec 2025). Alternatives using flexurally vibrating plates with Fresnel-inspired amplitude masks deliver focused beams over large scales with low cost and fabrication complexity, though they lack dynamic reconfigurability (Hasegawa et al., 2024). Large-area coverage is also achieved by robotically actuated ultrasound modules (UltraBots), addressing workspace constraints in VR and interactive installations (Faridan et al., 2022).

Latency, spatial resolution (limited by diffraction and element spacing), and intensity safety remain open technical bottlenecks. Ghost foci and artifacts from multi-focus beamforming pose perceptual challenges requiring further optimization in array geometry and pattern design.

7. Future Directions and Research Opportunities

Key research trajectories include shrinking focal spots below current limits (for finer texture rendering), deep-learning–based ultra-fast beamforming, multi-user spatial addressing, and fusion with other non-contact haptic modalities (air jets, thermal, electrical). Physiological sensor integration (heart rate, EDA) for bio-responsive haptic experiences is emerging (Romanus et al., 2020), as is cross-modal perceptual optimization (audio–haptic, visuo–haptic) (Fleig, 8 Dec 2025).

The field continues to advance via interdisciplinary collaboration in applied physics, real-time optimization, materials science, human factors, and design. A plausible implication is that continued improvements in scalable hardware, multimodal systems integration, and adaptive interaction paradigms will further expand the immersive and informational capacity of ultrasound midair haptic technologies.