AirTouch: Air-Based Contactless Interaction
- AirTouch is a contactless interaction paradigm that uses directed airflow and infrared marker tracking to create tactile safety barriers in human-robot interaction.
- The prototype integrates a UR10 manipulator with an IR camera and impeller-based airflow to maintain safe distances, demonstrating statistically significant perceptual discrimination in experiments.
- AirTouch also spans mid-air haptics and drone edge detection, reflecting a broad design space where air functions as both a tactile feedback channel and a sensing medium.
to=arxiv_search.search 彩神争霸大发快三 json
{"query":"AirTouch arXiv", "max_results": 10, "sort_by":"relevance"}{"query":"\"AirTouch\" arXiv human robot air pressure IR mocap", "max_results": 10, "sort_by":"relevance"}{"query":"\"AirTouch\" aviation mid-air haptics arXiv", "max_results": 10, "sort_by":"relevance"}{"query":"\"AirTouch\" drone ground effect arXiv", "max_results": 10, "sort_by":"relevance"}{"query":"mid-air haptics ultrasound tactile arXiv AirTouch related", "max_results": 10, "sort_by":"relevance"}{"query":"air vortex haptic arXiv DHH attention air touch", "max_results": 10, "sort_by":"relevance"}{"query":"on-body touch AR headset cameras arXiv EgoTouch", "max_results": 10, "sort_by":"relevance"}{"query":"touchscreen tactile feedback moving vehicles arXiv FITS", "max_results": 10, "sort_by":"relevance"}{"query":"AirWare in-air hand gesture arXiv", "max_results": 10, "sort_by":"relevance"}{"query":"AirPen touchless fingertip gesture arXiv", "max_results": 10, "sort_by":"relevance"}天天啪 to=arxiv_search.search 大发快三是什么json
{"query":"\"AirTouch: Towards Safe Human-Robot Interaction Using Air Pressure Feedback and IR Mocap System\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"Mid-Air Haptics in Aviation -- creating the sensation of touch where there is nothing but thin air\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"Flight Dynamics to Sensing Modalities: Exploiting Drone Ground Effect for Accurate Edge Detection\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"Tap tactile presentation by airborne ultrasound\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"SHITARA: Sending Haptic Induced Touchable Alarm by Ring-shaped Air vortex\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"AirWare: Utilizing Embedded Audio and Infrared Signals for In-Air Hand-Gesture Recognition\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"AirPen: A Touchless Fingertip Based Gestural Interface for Smartphones and Head-Mounted Devices\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"EgoTouch: On-Body Touch Input Using AR/VR Headset Cameras\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"EclipseTouch: Touch Segmentation on Ad Hoc Surfaces using Worn Infrared Shadow Casting\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"Airborne Ultrasonic Tactile Display Brain-computer Interface Paradigm\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"FITS: Ensuring Safe and Effective Touchscreen Use in Moving Vehicles\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"AiR -- An Augmented Reality Application for Visualizing Air Pollution\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"Touch-And-Guard\" touch resonance wearable pairing arXiv", "max_results": 5, "sort_by":"relevance"}{"query":"\"AirGuard -- Protecting Android Users From Stalking Attacks By Apple Find My Devices\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"Airborne Ultrasonic Tactile Display BCI\"", "max_results": 5, "sort_by":"relevance"}{"query":"\"InfraredTags\" reflective AprilTag 850 nm arXiv", "max_results": 5, "sort_by":"relevance"} to=arxiv_search.search 】【。】【”】【json
{"query":"human-robot interaction airflow tactile safety barrier arXiv 2023 IR camera AprilTag", "max_results": 10, "sort_by":"relevance"}{"query":"ultrasonic mid-air haptics tactile hologram aviation arXiv 2020", "max_results": 10, "sort_by":"relevance"}{"query":"ground effect sensing edge detection drone IMU motor arXiv 2025", "max_results": 10, "sort_by":"relevance"}{"query":"contactless tactile haptics palm phased array ultrasound arXiv 2014 2015", "max_results": 10, "sort_by":"relevance"}{"query":"AR on-body touch RGB headset cameras arXiv 2025", "max_results": 10, "sort_by":"relevance"}{"query":"ad hoc surface touch infrared shadow casting arXiv 2025", "max_results": 10, "sort_by":"relevance"}{"query":"in-air gesture smartphone audio infrared arXiv 2021", "max_results": 10, "sort_by":"relevance"}{"query":"touchless fingertip smartphone head-mounted devices arXiv 2019", "max_results": 10, "sort_by":"relevance"}{"query":"air vortex ring haptic alarm DHH arXiv 2023", "max_results": 10, "sort_by":"relevance"}{"query":"moving vehicles touchscreen tactile feedback BDFT arXiv 2024", "max_results": 10, "sort_by":"relevance"} 天天中彩票有 to=arxiv_search.search 大发娱乐 天天中彩票任选json
{"query":"AirTouch human-robot infrared camera wearable marker airflow potential field", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch safe human-robot interaction air pressure feedback IR mocap system", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch touchless tactile cockpit interaction ultrasound hand tracking", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch drone edge detection ground effect IMU motor signals", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch tactile air pressure barrier urban robot public spaces", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch tactile warning barrier robot shared spaces arXiv", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch focused ultrasound tactile hologram arXiv aviation", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch haptic virtual buttons dials ultrasound aviation arXiv", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch safety barrier airflow potential field robot public urban spaces arXiv 2308.00096", "max_results": 10, "sort_by":"relevance"}{"query":"AirTouch drone sensing by touching with air arXiv 2509.21085", "max_results": 10, "sort_by":"relevance"}
AirTouch is a term used in arXiv literature for several contactless interaction paradigms in which air is treated as an active medium rather than a passive background. In its most specific and experimentally grounded use, it denotes a human-robot safety system that combines infrared marker tracking with directed airflow to warn a person who is entering a hazardous robot workspace [2308.00096]. The same term has also been used for touchless tactile cockpit interaction based on gesture recognition and mid-air haptics in aviation [2001.01445], and for a drone edge-detection method that exploits ground effect as a sensing modality [2509.21085]. Across these uses, air functions either as a feedback channel or as a sensing medium. This suggests a broader research motif rather than a single standardized framework.
1. AirTouch as a human-facing safety layer for robots
The 2023 AirTouch system was proposed for human-robot interaction in public or urban environments where people may be distracted, inattentive, or unaware of robot motion. Its motivating example is an autonomous robot that charges electric vehicles in a shared environment. In that setting, conventional safety measures are described as primarily robot-side: slowing, scaling trajectories, or stopping when a person gets too close. AirTouch was introduced as a complementary human-facing layer that actively informs the person rather than only constraining the robot [2308.00096].
The central design claim is that visual or auditory warnings can fail in public environments because people may not be looking at the robot, may be distracted, or may be wearing headphones. AirTouch therefore uses directed airflow as a non-contact tactile cue. When a person approaches a dangerous region, the system renders an airflow-defined safety barrier intended to signal hazardous proximity before physical contact occurs. The paper frames this barrier as an “airflow potential field,” but the field is presented conceptually rather than through a formal control law [2308.00096].
A second motivation concerns sensing robustness. The system uses an infrared approach because it is presented as more robust than visible-light cameras under changing illumination and capable of operating in low-light or nighttime settings. Compared with depth cameras, the monocular IR camera plus reflective marker is described as lower-cost and easier to integrate into outdoor environments, although this comes with the explicit assumption that the user wears a marker and that the camera maintains line of sight to it [2308.00096].
2. Prototype architecture and control logic
The AirTouch prototype combines an infrared marker-based tracking subsystem with an airflow feedback subsystem mounted on a UR10 collaborative manipulator [2308.00096].
| Subsystem | Implementation | Function |
|---|---|---|
| Tracking | Monocular IR camera + IR-visible AprilTag wrist marker | Human localization |
| Airflow feedback | Impeller-based airflow generator on robot | Tactile warning barrier |
| Control software | Python on PC + Arduino + Raspberry Pi | Distance check and actuation |
Human position is tracked using a monocular IR camera and an IR-visible AprilTag marker. The marker is printed on reflective tape and embedded under an 850 nm IR filter, following prior “InfraredTags”-style ideas cited by the authors. In the prototype it is worn as a fabric wristband, although the paper states that it could be integrated into ordinary clothing as a decorative element. The imaging hardware consists of a Raspberry Pi Camera Module 2 NoIR augmented with eight 850 nm IR LEDs, with image processing running on a Raspberry Pi 4 B. The reported time to estimate marker pose from a raw image is (30 \pm 2) ms [2308.00096].
The feedback mechanism is an impeller-based airflow generator. The prototype uses an FMS ducted fan jet EDF unit (11-blade) with a 2840 KV3900 motor, driven by a SimonK 30A electronic speed controller and commanded by an Arduino Uno. The impeller creates a directed stream of air that acts as a physical or tactile warning barrier around dangerous robot regions. The figures depict the impeller’s effective working zones as ellipses around the manipulator, but the paper does not provide an explicit mathematical potential-field equation, force law, or airflow control function [2308.00096].
The control loop is threshold-based. Software is written primarily in Python on a PC, handling image acquisition, marker pose estimation, robot control, and communication with the Arduino, and Python’s multiprocessing library is used to reduce processing time and activation delays. The system computes the distance between the robot Tool Center Point (TCP) and the tracked marker position. If this distance falls below a preset haptic activation distance (HAD), the impeller is activated. In the main study, the HAD is (0.35) m, while a smaller dangerous-proximity distance is defined as (0.25) m. The article does not specify continuous modulation of airflow intensity as a function of distance, directional airflow control beyond physical impeller placement, localization equations, coordinate-transform formulas, or a full latency budget beyond marker-pose computation time [2308.00096].
3. Experimental studies and quantitative findings
The empirical evaluation consists of two studies: a preliminary airflow-perception study and a human-robot interaction study in an inattentive-person scenario [2308.00096].
| Study | Conditions | Key result |
|---|---|---|
| Study I | Perception of 0.25 m vs 0.35 m airflow distances | Significant discrimination, (T=-6.37,\ p<.001) |
| Study II | Visual-only (V) vs visual + airflow (VA) | Safer mean distance in VA, (T=-3.52,\ p=.006) |
In Study I, ten right-handed participants, including 2 female participants, wore the marker wristband and an opaque mask so that only airflow could be used. The impeller ran at a constant rotation regime. Participants were trained on two reference distances from the impeller, (0.25) m and (0.35) m, and then asked to reposition the hand to those locations based only on felt airflow. Each participant completed 20 attempts total, 10 per distance. The absolute mean error was (0.035 \pm 0.025) m for (0.25) m and (0.051 \pm 0.035) m for (0.35) m. Shapiro-Wilk normality tests gave (p=.58) and (p=.14), and the paired-samples t-test found a statistically significant difference between perceived positions: (T=-6.37,\ p<.001). The reported interpretation is that the airflow cue was strong and discriminable enough to render a tactile safety boundary [2308.00096].
Study II evaluates AirTouch in a more realistic imitation of an inattentive person near an autonomous EV charging robot. Again, 10 right-handed participants, including 2 female participants, were tested. The participant collected screwdriver bits into a toolbox while sharing workspace with the UR10, which simulated the charging plug-in operation. Two conditions were compared: visual feedback only (V) and visual plus air-pressure feedback (VA). In the VA condition, the impeller activated when the wrist marker crossed the (0.35) m HAD relative to the robot TCP. To simulate inattentiveness, participants wore headphones playing white noise and were asked to attend to a marker on a screen as a distraction. Each participant performed one trial per condition after training [2308.00096].
The dependent variable was the participant’s distance to the robot TCP. For analysis, samples above the HAD were discarded and the remaining near-robot samples were averaged per participant. The average distance in the dangerous-proximity regime was (0.307 \pm 0.014) m for V and (0.326 \pm 0.010) m for VA. Shapiro-Wilk tests gave (p=.75) and (p=.72), and a paired-sample t-test found the increase in distance with AirTouch statistically significant: (T=-3.52,\ p=.006). Since dangerous proximity was defined as (0.25) m and activation as (0.35) m, the airflow cue shifted behavior in the safer direction. In absolute terms, the mean distance increased by about (1.9) cm [2308.00096].
4. Limitations, misconceptions, and proof-of-concept status
A recurrent misconception is to read the phrase “airflow potential field” as if the paper presented a formalized potential-field controller. It does not. The 2023 AirTouch article provides a conceptual depiction of a tactile barrier, but no explicit potential-field equation, force law, or graded airflow-control function [2308.00096].
A second misconception is that the system is device-free. The prototype depends on a wearable IR marker, worn as a wristband in the reported experiments. The authors argue that the marker could be hidden in clothing and made natural-looking, but the system still assumes marker wear and camera visibility. Occlusion, marker orientation, and line-of-sight constraints remain implicit vulnerabilities [2308.00096].
The paper also does not characterize outdoor robustness to wind, weather, turbulence, or ambient environmental airflow, even though the intended deployment context is urban and public. The control logic remains simple threshold activation rather than a fully modeled field with graded intensity or multidirectional shaping. Experimental generalization is limited by the small sample size and by the fact that the second study is a simplified laboratory imitation rather than a real deployment. The paper also does not analyze robot speed effects, multi-person interactions, whole-body human motion, or cases in which a person approaches with a body part other than the marker-bearing wrist [2308.00096].
For those reasons, the article is best read as an initial proof of concept. Its conclusion is feasibility rather than formal closure: airflow can be used as a tactile safety channel, and the combination of IR marker tracking with non-contact air-pressure cues can improve maintenance of safer distances around a working robot. Proposed future work includes controlling multiple impellers simultaneously and extending tracking to whole-body coverage using multiple markers and IR cameras [2308.00096].
5. AirTouch in mid-air haptics and virtual control
In a different research context, “AirTouch” refers to a touchless tactile interaction paradigm for aviation. The 2020 concept paper on mid-air haptics in aviation argues that pilots could interact with virtual cockpit controls in empty space using hand tracking and focused-ultrasound haptic feedback, especially ultrasonic phased arrays. In that framing, users can “touch and feel virtual 3D holograms such as buttons and dials with their bare hands, without having to wear or hold any specialised controllers.” The paper distinguishes amplitude modulation, roughly (10)–(400) Hz, from spatio-temporal modulation, in which focal points move to create tactile patterns or a “tactile hologram.” It also states explicitly that this is not classical force feedback: the system stimulates cutaneous receptors rather than generating strong kinesthetic forces [2001.01445].
The proposed hardware stack centers on the Ultrahaptics STRATOS Inspire platform, described as using 256 Murata MA40S4S piezoelectric transducers at a 40 kHz carrier frequency, with sensations projected up to 70 cm, within a 90° cone angle, and with pressure-point motion at 4 mm resolution in a 120° field of view and position updates up to 40 kHz. The simulator concept uses four such devices, an integrated Leap Motion sensor for hand and finger tracking, two Lighthouse positional tracking devices, and software support through an SDK in C++ or C#, a sensation core library, and Unity or Unreal Engine. The intended controls include virtual buttons, rotary dials, switches, landing gear levers, and throttle-related controls, while primary flight controls are treated as substantially harder and less feasible under current certification constraints [2001.01445].
Related airborne-ultrasound work strengthens the tactile side of this interpretation. Contactless stimulation of six palm positions at (50) Hz has been shown sufficient to support a six-command tactile BCI paradigm, demonstrating that focused airborne ultrasound can create localized tactile sensations on the palm without physical contact [1404.4184]. Another line of work on tap tactile presentation by airborne ultrasound separates a tap into an attenuation collision phase and a stationary phase, using Amplitude Modulation for soft, deformable-surface sensations and Lateral Modulation for rigid, resonant-surface sensations; for sustained contact, it reports a pressure-like regime with (f_{\mathrm{LM}} = 5\text{--}15\,\mathrm{Hz}) and size (< 1\,\mathrm{mm}) [2411.06653].
6. Broader landscape: air as sensing medium, notification channel, and contrast class
AirTouch also names a drone sensing system that exploits ground effect for edge detection. In that work, the drone does not deliver tactile feedback; instead, it interprets changes in rotor-wash interaction with the surface as a proprioceptive signal. AirTouch estimates disturbance signatures from IMU measurements and motor PWM signals, extracts fluctuation features in the frequency domain, fuses them with a cascaded cross-spectrum representation, and classifies edge versus non-edge. On a Crazyflie 2.1, the reported overall mean absolute detection distance error is (0.051) m, with an 86% improvement over the baseline and an added sensing and processing overhead of 43 mW [2509.21085].
A different air-mediated branch concerns remote tactile notification. SHITARA uses ring-shaped air vortexes aimed at the head or hair of d/Deaf and hard-of-hearing users, reporting that the stimulus was noticeable up to 2.5 m away and that the back of the head was especially effective and comfortable. That work emphasizes distance-dependent tuning, coarse directional cueing, and the tradeoff between louder, stronger vortex generation and quieter but weaker emission [2301.08107].
Adjacent research also shows that many “AirTouch” design goals can be approached without pure free-space interaction. EclipseTouch performs hover-distance and touch inference on ad hoc surfaces using synchronized infrared shadow casting from a headset, reporting a mean hover-distance error of 6.9 mm and 98.0% touch accuracy, while EgoTouch treats the user’s own skin as a tactile surface and reports 94.9% mean touch-classification accuracy with 6.8% mean absolute force-estimation error from headset RGB video [2509.03430] [2509.01786]. This suggests that some interaction problems often assigned to mid-air systems may migrate toward near-surface or on-body designs when tactile grounding, force estimation, and ergonomic stability are prioritized.
Commodity sensing papers reach a similar conclusion from another direction. AirWare, which fuses smartphone audio Doppler and infrared proximity sensing, is described as not reliable enough for a deployable 21-gesture in-air vocabulary, but it achieves average true positive rates above 80% on reduced gesture sets of 4–7 gestures. AirPen, by contrast, uses monocular RGB fingertip tracking and a Bi-LSTM trajectory classifier on mobile hardware, reporting 80% end-to-end accuracy with an average latency of 0.12 s [2101.10245] [1904.06122]. Taken together, these results indicate that “AirTouch” does not denote a settled architecture. It marks a design space in which air, contactlessness, and tactile or proprioceptive mediation are combined in different proportions depending on whether the task is robot safety, virtual control, environmental sensing, or notification.