ANCILE: Prototyping a wearable active protection system for personal defense

Abstract

Active protection systems have long been employed on military vehicles and installations but have traditionally been too bulky for individual use. To enhance personal safety across military and industrial applications, the Active Neutralization of Celeritous Impacts by Lateral Expulsion (ANCILE) system was developed using open-source, commercially available components. The ANCILE system is a proof-of-concept for a semi-modular interception platform that can find and deflect incoming projectiles and debris. This gives the wearer an extra layer of protection, augmenting traditional passive protection. The system employs dual cameras mounted on a wearable, turret-like mechanism that can pneumatically deploy a Kevlar sail to intercept incoming threats. The experimental testing demonstrated reliable detection and interception of objects traveling up to 7.5 m/s, with an average interception probability of 28.6% (±8.3%) at higher velocities. Component-level stress testing was employed on the Kevlar sail, indicating it could physically resist standard ballistic projectiles ranging from .22 Long Rifle to .357 Magnum. While the hardware still limits the current detection and response performance, further refinement with specialized sensors and actuators could potentially enable higher-speed operation. While tested against slow-moving projectiles, this preliminary work validates the viability of a low-cost, wearable active protection system, which could find potential applications in construction, manufacturing, aerospace, and defense.

Keywords

wearable active protection ballistic deflection projectile interception pneumatic deployment and personal safety systems

Introduction

Individually wearable “force fields” have long been the domain of science fiction. However, the concept of using directed energy and active interception for defense originated in mid-twentieth-century military research, notably during the Cold War era of reactive armor development (Motyl et al., 2017) [1]. In the following decades, point-defense systems, which detect, track, and neutralize incoming projectiles in real time, have evolved from large vehicle- or ship-mounted configurations into sophisticated, autonomous systems integrated into modern defense infrastructure. These active systems (e.g., naval close-in weapon systems and active protection systems for armored vehicles) are optimized for intercepting macroscopic threats, including missiles, rockets, and artillery shells (Chia, 2019; Rosenberg et al., 2005; Wickert, 2006) [2, 3, 4]. Active systems augment passive protective measures, such as armor. However, their principles suggest potential adaptation to smaller-scale, human-centered protection systems that can respond to high-velocity, short-range, ballistic or impact hazards (Cooper and Costello, 2004; Corbin et al., 2018; da Rocha et al., 2020; Faraji et al., 2015; Hill et al., 2015; Norris et al., 2020; Thompson et al., 1972) [5, 6, 7, 8, 9, 10, 11].

The Active Neutralization of Celeritous Impacts by Lateral Expulsion (ANCILE) system is an early prototype exploring this concept, intended to augment conventional passive, wearable protection. Constructed from open-source, commercially available hardware and software, the ANCILE system explores the feasibility of a wearable active ballistic interception system. The shoulder-mounted configuration employs a closed-loop control scheme that integrates camera-based visual detection with rapid-response actuation. When the device sees an incoming projectile, it uses a compressed-air launcher to send a Kevlar shield along the predicted impact vector. This deflects or dissipates kinetic energy before the impact. This proof-of-concept platform lays the groundwork for making personal active-defense technology smaller and better, with uses in military, industrial, and high-risk operational settings.

Background

Point-defense systems

The concept of individual force fields, popularized in Frank Herbert’s Dune (1965), introduced a speculative defensive technology capable of selectively blocking projectiles traveling faster than approximately 9 cm/s (Herbert, 1965) [12]. Although purely fictional, this concept anticipated real-world innovations in active and reactive protection technology. Since the early Cold War, military engineers have developed increasingly sophisticated systems to shield armored vehicles from kinetic and chemical energy threats (Mayseless, 2011; Motyl et al., 2017) [13, 1]. For instance, reactive armor employs explosive layers that detonate outward upon impact, neutralizing or deflecting the penetrating jet of shaped-charge warheads (Mayseless, 2011) [13]. Building on these principles, active protection systems emerged to detect, track, and intercept incoming munitions using radar sensors and countermeasure projectiles, providing dynamic protection that augments traditional steel and composite armor (Chia, 2019; Rosenberg et al., 2005; Wickert, 2006) [2, 3, 4]. While military vehicles possess active protection systems, they augment, rather than replace, traditional passive measures.

Researchers have conducted parallel research into electromagnetic, laser, and electric field-based defenses to apply rapid energy projection to disrupt or deflect threats without physical contact (Ahmed et al., 2021; Chia, 2019; Mayseless, 2011) [14, 2, 13]. However, to date, the extreme power demands and limitations of compact, mobile energy sources have confined such concepts primarily to experimental phases. At the systems level, naval vessels and fixed installations have employed analogous point-defense technology (e.g., radar-guided close-in weapon systems) to engage incoming missiles, artillery shells, and drones before impact (Chia, 2019) [2]. Although highly effective, these systems rely on a substantial power-generation and stabilization infrastructure that remains infeasible for dismounted infantry applications. Therefore, individual soldiers continue to depend predominantly on passive ballistic protection, ranging from Kevlar-based vests to cutting-edge composite body armor, to mitigate the effects of small-arms fire and fragmentation (Anh et al., 2018; Ballistics 101, 2009; Sockalingam et al., 2017) [15, 16, 17]. Despite persistent exploration of personal electromagnetic or plasma shielding, the engineering challenges of energy density, reaction time, and portability remain principal obstacles to realizing a true “individual forcefield” in the foreseeable future (Motyl et al., 2017) [1]. Such active protection technology could supplement individual survivability, covering gaps in passive protective gear.

Consumer technology

In civilian applications, as in military contexts, the evolution of individual personal protective equipment (PPE) has focused on passive defenses: mechanical barriers (e.g., helmets, face shields, and ballistic fabrics) designed to absorb or deflect external forces (Peatie et al., 2024) [18]. These systems rely on material science advancements rather than dynamic interaction with threats, and they do not cover the entirety of the body and head for practical reasons (e.g., constrainted movement, limited vision, and/or heavy weight). However, the emergence of portable directed-energy technology has prompted the examination of the role of these systems in creating a responsive interface between the human body and its environment. An evolving area of research explores their potential to generate controlled tactile stimuli, leading to novel forms of haptic feedback that simulate physical contact or impact without direct mechanical coupling. Several actuation modalities have been demonstrated, most notably pneumatic, acoustic, and optical actuation systems capable of delivering discernible impulses via the air or radiation pressure (Shtarbanov, 2018; Shusser and Gharib, 2000) [19, 20].

A well-documented example of pneumatic haptic feedback is Disney Research’s AIREAL (known as AirTap), which employs vortex-ring propulsion to deliver brief, low-energy tactile sensations at a distance (Shtarbanov, 2018; Shusser and Gharib, 2000; Sodhi et al., 2013) [19, 20, 21]. The mechanism involves toroidal air pulses generated by a controlled diaphragm that could convey spatially localized feedback to a user’s skin (Akhmetov et al., 2019; Shtarbanov, 2018) [22, 19]. Similar systems have appeared in interactive entertainment, training simulators, and virtual environments, reinforcing the value of mid-air haptic effects in enhancing immersion (Ramírez et al., 2018; Shusser and Gharib, 2000) [20, 23]. Parallel research in acoustic actuation has focused on using ultrasound phased arrays to produce focused pressure points, enabling touch perception without physical contact (Shusser and Gharib, 2000) [20]. Optical or laser-based systems have also demonstrated momentum transfer sufficient to trigger tactile sensations, typically through photon pressure or localized thermal gradients (Motyl et al., 2017) [1].

Despite these advancements, such technology remains constrained regarding range and energy-transfer capacity. Energy dissipation in air, beam divergence, and safety limitations restrict operation to low-force interactions, making such systems unsuitable for defensive or protective applications. Some experimental systems have attempted to overcome these limits by incorporating aerosolized, combustible, or particulate payloads into pneumatic vortices to increase mass and energy density, although this approach introduces complexities in control, safety, and reproducibility (Akhmetov et al., 2019; Anishchanka and Loktionov, 2020; Ishizuka et al., 2013) [22, 24, 25]. Achieving individualized active protection, where the PPE senses an impending ballistic or high-energy threat and intercepts or neutralizes it, requires an integrated network of high-speed sensors, predictive algorithms, and compact interception mechanisms. Even low-energy lasers igniting combustible aerosols add unpredictability where consistency is required (Boonsri et al., 2012; Gargan-Shingles, 2017; Ildar, 2021; Kopecek et al., 2003; Ma et al., 1998; Weinrotter et al., 2005) [26, 27, 28, 29, 30, 31]. This transition from wholly passive to both passive and active PPE parallels the broader shift from static defense to dynamic survivability engineering, a frontier in which sensing, computation, and energy delivery must converge according to strict temporal and ergonomic constraints.

Ballistic Interception

Individualized action protection systems represent a developing frontier in applied defense engineering, with direct implications for military and civilian safety applications (Bigdeli et al., 2025)[32]. Although environmental and operational requirements differ between these domains, the underlying physics of projectile interception remain focused on ballistic dynamics. In the military context, projectile motion stability is critical in determining the effective range, impact precision, and kinetic efficiency of a weapon system (Fedorov et al., 2021; Yang et al., 2021) [33, 34]. Rifled barrels, aerodynamic shaping, and rotational stabilization reduce yaw, precession, and drag, ensuring that the projectile maintains a predictable, stable trajectory over long distances (Hatakeyama and Mochiyama, 2010) [35]. In contrast, civilian, disaster, or industrial scenarios typically involve unintended or accidental projectiles (e.g., bolts, fragments, or shards) that are launched unpredictably during mechanical failures, explosions, or high-energy tool discharges (Peatie et al., 2024; Taddeucci et al., 2017; Xue et al., 2022) [36, 18, 37]. These objects rarely follow a stable ballistic path; instead, they exhibit chaotic motion dominated by an asymmetric mass distribution and nonuniform aerodynamic loading (Taddeucci et al., 2017) [37].

Computational simulations and wind-tunnel experiments performed on thin-walled shells have demonstrated that even marginal lateral forces (on the order of a few newtons) can induce significant trajectory deviation (Li et al., 2019) [38]. Such disturbances reduce accuracy by introducing rapid angular drift, neutralizing the ballistic stability initially imparted by spin or shaping (Hahn et al., 2009; Peatie et al., 2024) [39, 18]. Destabilizing a conventional projectile requires applying corrective forces comparable to or greater than the Coriolis terms governing two-dimensional (2D) lateral motion. Commercial sensors could potentially detect and respond to firearm bullets, given the specific calibration (Scannapieco et al., 2016) [40]. The scalability of this principle to smaller calibers, including 5.56 or 9 mm rounds, suggests a theoretical pathway for developing interception systems to deflect or degrade projectile momentum via controlled micro-actuation or induced, localized turbulence (Faraji et al., 2015; Rosenberg et al., 2005; Slegers et al., 2008) [11, 4, 41].

If a wearable system could rapidly detect an incoming projectile, compute its trajectory in real time, and deploy a mechanical or electromagnetic countermeasure, it could feasibly neutralize high-velocity rifle rounds and slower, erratically moving industrial debris. Even commercial sensors could potentially detect conventional ballistics (Mérelle et al., 2016) [42]. Given that thin Kevlar laminates can dissipate sufficient kinetic energy to halt smaller projectiles, coupling such passive materials with active sensing and actuation layers presents a promising design approach. The integration of radar, lidar, or high-speed optical sensors would allow the system to anticipate impact vectors within milliseconds, expanding the individual protection from entirely passive resistance to integrated active and passive defense (Scannapieco et al., 2016; Slegers, 2008) [40, 41]. This convergence of real-time detection, dynamic response, and scalable material science could define the next generation of individualized protective technology for application in combat theaters and high-risk industrial environments.

Methods

Overview

The ANCILE active-interception module employs a pneumatic actuation mechanism driven by compressed-air reservoirs, selected for their high power density and rapid-response characteristics. Stepper motors that provide fine angular resolution and synchronized motion control complement this module, allowing the turret to reposition and respond with the interception system within milliseconds. The interception interface comprises a lightweight Kevlar sail engineered for high tensile strength and energy absorption while maintaining an exceptionally low mass-to-area ratio (Sockalingam et al., 2017) [17]. As an early stage prototype, this system implementation must enable the effective deflection or dissipation of kinetic energy from an incoming projectile while ensuring that the overall platform remains compact and wearable, integrating both detection and actuation Mounting configurations include shoulder, helmet, and backpack integrations, providing adjustable coverage angles and adaptable ergonomic balance. The closed-loop control algorithms dynamically fuse visual feedback and actuator position data to ensure continuous alignment between the detection and interception subsystems (Islam et al., 2020; Li et al., 2019; Qin et al., 2019) [38, 43, 44]. The result is a highly responsive, autonomous protection unit that compensates for wearer movement and environmental variability while maintaining a reliable interception probability across a broad range of engagement scenarios (Islam et al., 2020; Li et al., 2019; Qin et al., 2019) [38, 43, 44].

The ANCILE device is a proof-of-concept for an individually-worn, closed-loop active-interception system intended for individual use. The closed-loop control algorithms must dynamically fuse visual feedback and actuator position data to ensure continuous alignment between the detection and interception subsystems (Islam et al., 2020; Li et al., 2019; Qin et al., 2019) [38, 43, 44]. To these ends, the system employs camera-based object tracking to detect threats and moves a turret to engage them. The system applies compressed air as its primary actuation method and employs coordinated stepper-motor positioning to aim its interception mechanism rapidly. Two high-speed global-shutter cameras, operating at 140 frames per second (fps), provide low-latency sensing with minimal motion blur to reliably detect and track incoming objects. A lightweight Kevlar sail serves as the interception surface, absorbing or deflecting impacts while maintaining a compact and portable platform (Sockalingam, at el. 2017) [17]. As detailed in supplemental materials, the entire prototype system cost slightly under $400 at $395.55.

System structure

As presented in Figure 1, the ANCILE system was built around two cooperating subsystems: a GRBL-based Arduino (Arduino, Turin, Italy) motion controller that drives the mechanical positioning hardware, and a Raspberry Pi 5 (Raspbery Pi, Cambridge, UK) vision subsystem that processes camera data, estimates trajectories, and generates the corresponding GRBL instructions (grbl, 2009; Jeon, 2002) [45, 46]. Combined, these components form an integrated, autonomous protective system to identify incoming hazards and dynamically react in real time.

Figure 1.

Detection and interception system logic.

Processing software

Initially employed in computer numerical control (CNC) machining, GRBL was applied in ANCILE as the firmware that drives the Arduino-based motion system. As a CNC framework, the system interprets incoming G-code commands and converts them into precise step and direction signals for stepper motors, allowing the interception mechanism to move quickly and predictably. Because GRBL handles timing, acceleration, and coordinated motion on the microcontroller, the Raspberry Pi can focus on sensing and decision-making while the Arduino executes the low-level motor control required for rapid, accurate positioning.

Image processing

The ANCILE system employs two IMX296 cameras (Sony, Tokyo, Japan) connected to a Raspberry Pi 5 to sense incoming projectiles and estimate their paths in real time. The cameras provide YUV420 frames, which are processed using a motion- and color-based detector at 140 fps. A commercial PC laptop was used for processing. In Figure 2, motion detection identifies fast-moving objects between frames based on a target velocity threshold, and an integrated hue, saturation, value (HSV) filter isolates the cyan-colored projectile tip. Combining these techniques allows the system to ignore background movement, lighting changes, and noise while reliably locking onto the projectile (Islam et al., 2020; Qin et al., 2019) [43, 44]. Figure 2 presents the visual processing components employed in the ANCILE system, showing the process applied to each frame.

Figure 2.

Image processing flow from two-dimensional cameras to the stereo triangulation of three-dimensional coordinates.

When both cameras detect the projectile in the same frame interval, the software performs stereo triangulation to convert the two-pixel locations into a 3D position (Qin et al., 2019) [43]. In each dimension, the Raspberry Pi stores several of these 3D points ( $v_{x}, v_{y}, v_{z}$ ) across consecutive frames and fits a simple linear trajectory model to estimate the projectile speed and travel direction (Islam et al., 2020; Qin et al., 2019) [43, 44]. In equation (1), this process provides the system with a short-term prediction of how the projectile moves through space ( $‖ v ‖$ ):

‖ v ‖ = \sqrt{v_{x}^{2} + v_{y}^{2} + v_{z}^{2}}

(1)

Using this prediction, the Raspberry Pi determines where the projectile intersects a fixed interception plane in front of the CNC mechanism. As this technique is well-documented by prior work, more information is available in the supplemental data (Islam et al., 2020; Qin et al., 2019) [43, 44]. If the predicted interception location is reachable within the available time, the Pi sends G-code commands to GRBL so the CNC can move into position and prepare to fire. The GRBL module executes the motion with correct step timing and coordinated axis control, and the launcher is activated using standard coolant commands (grbl, 2009; Jeon, 2002) [46, 45].

In summary, each camera scanned for objects moving in a certain velocity using optical follow. The distinctive colors of the dart and tip were used as a second threshold. The two camera samples and their fixed distance was used to triangulate the projectile. If the location was confirmed by both cameras, the projectile was observed for a short window (<2 ms). A simple linear trajectory was fitted to the projectile’s trajectory. If the tracked projectile was heading to the target zone, the processor used CNC code (via the GRBL module) to move the turret and launch the sail at the incoming projectile. Image processing took 8.5 ± 5 ms, projectile triangulation took <2 ms, and fitting a path and solution took 10 ms. Ideally, the system had a latency below 20 ms. While the position of the turret added high variability, the longest average overall latency, from overall detection to deployment, was 418 ± 25 ms. In short, after velocity and HSV filtering, two detections of the projectile were used to interpolate the path and initiate interception.

Launcher design

The 1.26 kg launcher was intended as an exploration of concept viability. Standard school backpacks and PPE weigh far more than a pair of shoulder-mounted launchers (Peatie et al., 2024) [18]. The launcher comprises an acrylic hose mounted to a tilting frame that serves as the y-axis of the CNC assembly. Figure 3 presents a breakdown by component weight. The turret was designed to tilt at 50° for vertical motion and 120° for horizontal motion. The turret base is 200 mm high, 145 mm long, and 139 mm wide.

Figure 3.

Turret diagram and component list, including the base, body, bearing, stepper motor, Arduino Uno, cameras, Raspberry Pi, lever mount, and drive gear.

Figure 4 depicts the launcher view from the front. As shown, the tethered Kevlar sail is placed in line with the mounted cameras to maximize accuracy.

Figure 4.

Assembled ANCILE launcher viewed from the front. The two mounted cameras (upper left) track incoming projectiles. The camera feed connected to a board (lower left), and the main processor (unseen) would determine whether to intercept an object or not. When intercepting, the kevlar sail (upper right) is launched using compressed air from the pneumatic hose (lower right).

Figure 5 depicts the launcher view from the top, demonstrating the base of the device and the exposed electronics. The pneumatic hose was attached to the launcher with zip ties to minimize recoil from compressed air release.

Figure 5.

Assembled ANCILE launcher viewed from the top, detailing how the pneumatic hose is secured to the launcher with zip ties.

Figure 6 depicts the launcher view from the back, depicting the vertical mounting for the stereo camera arrangement.

Figure 6.

Assembled ANCILE launcher viewed from the back, detailing the integrated camera mounting.

The hose is connected to an air compressor via a solenoid valve, which is wired to the coolant output on the Arduino CNC shield. When the coolant is turned on, compressed air is released through the hose. In Figure 7, a lightweight tethered Kevlar sail, about 100 × 100 mm and weighing 40 g, is placed above the hose, and the tilt angle determines the direction of the air burst. When the system fires, the air jet strikes the sail at 0.689 MPa (∼100 PSI), swinging it into the projectile path and intercepting it.

Figure 7.

Kevlar sail for ballistic interception.

Interception testing

Figure 8 illustrates the testing of a modular spring-based launcher system that enables precise control of the projectile velocity. The test projectile was loaded in, and the spring was set to the desired kinetic energy. The test projectile was then launched in front of the device.

Figure 8.

Illustrating a modular spring launcher designed for interception tests.

By adjusting the internal spacers and outer extensions, varying the potential energy of the spring can produce a range of speeds. The spring-based launcher discharged a (12.7-mm-diameter, 72-mm-long) foam dart with an M4 x 12 mm bolt inserted, weighing 0.0114 kg. This approach enables an evaluation of the ANCILE system performance across various projectile velocities, travel distances, and timing conditions.

In Figure 9, the turret was positioned perpendicularly to the launcher. The vertical and horizontal distances between the two devices were 43 and 167.6 cm, respectively. A velocity of 20 m/s was initially employed, decreasing by 2.5 m/s if the tests failed. The lowest velocity was 5 m/s. Three trials were conducted at each velocity, and the results were averaged.

Figure 9.

Top-down view of the turret positioned relative to the spring launcher.

Interception scoring

Interception accuracy was determined by the percentage of hits at each velocity. A lidar range finder was used to verify the distance at the beginning of testing. Testing occured each day in the afternoon, to minimize noted low-light errors. The velocity of each test projectile was verified by a chronograph. A hit was counted as a successful the sail deployment that either stopped or deflected the test projectile from hitting the target (the launcher or behind it). Each test was confirmed by direct observation. If the sail did not deploy or the projectile reached the target, it was counted as a miss. The sail and entire system were reset for each trial. The total percentage of hits and misses was calculated for each velocity. The interception rate was calculated as an average across velocities.

Sail testing

In addition to launcher testing, component-level stress testing was conducted with the Kevlar sail against conventional ballistic projectiles, evaluating the limits of its physical durability. As the interception method, its durability and robustness were imperative for any successful worn active protection. In the component level tests, the sail was deployed on a string and applied as a ballistic pendulum. The experiment tested the velocity imparted to the pendulum and the potential failure conditions of the Kevlar sheet. To estimate the energy imparted to the sail, the mass of each bullet was compared pre- and post-interception. The cartridges include .22 Long Rifle (LR), .38 Special, .45 Automatic Colt Pistol (ACP), and .357 Magnum. The ammunition brands were Federal AutoMatch .22 LR, and hand-loaded.38 Special, Federal Range and Target .45 ACP, and hand-loaded .357 Magnum. Two cartridges of each brand were used. All bullets were discharged from a 5-inch handgun at 3.4 m. Given the high speed of bullets and short distance, the effects of drag and projectile drop were assumed to be minimal.

A ballistic pendulum of mass 2.89 kg was employed to measure the velocity of each bullet to compare the use of the Kevlar sail. In equation (2), the kinetic energy KE (Joules) results from the relationship between velocity $v$ and mass $m_{o b j}$ (Milutinovic et al., 2019) [47]:

K E = (. 5) * m_{o b j} v^{2} .

(2)

In the pendulum, the potential energy at its maximum swing

U

is given in equation (3), which is a product of the mass m, height h, and gravity g (Milutinovic et al., 2019) [47]:

U = m g h .

(3)

In addition, equation (2) can be substituted into equation (3) for equation (4) at the maximum swing, representing the energy impulse of the impact:

(. 5) * m_{o b j} v^{2} = m g h

(4)

Although some energy is lost, the ballistic pendulum equation in equation (5) was streamlined for calculation (Milutinovic et al., 2019) [47]:

v = \frac{m}{m_{o b j}} \sqrt{g h}

(5)

The mass of the bullet was measured after recovery to determine whether the amount of mass loss was substantial. If the bullet lost more than 5% of its mass, it shattered and fragmented instead of transferring its energy. Observed with a camera, the calculated pendulum impulse could be assumed to be at a minimum due to the physical setup constraints (Milutinovic et al., 2019) [47].

Statistical testing and hypotheses

Statistical testing was independently tracked for interception trials and component-level sail testing. Due to camera and system latency limitations, it was hypothesized that this system would not operate reliably above a certain velocity. A velocity threshold that the device could not intercept at or above indicated a hard limit to the system’s function. As a direct and proven measure, a paired t-test was conducted to determine whether the differences between velocities with successful and unsuccessful hits were significant [38, 11]. If the values differed significantly, the velocity threshold was considered a hard limit to the system’s effectiveness. The component-level ballistic testing employed a statistical correlation, calculating Pearson’s correlation coefficient for the impacting bullet energy and imparted energy. Consistency was sought in Kevlar sail testing, so high correlation between energy imparted and bullet kinetic energy was preferred. A higher coefficient value indicated a reliable Kevlar sail, resulting in a coefficient closer to 1. It is hypothesized that the sail exhibited a high correlation between imparted energy and the projectile velocity during ballistic testing (Sockalingam et al., 2017) [17], and if not, the protectiveness of the sail was unreliable.

Results

Overview

Interception tests were the primary metric observed, independently of component-level ballistic stress tests on the Kevlar sail. . For the interception tests, the average results from each of the three trials were compared. For the ballistic stress tests on the Kevlar sail, the amount of kinetic energy imparted was calculated. Failure modes were also noted.

Interception testing results

The mean velocity of the trials was 12.5 ± 5.40 m/s. Across all cases in Figure 10, the overall hit rate was 28.6% ± 8.3%. The highest projectile velocity that could be reliably intercepted was 7.5 m/s. The average velocity of the successfully intercepted projectiles was 6.8 ± 0.5 m/s. The average velocity of the missed interceptions was 15.4 ± 0.6 m/s. The difference between the hit and miss velocities was significant (t = 21.2, p < 0.001).

Figure 10.

Average velocities of successful interception hits and misses.

Kevlar sail ballistic testing results

Table 1 lists the ballistic results for the Kevlar sail. The bullet fragment losses were all less than 5%. The highest impulse energy was imparted by the .357 Magnum, and the minimum energy was 26.2 ± 1.7 J. The least energy imparted was by the .22 LR, at a minimum of 0.49 ± 0.02 J.

Table 1.

Ballistic test results on the Kevlar sail.

Caliber	Energy (J)	Mass (g)	M_delta (%)	E_impulse (J)
.22 LR	108.00	2.49	0.04	0.49 ± 0.02
.38 special	254.00	10.19	0.00	6.80 ± 0.4 J
.45 ACP	352.00	14.90	0.01	15.40 ± 0.9 J
.357 magnum	780.00	10.14	0.01	26.20 ± 1.7 J

Note. LR: Long Rifle.

Each round was placed beside recovered bullets in Figure 11 to compare the visual effects of the impact with the Kevlar sheet. Figure 12 illustrates the repeated ballistic impact damage to the sail. The Pearson correlation coefficient was calculated as r = 0.971, confirming the experimental hypothesis regarding the relationship between energy imparted by ballistic impact.

Figure 11.

A visual comparison between standard rounds (A, top row) and recovered bullets (B, bottom row). From left to right, the cartridges are: .22 Long Rifle (LR), .38 Special, .45 Automatic Colt Pistol (ACP), and .357 Magnum.

Figure 12.

Visual damage to the Kevlar sail after ballistic testing.

Failure modes

During interception testing, specific failure causes were noted. One was the influence of ambient lighting on object tracking. As the tests were conducted outdoors, the time of day resulted in a number of erroneous detection errors until a daily time was standardized. Another issue was due to the stereo camera setup, as the vertical stacking added complexity to calibration, relative to a horizontal alignment. Tracking and timing were also inconsistent, owing to delays in translation and actuator control.

Discussion

Interpretation

While using low velocity projectles, the ANCILE system demonstrated a proof of concept for a low-cost, wearable active-projectile interception system to augment conventional PPE. The overall interception rate was 28.6% ± 8.3% across the tested velocity spectrum; however, the reliability exhibited a pronounced transition at a distinct threshold velocity. In contrast to Dune’s fictional defensive fields, which nominally arrest objects at approximately 9 cm/s, the ANCILE system consistently intercepted projectiles at or below 7.5 m/s, a regime representative of many unintentional industrial projectiles (e.g., tool fragments or ricochets) (Herbert, 1965; Peatie et al., 2024) [12, 18]. While the overall interception rate was low, the Kevlar sail withstood component-level ballistic stress testing. Given the strong correlation between the impulse and projectile kinetic energy, the demonstrated ability of the Kevlar sail to withstand impacts from handgun rounds up to .357 Magnum suggests its substantial robustness to conventional ballistic threats and accidental projectiles in industry (Sockalingam et al., 2017) [17]. Currently, contemporary worn PPE is predominantly passive, whereas the ANCILE system establishes a proof-of-concept that an actively controlled, body-worn interceptor is technically feasible using low-cost, commercially available, or open-source components. While a limited prototype, it demonstrates the potential use of individually worn active protection to cover gaps in passive protection. The resulting 1.26-kg prototype exhibited multiple technical limitations requiring further engineering refinement.

Limitations

The ANCILE system is a limited prototype with many opportunities for improvement, as it is currently not a viable active protection system. The first is the sensor selection, based on relatively slow, low-cost cameras. The velocities of conventional bullets are far higher by two orders of magnitude than the test projectiles, greatly reducing the detection and response window. The low interception rate, even for a low-velocity projectile, would need to be greatly improved for real use. Space debris and conventional bullets require far more precise sensors for a rapid, timely response (Bigdeli et al., 2025; Hahn et al., 2009) [32, 39]. Trajectory estimation and deployment determination must be performed more quickly than the 140 fps cameras allow. The long latency due to movement, vertical camera array calibration difficulties, and low-light difficulties were previously noted. In addition to moving to non-optical tracking methods (e.g., radar), improving the algorithms (e.g., those with Kalman filtering or fitted functions) could improve resilience (Islam et al., 2020; Qin et al., 2019) [43, 44]. The deployment system has flaws in the turret and propulsion mechanism, as faster alternatives to pneumatic deployment could cut the system latency (e.g., electromagnetic or pyrotechnic). As rapid response would be required, the turret mechanisms constrain the ability to swivel, pan, and pivot, limiting the field of view of the sensors. As the trend in PPE is towards lighter materials, the pneumatic system may add extra weight relative to additional passive protection, which was not accounted for in the test model. Various projectile types require their own interception methods, some of which may be more difficult to intercept than a foam dart (Hahn et al., 2009; Slegers, 2008) [39, 41]. The sail geometry and materials could be optimized to achieve a greater coverage area, especially in vulnerable areas (Mayseless, 2011; Wickert, 2006) [13, 3]. The lack of a retraction and reset system also limits its repeated use in the field. Testing at a single distance and projectile vector was unrealistic, as an impact could strike from unpredictable range and velocity. Likewise, the weight, size, and shape could be optimized, as modern PPE has been based on lighter, flexible materials, rather than bulky wearables. A broader range of projectiles, including conventional and unconventional projectiles, could be explored to improve sampling efficiency (Anh et al., 2018; Norris et al., 2020) [15, 5]. Despite these limitations, the project possesses clear potential.

Future work

While a proof-of-concept, future refinement of the ANCILE system is necessary for practical use. The concept could be improved by reducing the system latency; optimizing the interception angle; broadening the sensor modalities and field of view; improving the detection algorithm; redesigning the Kevlar sail geometry; improving sail deployment time; testing a greater variety of projectiles at different ranges, velocities, and angles; and reducing the time by improving the system for sail deployment (Thompson et al., 1972; Yang et al., 2021) [6, 33]. A lower latency sensor (e.g., radar or lidar) with a far higher sampling rate would be required to intercept conventional ballistic projectiles. Low-cost, open-source components could be replaced with higher-grade components, as existing high-performance space robotics and aerospace systems can react in milliseconds (e.g., electromagnetic and pyrotechnic) (Bigdeli et al., 2025; Mérelle et al., 2016) [32, 42]. Reducing the weight and profile, as well as mounting system, would be required for a viable device. Beyond hardware and software improvements, determining the most ergonomic and comfortable way to wear and carry the device would involve a task-based biomechanical analysis. The system could be optimized for specific projectile types, from those in industrial mishaps and conventional ballistics to entertainment (Hill et al., 2015; Li et al., 2019; Rosenberg et al., 2005) [4, 7, 38]. An optimized, low-weight future device could also be mounted on a spacecraft to divert incoming debris. In addition to its size and weight, the device must also be tested to ensure it remains reliable while the user is in motion or shifts posture. In summary, the ANCILE system is a proof-of-concept that could bring the popular depiction of force fields closer to reality.

Conclusions

Despite the existence of active protection in military vehicles for decades, wearable PPE for individuals relies on passive properties (Chia, 2019; Peatie et al., 2024; Sockalingam et al., 2017) [2, 18, 17]. Covering gaps in passive individual protection, the ANCILE system demonstrates the concept of an active point-defense system to a wearable device, evaluating interception rate with a low-velocity projectile and component-level resilience to ballistic impacts. The device employed two cameras to control a pneumatic turret that launches a Kevlar sail. The interception rate was 28.6% ± 8.3% for velocities from 5.0 to 20.0 m/s. Velocities at or below 7.5 m/s were reliably hit, with reliability dropping above that velocity. As in prior literature, the Kevlar sail withstood pistol fire from .22 LR to .357 Magnum (Sockalingam et al., 2017) [17]. Despite its relatively low hit rate across all velocities, many dropped objects and unintentional projectiles were at or below 7.5 m/s. With further improvements to its hardware, design, mounting, cost reduction, and software, future wearable active protection systems, integrated with traditional passive protection, could be deployed to prevent impacts and injuries in industry, entertainment, emergency response, firefighting, aerospace, and defense domains (Bigdeli et al., 2025; Cooper and Costello, 2004; da Rocha, 2020; Peatie et al., 2024; Sodhi et al., 2013) [32, 21, 10, 18, 8].

Footnotes

Acknowledgments

The authors thank the Ohio State University. The authors disclose the use of Perplexity AI (Perplexity, 2026) for grammar and punctuation editing. The authors would like to acknowledge a prior preprint present at:

ORCID iDs

John LaRocco

Qudsia Tahmina

John Simonis

Author contributions

Conceptualization, J.L. and J.S.; methodology, J.S. and A.C.L.; software, J.S.; validation, J.L., J.S., and Q.T.; formal analysis, J.L.; investigation, J.S. and A.C.L.; resources, Q.T.; data curation, J.L.; writing—original draft preparation, J.L.; writing—review and editing, J.L.; visualization, J.S. and A.C.L.; supervision, J.L. and Q.T.; project administration, J.L. and J.S.; funding acquisition, Q.T. All authors have read and agreed to the published version of the manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data, models, and supplementary information in this work are available at .

References

Ahmed

Mohsin

Ali

SMZ

(2021) 'survey and technological analysis of laser and its defense applications. Defence Technology 17(2): 583–592. https://doi.org/10.1016/j.dt.2020.02.012

Akhmetov

Kotelnikova

Nikulin

, et al. (2019) 'generation of large-scale high-velocity vortex rings by initiating an explosive. Combustion, Explosion, and Shock Waves 55(4): 390–394. https://doi.org/10.1134/s0010508219040038

Anh

Dat

(2018) 'development of a framework for ballistic simulation. International Journal of Simulation Modelling 17(1): 623–632. https://doi.org/10.2507/ijsimm17(4)451

Anishchanka

Loktionov

(2020) Comparison of laser breakdown and laser ablation ignition thresholds of combustible gas mixtures. Journal of Physics: Conference Series 1556(1): 012012. https://doi.org/10.1088/1742-6596/1556/1/012012

Ballistics 101 (2009) 308 Winchester ballistics. Ballistics 101, 1 January. Available at: https://www.ballistics101.com/308_winchester.php. (Accessed 3 March 2022).

Bigdeli

Srivastava

Scaraggi

(2025) Mechanics of space debris removal: a review. Aerospace 12(4): 277. https://doi.org/10.3390/aerospace12040277

Boonsri

Sumriddetchkajorn

Buranasiri

(2012) 'Laser-based mosquito repelling module'. In: Proceedings of the Photonics Global Conference, pp. 1-4.

Chia

(2019) A Simulation of a Combined Active and Electronic Warfare System for the Defense of a Naval Ship Against Multiple low-altitude Missiles Threat. Monterey, CA. Naval Postgraduate School.

Cooper

Costello

(2004) Flight dynamic response of spinning projectiles to lateral impulsive loads. Journal of Dynamic Systems, Measurement, and Control 126(3): 605–613. https://doi.org/10.1115/1.1789976

10.

Corbin

Hanna

Royston

, et al. (2018) 'Impact-induced acceleration by obstacles. New Journal of Physics 20(5): 053031. https://doi.org/10.1088/1367-2630/aac151

11.

da Rocha

HFLP

(2020) Trajectory and Aerodynamic Analyses of Air Launched fire-extinguishing Projectiles. Covilha, Portugal. Universidade da Beira Interior.

12.

Faraji

Veile

Hemleben

, et al. (2015) Impulse redirection of a tethered projectile. In: Proceedings from the Conference on Dynamic Systems and Control, Vol. 57267, V003T43A002.

13.

Fedorov

Veldanov

Sotskiy

, et al. (2021) Jet thrust penetrators for sounding the surface layer of space bodies. Acta Astronautica 180(1): 189–195. https://doi.org/10.1016/j.actaastro.2020.12.030

14.

Gargan-Shingles

(2017) The Influence of Azimuthal Velocity and a Parallel Wall on Vortex Ring Evolution. Melbourne, Australia. Monash University. grbl (2009). grbl (Version v0.9j and earlier). Available at: https://github.com/grbl/grbl. (Accessed 5 October 2025).

15.

Hahn

Frederick

Slegers

(2009) Predictive guidance of a projectile for hit-to-kill interception. IEEE Transactions on Control Systems Technology 17(6): 745–755. https://doi.org/10.1109/tcst.2008.2004440

16.

Hatakeyama

Mochiyama

(2010) ‘Shooting manipulation'. In: Abstracts of the international conference on advanced mechatronics: toward evolutionary fusion of IT and mechatronics, pp. 456-461.

17.

Herbert

(1965) Dune. Boston, MA, USA. Chilton Books.

18.

Hill

Woodward

Arisumi

, et al. (2015) Wrapping a target with a tethered projectile. In: Proceedings of the IEEE International Conference on Robotics and Automation, pp. 1442–1447.

19.

Ildar

(2021) Raspberrypi for Mosquito Neutralization by Power Laser. arXiv preprint, arXiv:2105.14190.

20.

Ishizuka

Yamashita

Shimokuri

(2013) Further investigation on the enhancement of flame speed in vortex ring combustion. Proceedings of the Combustion Institute 34(1): 745–753. https://doi.org/10.1016/j.proci.2012.06.024

21.

Islam

Asikuzzaman

Khyam

, et al. (2020) Stereo vision-based 3D positioning and tracking. IEEE Access 8: 138771–138787. https://doi.org/10.1109/access.2020.3011360

22.

Jeon

(2002) Grbl: an open-source, embedded, high-performance g-code parser and CNC milling controller. Available at: https://github.com/gnea/grbl. (Accessed 31 December 2002).

23.

Kopecek

Maier

Reider

, et al. (2003) Laser ignition of methane–air mixtures at high pressures. Experimental Thermal and Fluid Science 27(4): 499–503. https://doi.org/10.1016/s0894-1777(02)00253-4

24.

Fan

(2019) Dynamic response analysis for a terminal guided projectile with a trajectory correction fuze. IEEE Access 7: 94994–95007. https://doi.org/10.1109/access.2019.2928718

25.

Alexander

Poulain

(1998) 'laser spark ignition and combustion characteristics of methane-air mixtures. Combustion and Flame 112(4): 492–506. https://doi.org/10.1016/s0010-2180(97)00138-7

26.

Mayseless

(2011) Effectiveness of explosive reactive armor. Journal of Applied Mechanics 78(5): 051006. https://doi.org/10.1115/1.4004398

27.

Merelle

Gaugue

Louis

, et al. (2016) UWB pulse radar for micro-motion detection. In: International Conference on Ultrawideband and Ultrashort Impulse Signals (UWBUSIS), pp. 152–155.

28.

Milutinovic

Hristov

Jerković

, et al. (2019) The application of the ballistic pendulum for the bullets velocity measurements. IOP Conference Series: Materials Science and Engineering 659(1): 012016. https://doi.org/10.1088/1757-899x/659/1/012016

29.

Motyl

Magier

Borkowski

, et al. (2017) 'theoretical and experimental research of anti-tank kinetic penetrator ballistics. Bulletin of the Polish Academy of Sciences, Technical Sciences 65(3): 1–6. https://doi.org/10.1515/bpasts-2017-0045

30.

Norris

Hameed

Economou

, et al. (2020) A review of dual-spin projectile stability. Defence Technology 16(1): 1–9. https://doi.org/10.1016/j.dt.2019.06.003

31.

Peatie

Haroglu

Umar

(2024) Barriers to achieving satisfactory dropped objects safety performance in the UK construction sector. Heliyon 10(17): e37413. https://doi.org/10.1016/j.heliyon.2024.e37413

32.

Qin

Wang

(2019) Triangulation learning network: from monocular to stereo 3d object detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 7615–7623.

33.

Ramírez

AGR

Luna

FJG

Villegas

OOV

, et al. (2018) Applications of haptic systems in virtual environments: a brief', advanced topics on computer vision. Control and Robotics in Mechatronics 2018(1): 349.

34.

Rosenberg

Surujon

Yeshurun

, et al. (2005) Ricochet of 0.3″ AP projectile from inclined polymeric plates. International Journal of Impact Engineering 31(3): 221–233. https://doi.org/10.1016/j.ijimpeng.2003.12.008

35.

Scannapieco

Renga

Fasano

, et al. (2016) Ultralight radar sensor for autonomous operations by micro-UAS. In: Proceedings from the International Conference on Unmanned Aircraft Systems (ICUAS), pp. 727–735.

36.

Shtarbanov

(2018) Airtap: A Multimodal Interactive Interface Platform with free-space Cutaneous Haptic Feedback via Toroidal air-vortices. Cambridge, MA, USA. Massachusetts Institute of Technology.

37.

Shusser

Gharib

(2000) Energy and velocity of a forming vortex ring. Physics of Fluids 12(3): 618–621. https://doi.org/10.1063/1.870268

38.

Slegers

(2008) Predictive control of a munition using low-speed linear theory. Journal of Guidance, Control, and Dynamics 31(3): 768–775. https://doi.org/10.2514/1.34329

39.

Sockalingam

Chowdhury

Gillespie Jr

, et al. (2017) Recent advances in modeling and experiments of kevlar ballistic fibrils, fibers, yarns and flexible woven textile Fabrics–A review. Textile Research Journal 87(8): 984–1010. https://doi.org/10.1177/0040517516646039

40.

Sodhi

Poupyrev

Glisson

, et al. (2013) AIREAL: interactive tactile experiences in free air. ACM Transactions on Graphics (TOG) 32(4): 1–10. https://doi.org/10.1145/2461912.2462007

41.

Taddeucci

Alatorre-Ibargüengoitia

Cruz-Vázquez

, et al. (2017) In-flight dynamics of volcanic ballistic projectiles. Reviews of Geophysics 55(3): 675–718. https://doi.org/10.1002/2017rg000564

42.

Thompson

(1972) 'Coriolis deflection of a ballistic projectile. American Journal of Physics 40(10): 1477–1483. https://doi.org/10.1119/1.1986873

43.

Weinrotter

Kopecek

Tesch

, et al. (2005) Laser ignition of ultra-lean methane/hydrogen/air mixtures at high temperature and pressure. Experimental Thermal and Fluid Science 29(5): 569–577. https://doi.org/10.1016/j.expthermflusci.2004.08.002

44.

Wickert

(2006) 'Electric armor against shaped charges: analysis of jet distortion with respect to jet dynamics and current flow. IEEE Transactions on Magnetics 43(1): 426–429. https://doi.org/10.1109/tmag.2006.887650

45.

Xue

Ren

, et al. (2022) Design, modeling and implementation of a projectile-based mechanism for USVs charging tasks. IEEE Robotics and Automation Letters 8(1): 288–295. https://doi.org/10.1109/lra.2022.3154036

46.

Yang

Wang

Chen

(2021) Movement characteristics of a dual-spin guided projectile subjected to a lateral impulse. Aerospace 8(10): 309. https://doi.org/10.3390/aerospace8100309