Integrated wildlife detection, deterrence, and avoidance system

An integrated system with a multi-sensor array, AI processor, and deterrent signal generation provides real-time wildlife detection and species-specific deterrence, improving safety by coordinating evasive maneuvers for low-altitude aviation.

WO2026241291A1PCT designated stage Publication Date: 2026-11-19MANTA LLC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/US2026/027843
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-12
Filing Date
2026-05-12
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing wildlife deterrent technologies are inadequate for airborne platforms, lacking real-time detection, species-specific deterrence, and coordinated evasive maneuver recommendations, posing a significant safety risk for low-altitude aviation operations.

Method used

An integrated system comprising a multi-sensor array, AI processor, deterrent signal generation, and flight control interface that detects wildlife, identifies species, emits species-specific deterrent signals, and provides evasive maneuver recommendations autonomously.

Benefits of technology

The system effectively detects and deters wildlife in real-time, adapting deterrent signals based on species behavior and environmental conditions, and coordinates evasive maneuvers, enhancing safety for low-altitude aviation operations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure US2026027843_19112026_PF_FP_ABST
    Figure US2026027843_19112026_PF_FP_ABST
Patent Text Reader

Abstract

A system for wildlife detection and deterrence for aircraft is disclosed. The system includes a multi-sensor array comprising a thermal imaging sensor and an optical imaging sensor mounted to an aircraft, an onboard artificial intelligence processor configured to run one or more machine learning models that detect wildlife presence in real time and identify wildlife species based on data from the multi-sensor array, a deterrent signal generation system configured to emit species specific ultrasonic and / or infrasonic signals to deter identified wildlife with self-calibrating output based on detected wildlife behavior, and a flight control interface configured to provide alert information and suggest evasive maneuvers based on detected wildlife trajectories.
Need to check novelty before this filing date? Find Prior Art

Description

INTEGRATED WILDLIFE DETECTION, DETERRENCE, AND AVOIDANCE SYSTEMCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 804,267, filed May 12, 2025. The entire contents of the above-identified application are hereby fully incorporated herein by reference.FIELD OF THE INVENTION

[0002] The present invention relates to safety systems, and in particular, safety systems that detect and deter birds and other animals from both mobile and stationary platforms, including aerial vehicles, motor vehicles, and fixed installations such as wind turbines or landing zones, using optical detection systems with an Al-powered response and sound dispersal mechanism.BACKGROUND OF THE INVENTION

[0003] The rapid expansion of urban air mobility — including air taxis, delivery drones, and other electric vertical take-off and landing (eVTOL) aircraft — has created a substantial and growing safety gap in aviation. Unlike traditional commercial aircraft, which operate primarily at cruise altitudes above wildlife-dense airspace, next-generation aircraft operate below 10,000 feet, at relatively low airspeeds, and in suburban and urban corridors previously unoccupied by commercial aviation. These operating characteristics expose eVTOL aircraft to substantially higher wildlife encounter rates: lower speeds reduce the effectiveness of passive deflection, lower altitudes place aircraft within the primary flight and foraging zones of birds and other animals, and higher operational frequency in wildlife-dense environments increases cumulative encounter risk substantially compared to conventional aircraft.

[0004] Bird strikes and wildlife collisions with aircraft have historically caused thousands of incidents annually, resulting in aircraft damage, flight delays, and occasional fatal accidents. The consequences are particularly acute for smaller, lighter aircraft: a bird strike that a large commercial airliner can absorb may be catastrophic for an eVTOL platform with smaller engines, reduced structural margins, and no co-pilot redundancy. As eVTOL operations expand into wildlife-dense low-altitude corridors on high-frequency schedules, the aggregate probability of a severe wildlife encounter increases substantially relative to traditional commercial operations.

[0005] Existing wildlife deterrent technology is almost entirely ground-based and unsuitable for deployment on aircraft. Laser-based systems, which represent the dominant commercial category, cannot legally be mounted on airborne platforms. Pyrotechnic deterrents and sonic cannons, while effective at fixed installations such as airports and wind farms, are passive broadcast devices with no detection, identification, or targeting capability, and are impractical for integration onto aircraft. Radar-triggered alerting systems can detect large wildlife but do not identify species, do not generate deterrence signals, and do not interface with flight controls. No currently available system is capable of performing wildlife detection and active deterrence from the aircraft itself while simultaneously coordinating species-specific response recommendations with the flight control system. This gap leaves pilots without an integrated, onboard solution for detecting, identifying, deterring, and coordinating an evasive response to wildlife in the dynamic environment of low-altitude flight.

[0006] Therefore, there remains a need for an integrated, platform-independent system capable of detecting and identifying wildlife in real time, generating species-specific deterrence signals, predicting wildlife response to those signals, and — where deployed on a mobile platform — providing coordinated evasive maneuver recommendations to the vehicle's navigation or flight control system, all autonomously and without requiring operator intervention in the detection-to-response cycle. Such a system would be deployable on airborne platforms including eVTOL aircraft and UAVs, as well as on stationary installations such as wind turbines and landing zones, and on mobile ground platforms including motor vehicles.SUMMARY OF THE INVENTION

[0007] Disclosed are a system, a method, and a non-transitory computer-readable medium for wildlife detection and deterrence for aircraft, together with mobile-platform and stationary embodiments of the disclosed system, addressing the wildlife encounter exposure of low-altitude aviation operations.

[0008] In an embodiment, a system for wildlife detection and deterrence for aircraft may comprise a multi-sensor array comprising a thermal imaging sensor and an optical imaging sensor mounted to an aircraft; an onboard artificial intelligence processor configured to run one or more machine learning models that detect wildlife presence in real time and identify wildlife species based on data from the multi-sensor array; a deterrent signal generation system configured to emitultrasonic and / or infrasonic signals to deter identified wildlife, wherein the deterrent signal generation system self-calibrates its output based on detected wildlife behavior; and a flight control interface configured to provide alert information and suggest evasive maneuvers based on detected wildlife trajectories.

[0009] In an embodiment, the system may further comprise a data recording and learning module configured to record wildlife encounter data and use the recorded data to update and improve the system's machine learning models, optionally including a fleet-learning configuration in which recorded encounter data is transmitted to a remote server and model updates derived from the combined encounter history of a plurality of deployed systems are distributed back to individual units. In an embodiment, the system may further comprise an environmental data input configured to supply wind speed, wind direction, atmospheric temperature, and atmospheric pressure data for real-time recalibration of detection, deterrence, and flight control parameters, in which the artificial intelligence processor may be configured to adjust deterrent signal beam direction, frequency, and amplitude based on the current wind vector and to compute a predicted evasive trajectory for detected wildlife that accounts for wind-influenced movement patterns.

[0010] In an embodiment, a method for wildlife detection and deterrence for aircraft may comprise detecting wildlife in the vicinity of an aircraft using a multi-sensor array; running one or more machine learning models on an onboard processor to detect wildlife presence in real time; fusing thermal imaging data and optical imaging data into a unified sensor data set; identifying wildlife species and differentiating between species based on shape, thermal signature, and movement characteristics; emitting species-specific ultrasonic and / or infrasonic signals to deter identified wildlife with self-calibration based on detected wildlife behavior; and providing alert information and suggested evasive maneuvers through a flight control interface based on detected wildlife trajectories.

[0011] In an embodiment, a non-transitory computer-readable medium may store instructions that, when executed by a processor, cause the processor to perform a method for wildlife detection and deterrence for aircraft, the method comprising analyzing data from a multi-sensor array, running one or more machine learning models to identify wildlife species, fusing thermal and optical imaging data into a unified sensor data set, prioritizing threats, selecting species-specific deterrent frequencies, controlling a deterrent signal generation system with self-calibration, and providing alert information and suggested evasive maneuvers through a flight control interface.

[0012] In an embodiment, a mobile wildlife detection and deterrence system may comprise a multi-sensor array mounted to a mobile platform, an artificial intelligence processor configured to identify detected wildlife and determine species-specific deterrent responses, a deterrent signal generation system, and a flight or navigation interface configured to issue alerts or control signals to the vehicle in response to wildlife threat data. In an embodiment, a stationary wildlife detection and deterrence system may comprise a multi-sensor array mounted on a fixed structure, an artificial intelligence processor configured to detect and identify wildlife approaching the structure, a sound emitter configured to emit ultrasonic or infrasonic signals in response to detected threats, and a data logging module configured to record detection events and system responses, optionally configured to transmit detection event data to an urban air traffic management system or centralized airspace management platform to enable airspace-level wildlife hazard awareness across multiple aircraft.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIG. l isa block diagram depicting an aircraft wildlife detection and deterrence system. Elements shown in dashed lines represent optional components that are not required elements of the broadest independent claim.

[0014] FIG. 2 illustrates the placement of sensors and deterrent signal generators on an example aircraft.

[0015] FIG. 3 is a block flow diagram depicting a wildlife detection process flow.

[0016] FIG. 4 is a block flow diagram depicting a wildlife identification and deterrence process flow.

[0017] FIG. 5 is a block flow diagram depicting a flight control integration process flow.

[0018] FIG. 6 is a block diagram depicting an intelligent animal detection and deterrence system.

[0019] FIG. 7 is a block flow diagram depicting an intelligent animal detection and deterrence method.

[0020] FIG. 8 depicts a block diagram of a computing machine and a module.DETAILED DESCRIPTION

[0021] The embodiments disclosed herein address wildlife encounter exposure that arises as aircraft operate at lower altitudes and at lower airspeeds. Wildlife, as used herein, refers broadly to any non-human animal that may enter or occupy airspace, a runway environment, or other operational area of an aircraft or fixed infrastructure, and may include, without limitation, wild birds, raptors, waterfowl, and migratory species; terrestrial wild mammals; feral animals; and domesticated animals such as livestock, working animals, and companion animals that are present in or near the operational area, whether or not under human control at the time of detection. The embodiments disclosed herein may be applied including during the takeoff, initial climb, approach, and landing phases of commercial and general aviation operations, and throughout the primary operating envelope of air taxis, urban air mobility platforms, electric vertical take-off and landing (eVTOL) and vertical take-off and landing (VTOL) aircraft, ultralight aircraft, and unmanned aerial vehicles (UAVs). The technology disclosed herein may be deployed on any aircraft platform that operates within wildlife-dense low-altitude airspace, regardless of aircraft category or size. Wildlife strikes in these flight regimes can result in airframe damage, flight delays, and in some cases serious incidents, and approaches that have been described for managing wildlife encounters from ground-based platforms may not address the dynamic, in-flight conditions presented by such aircraft. The technology disclosed herein provides an integrated, onboard solution that may detect wildlife in the vicinity of an aircraft, identify the species of the detected wildlife, emit a speciesspecific deterrent signal, predict the post-deterrence response of the deterred wildlife, and coordinate a corresponding evasive maneuver through the aircraft's flight control system, in a single autonomous loop.

[0022] In an embodiment, the disclosed system may include multiple integrated subsystems operating together onboard an aircraft. A multi-spectral sensing system may include a multi-sensor array comprising at least a thermal imaging sensor and a high-definition optical camera, and may further include additional sensors selected from radar, sonar, ultraviolet, microwave, and radiowave sensors. An artificial intelligence processor may run one or more machine learning architectures, including convolutional neural networks, trained to identify wildlife species from fused thermal and optical sensor data, to assign behavior scores reflecting proximity, trajectory, and species characteristics, and to predict the response of an identified species to a particular deterrent signal. A deterrent signal emission system may include an array of speakers configuredto emit ultrasonic and infrasonic signals in species-specific frequency ranges, and may selfcalibrate the volume, tone, frequency, and pattern of the emitted signals in real time based on the observed behavioral response of the detected wildlife. A flight control interface may receive the deterrence output and the predicted post-deterrence wildlife trajectory and may either provide an alert to the pilot with a suggested evasive maneuver or, where the airframe permits, provide direct inputs to the aircraft's flight control system for automatic evasive maneuvering. A data recording and learning module may capture data describing each wildlife encounter and may feed the captured data back to the artificial intelligence processor for refinement of subsequent identification, deterrence, and evasive performance.

[0023] The disclosed technology may offer several advantages over approaches that have been described. By relying on directional ultrasonic and infrasonic acoustic deterrence, the disclosed technology may be deployed from the aircraft itself in flight, and the directional emission cone may concentrate deterrent energy on the target wildlife while limiting acoustic energy directed toward non-target areas. By identifying the species and selecting species-specific deterrent frequencies — and, when more than one species is concurrently identified, by pulsating through multiple species-specific frequency ranges — the disclosed technology may achieve a behavioral response calibrated to the identified species, and the closed-loop self-calibration mechanism may further refine the emitted signal in real time as the wildlife response is observed. By coordinating the predicted post-deterrence response of the deterred wildlife with a specific evasive maneuver communicated to the pilot or directly to the flight control system, the disclosed technology may close the safety loop in a manner that detection-only or detection-plus-deterrence approaches may not. By fusing thermal and optical imagery into a unified sensor data set prior to species identification, the disclosed technology may achieve effective detection in low-light, nighttime, and adverse-weather conditions in which optical-only detection may be degraded, and by performing real-time recalibration for wind, atmospheric temperature, and atmospheric pressure, the disclosed technology may maintain detection and deterrence performance across a range of operating environments. By recording each wildlife encounter and the corresponding deterrent and evasive outcomes, the disclosed technology may improve its identification and response performance over the course of operational deployment.

[0024] The technology is described in the sections that follow through several embodiments. An aircraft-mounted wildlife detection and deterrence system embodiment is described withreference to FIGS. 1 and 2, and the operation of that embodiment is described through a multistage process with reference to FIGS. 3 through 5. A further intelligent animal detection and deterrence system embodiment, presented in a simplified configuration, is described with reference to FIGS. 6 and 7. A method of wildlife detection and deterrence is also described, as is a non-transitory computer-readable medium storing instructions that, when executed, may cause one or more processors to perform the method. Mobile-platform and fixed-structure embodiments suited to ground vehicles, marine vessels, wind turbines, eVTOL landing zones, and analogous deployments are also described.WILDLIFE DETECTION AND DETERRENCE SYSTEM

[0025] In an embodiment, the disclosed system may provide an airborne, integrated solution to the wildlife detection and deterrence. The system 100 may combine multi-spectral sensing, AI-based species identification, directional acoustic deterrence, environmental adaptation, and flight control coordination in a single, lightweight, onboard architecture that may operate autonomously without requiring pilot intervention in the detection-to-response cycle. The system may offer several advantages over approaches that have been described, including airborne deployability without the legal restrictions associated with laser-based systems, species-specific deterrence adapted in real time to observed animal behavior, and a coordinated evasive maneuver recommendation that accounts for the predicted post-deterrence trajectory of the identified wildlife. In an example embodiment, the system may comprise six integrated subsystems, each corresponding to a functional element of the claimed system. The following paragraphs introduce each subsystem and its role within the overall architecture. The subsystems are described in further detail in the sections that follow.Multi-Sensor Array

[0026] Referring to FIG. 1, in an embodiment, the system 100 may include a multi-sensor array comprising at least a thermal imaging sensor 112 and an optical imaging sensor 114, both mounted to the aircraft. The thermal imaging sensor 112 may detect heat signatures emitted by warm -bodied animals in the surrounding flight environment, enabling detection at long range and under low-light or nighttime conditions. The optical imaging sensor 114 may provide high-definition visual data to the Al processor for species-level identification, capturing shape, outline, and movement patterns of detected animals in the visible spectrum. The two sensors may operate in a fused configuration in which the thermal channel may provide coarse detection cues and theoptical channel may provide fine-grained data for species differentiation. Their combination, processed by the Al processor as described herein, may produce detection and classification performance across varied lighting and environmental conditions that may not be achievable by either sensor operating independently.

[0027] In an embodiment, the multi-sensor array may be configured in a forward-facing arrangement in which the thermal imaging sensor 112 and the optical imaging sensor 114 are oriented to detect wildlife within the primary flight corridor ahead of the aircraft. In an example embodiment, the multi-sensor array may alternatively be configured as a 360-degree array in which sensors are positioned to provide continuous detection coverage of the full flight environment surrounding the aircraft, providing full flight awareness regardless of the direction of an approaching wildlife threat.

[0028] In an embodiment, the thermal imaging sensor 112 may serve as the primary detection sensor during night operation and under conditions of reduced ambient light, in which the optical imaging sensor 114 may have reduced detection capability due to insufficient illumination. The multi-sensor array may be configured for all-weather, all -conditions detection operation, and the fused sensor data set produced by the thermal imaging sensor 112 and the optical imaging sensor 114 may enable the system 100 to maintain detection and species identification performance across a range of environmental conditions including low light, darkness, fog, and precipitation.

[0029] In an embodiment, the fused dual-channel architecture of the multi-sensor array may be particularly suited to the operating conditions of low-altitude aviation, including eVTOL and UAV operations. Thermal-only detection at low altitudes may produce elevated false-positive rates because warm ground surfaces, vehicle engines, building HVAC exhaust, and other nonwildlife thermal sources may generate heat signatures within the detection field that are not readily distinguishable from warm-bodied animals at the resolution and range relevant to eVTOL operations. Optical-only detection may perform adequately in full daylight but may exhibit reduced detection reliability under low-light conditions, at dawn and dusk, in overcast conditions, and against backgrounds with low contrast between the animal and its surroundings, which are common conditions in the low-altitude suburban and urban corridors where eVTOL aircraft frequently operate. The fused configuration may address both limitations simultaneously: the thermal channel may flag candidate detection events in the field of view, and the optical channel may provide the visual resolution and species-differentiating detail needed to confirm thedetection, classify the species, and reject non-wildlife thermal sources. Because the thermal channel functions as a spatial gate rather than a classifier, it may not be required to distinguish between a warm-bodied animal and a non-wildlife thermal source in low-light conditions; that discrimination may be performed by the optical classifier operating on the thermally-gated region of interest. Under low-light conditions where full-frame optical analysis may exhibit reduced reliability, restricting the optical classification model to a spatially gated sub-region defined by the thermal detection event may substantially reduce the classification task from a full-scene analysis to a localized region where a candidate thermal source has already been identified, so that the optical model may apply its classification resources to the specific image region of interest rather than the full frame, improving both processing efficiency and classification accuracy under the low-light conditions in which unaided optical detection may be less reliable. In an example embodiment, the data fusion module may apply the thermal channel output as a spatial attention mask that directs the optical classification model to focus computational resources on regions of the image where a thermal anomaly has been detected, reducing the processing load associated with full-frame optical analysis while maintaining the classification accuracy needed for speciesspecific deterrence signal selection.

[0030] In an embodiment, the system 100 may further include a data fusion module, which as used herein refers to a processing component configured to merge thermal imaging data from the thermal imaging sensor 112 and optical imaging data from the optical imaging sensor 114 into a unified sensor data set prior to species identification by the Al processor 120. The data fusion module may combine the complementary spatial, spectral, and temporal information from both sensor channels so that the Al processor 120 may analyze a single, integrated data representation rather than processing each sensor stream independently. In an example embodiment, the data fusion module may reside within or communicate with the Al processor 120 and may output a fused data representation to the species identification and classification functions of the Al processor 120.

[0031] In an embodiment, the multi-sensor array 110 may further include one or more supplemental sensing modalities in addition to the thermal imaging sensor 112 and the optical imaging sensor 114, selected to expand the system's detection envelope under conditions where thermal and optical sensing alone may have reduced effectiveness. In an example embodiment, the multi-sensor array 110 may include a radar sensor configured to detect the presence and rangeof wildlife at extended distances and in conditions of reduced optical and thermal visibility, including dense fog, heavy precipitation, and complete darkness, in which the radar return from a large bird or flock may provide a detection cue at ranges beyond the effective detection distance of the thermal channel. In an example embodiment, the multi-sensor array 110 may include one or more sonar or ultrasonic ranging sensors configured to provide precision short-range distance measurements to detected objects, complementing the longer-range detection capability of the thermal and radar channels with close-proximity ranging accuracy relevant to final deterrence timing. In an example embodiment, the multi-sensor array 110 may include an ultraviolet imaging sensor configured to capture ultraviolet reflectance patterns from detected wildlife, which may enable species differentiation based on UV-reflective plumage or body surface characteristics that are not detectable in the visible or thermal spectrum and that may distinguish between species that appear substantially similar in optical and thermal imagery. In an example embodiment, the multisensor array 110 may include one or more microwave or radio-wave sensors configured to detect the presence and movement of wildlife based on reflected electromagnetic energy in frequency ranges that penetrate atmospheric conditions, including precipitation and fog, that may attenuate optical and thermal sensor performance. Where one or more supplemental sensing modalities are present, the data fusion module may be configured to incorporate the output of each supplemental sensor into the unified sensor data set alongside the thermal and optical sensor streams, so that the Al processor 120 may analyze fused multi-modal sensor data that reflects the combined detection input of all active sensing channels.

[0032] Example technical parameters of the thermal and high-definition cameras are provided below. The parameters disclosed include resolution, thermal sensitivity, lens characteristics, frame rate, field of view, detection capability, and low-light operational performance.Parameter Thermal Camera HD Camera Resolution 640 X 512 1920 X 1080 or higher Thermal Sensitivity < 50 ml< N / A(NETD)Lens 13 -25 mm N / AFrame Rate N / A 30 - 60 Frames per second Field of View N / A 90 - 120 degrees Detection Capability High signature differentiation at long N / ArangeLow-Light Primary sensor for night operation Enhanced night operationCapabilityOnboard Al Processor

[0033] In an embodiment, the system 100 may include an onboard artificial intelligence processor 120 configured to run a machine learning algorithm configured to detect wildlife presence in real time and identify wildlife species based on data from the multi-sensor array. The Al processor 120 may communicate with the multi-sensor array 110, deterrent signal emission system 130, flight control interface 140, environmental sensors 160, and data recording and learning module 150 via a wired connection, a wireless connection, or any suitable combination thereof, and the communication link may be configured to transmit sensor data from the multisensor array to the Al processor 120 at a rate sufficient to support real-time detection and species identification. In an example embodiment, the communication link may operate at a data rate of at least approximately 100 Mbps, and may support data rates of 1 Gbps or higher, depending on the sensor configuration, compression format, and frame rate employed.

[0034] The Al processor 120 may execute one or more machine learning models trained on datasets containing the visual and thermal characteristics, movement patterns, and behavioral profiles of wildlife species. As used herein, "movement patterns" refers to the spatiotemporal characteristics of animal motion observable from sensor data, including wingbeat frequency and amplitude, flight velocity and acceleration, trajectory shape, flock geometry and cohesion, and micro-oscillatory body movements characteristic of particular species. As used herein, "behavioral profiles" refers to species-level behavioral characteristics relevant to threat assessment and deterrence response prediction, including typical response to threat stimuli, species-characteristic altitude and approach angle distributions, flock response dynamics, and historical deterrence response data associating particular species with effective deterrent signal parameters.

[0035] In an example embodiment, the machine learning models may comprise convolutional neural networks; however, other deep learning architectures may also be used, including but not limited to vision transformers, recurrent neural networks, long short-term memory networks, temporal convolutional networks, graph neural networks, or any combination of machine learning methods suitable for real-time image classification and behavioral prediction. In an example embodiment, the Al processor 120 may employ a hybrid architecture in which a convolutional feature extractor processes spatial image data from the optical and thermal sensors and a transformer-based attention module processes temporal behavioral sequences, enabling simultaneous spatial classification and temporal behavioral prediction within a single inferencepass. In an example embodiment, the Al processor 120 may run an ensemble of machine learning models in parallel, combining the outputs of independently trained classifiers to improve species identification confidence under conditions where a single model may produce low-confidence predictions, such as partial occlusion, non-canonical viewing angles, or novel species not well-represented in the training dataset. In an example embodiment, the machine learning models may be selected from lightweight, edge-optimized architectures, including but not limited to MobileNet-class, EfficientDet-class, or YOLO-class models, configured to operate within the compute and power budget of the onboard Al processor 120 while maintaining the inference latency required for real-time wildlife detection and deterrence response.

[0036] In an embodiment, the selection of a machine learning architecture for the Al processor 120 may involve consideration of one or more of the following factors. First, inference latency: the end-to-end detection-to-deterrence response cycle of the system 100 may impose a latency constraint on the machine learning pipeline, and architectures that achieve classification within the latency budget of the system may be selected over architectures that achieve higher accuracy at the cost of response time. Second, input modality fusion strategy: because the multi-sensor array 110 may supply both thermal and optical data simultaneously, the machine learning architecture may be configured for early fusion, in which thermal and optical inputs are combined prior to the first processing layer; late fusion, in which separate models process each modality and their outputs are combined at the classification stage; or intermediate fusion, in which a shared backbone processes both modalities up to a certain depth and modality-specific processing heads diverge thereafter. Third, training data availability: wildlife encounter data collected from airborne platforms operating at low altitudes may be limited in volume relative to ground-based wildlife datasets, and architectures with strong transfer learning characteristics or few-shot learning capability may be selected to enable effective model training from smaller labeled datasets. Fourth, hardware resource constraints: the machine learning architecture may be selected to operate within the compute throughput, memory bandwidth, and power consumption profile of the onboard Al processor 120, such that real-time inference may be sustained across the full operational envelope of the aircraft platform on which the system is installed.

[0037] The Al processor 120 may perform species differentiation, assign behavior scores reflecting proximity and trajectory, determine the threat level of detected animals, and select deterrent signals appropriate to the identified species. In an example embodiment, the Al processor120 may perform approximately 72 trillion computations per second and may weigh less than approximately 300 grams, making the system suitable for installation on lightweight aircraft including eVTOL vehicles, UAVs, and ultralight aircraft.

[0038] In an embodiment, the Al processor 120 may comprise one or more machine learning models trained to detect animal signatures and differentiate between wildlife species based on shape, thermal signature, visual outlines, and movement characteristics. In an example embodiment, the machine learning models may comprise convolutional neural networks; in other embodiments, vision transformers, recurrent neural networks, or other deep learning architectures configured for image-based wildlife classification may be used. The machine learning models may be trained on datasets containing labeled wildlife imagery and thermal data representing a range of species, environmental conditions, distances, and behavioral states, enabling the Al processor 120 to identify detected animals with a calculated confidence level and to select from a probability distribution of possible species identifications when a definitive single-species determination may not be achievable.

[0039] In an embodiment, the Al processor 120 may be further configured to determine the range, trajectory, and threat level of detected wildlife, and to predict the response of identified wildlife to deterrent signals. As used herein, "threat level" refers to a computed assessment of the probability and severity of a collision event between the aircraft and detected wildlife if no deterrence or evasive action is taken, based on one or more factors including the current distance and closing rate of the detected wildlife, the projected intersection of the wildlife's trajectory with the aircraft's flight path, the species characteristics of the detected wildlife including mass and maneuverability, and the estimated time to potential intercept. The Al processor 120 may prioritize threats based on proximity, trajectory, and species characteristics, and may assign behavior scores reflecting the likelihood and severity of a collision event. In an example embodiment, the Al processor 120 may use the predicted wildlife response to a given deterrent signal — including the expected direction and magnitude of evasive movement — as an input to the evasive maneuver recommendation generated by the flight control interface 140, so that the suggested pilot maneuver may be coordinated with the expected post-deterrence trajectory of the detected wildlife.

[0040] In an embodiment, the Al processor 120 may be configured to identify multiple wildlife species simultaneously when more than one species is detected within the monitored flight environment. In an example embodiment, the Al processor 120 may concurrently maintainindependent identification threads for each detected animal or animal group, assigning separate behavior scores and threat assessments, and may direct the deterrent signal generation system 130 to emit multiple species-specific deterrent signals simultaneously or in rapid alternation to address all identified threats within a single response cycle.Deterrent Signal Generation System

[0041] In an embodiment, the system 100 may include a deterrent signal generation system 130 configured to emit ultrasonic and / or infrasonic signals to deter identified wildlife, wherein the output parameters of the deterrent signal generation system 130 may be dynamically adjusted by the Al processor 120 based on observed wildlife behavioral response. In certain embodiments, the deterrent signal generation system 130 may also be referred to as a directional sound emitter. The deterrent signal generation system 130 may include an array of speakers capable of producing a wide range of frequencies, including ultrasonic and infrasonic frequencies. The Al processor 120 may select deterrent frequencies based on the species identified, and may direct the deterrent signal generation system 130 to emit a single species-specific signal, emit multiple signals targeting one or more identified species simultaneously, or pulsate through frequency ranges to cover the deterrence bands of all concurrently detected species. The Al processor 120 may continuously monitor observed animal behavior via the multi-sensor array and may adjust the output parameters of the deterrent signal generation system 130 — including volume, tone, frequency, and modulation pattern — in real time, enabling the system 100 to escalate or modify the deterrent signal based on observed behavioral response.

[0042] The deterrent signal generation system 130 may include feedback-driven calibration that may adjust volume, tone, frequency, and pattern based on observed animal reactions. In an example embodiment, the feedback-driven calibration may operate by continuously monitoring the behavioral state of detected wildlife — including flight trajectory, velocity, and direction — and comparing the observed behavioral response against an expected response profde stored in the Al processor 120 for the identified species, adjusting one or more signal output parameters when the observed response deviates from the expected profile.

[0043] In an embodiment, the self-calibration mechanism of the deterrent signal generation system 130 may not only determine whether to escalate deterrent output, but may also determine which output parameter to adjust based on the species-specific response mode of the identified wildlife. As used herein, "response mode" refers to the primary behavioral mechanism by whicha given species responds to deterrent stimuli, which may vary across species and may influence which signal parameter is most effective in producing a trajectory change. In an example embodiment, species such as Canada Geese, which may respond more reliably to directional sound concentration than to broadband signal amplitude, may cause the self-calibration mechanism to prioritize beam steering adjustment and directional targeting before escalating overall amplitude. In another example embodiment, species such as European Starlings, which may exhibit heightened response to pulse modulation, may cause the self-calibration mechanism to prioritize pulse rate adjustment before escalating frequency range. In another example embodiment, species such as Pigeons, for which rapid frequency sweeps across the operative range may produce improved avoidance behavior, may cause the self-calibration mechanism to prioritize frequency sweep rate before escalating amplitude. In an embodiment, where multiple species are identified simultaneously, the self-calibration mechanism may apply species-specific escalation sequences independently for each identified species, and the deterrent signal generation system 130 may execute the resulting multi-species escalation protocol concurrently. The species-specific response mode data may be stored as a parameter of the behavioral response model of the Al processor 120 and may be updated by the data recording and learning module 150 as encounter data accumulates.

[0044] In an embodiment, the deterrent signal generation system 130 may use either a stationary 360-degree array in which speakers are distributed around the aircraft to provide omnidirectional deterrent coverage, or a servo-controlled or gimbaled directional array in which one or more speakers may be mechanically oriented toward detected wildlife to concentrate deterrent energy within a directional sound cone. In an example embodiment, the deterrent signal generation system 130 may steer the direction of deterrent signal emission through electronic beam steering without mechanical actuation, in which the relative phase and amplitude of signals emitted from multiple speakers in the array may be adjusted to produce constructive interference in the direction of the target wildlife and destructive interference in non-target directions.

[0045] In an embodiment, the deterrent signal generation system 130 may comprise an array of between one and more than six speakers positioned at one or more locations on the aircraft, including one or more of a bottom surface, a top surface, or a nose of the aircraft. In an example embodiment, a typical configuration may comprise between five and six speakers, though the number of speakers and their placement may be customized based on the specific airframe type and the operational requirements of the deployment.

[0046] In an embodiment, the deterrent signal generation system 130 may be configured to emit deterrent signals in species-specific frequency ranges corresponding to the identified species. The species-specific frequency ranges may be determined from training data and behavioral response data associated with each identified species, and may be stored as a lookup table or model parameter accessible to the Al processor 120. In an example embodiment, the species-specific frequency ranges may include ultrasonic frequencies in the range of approximately 18-35 kHz for white-tailed deer, frequencies of approximately 2-6 kHz plus harmonics for Canada geese, approximately 3-8 kHz for European starlings, approximately 1-5 kHz for pigeons, and approximately 2-7 kHz for seagulls, though the disclosed system is not limited to these species or frequency ranges.

[0047] In an embodiment, the self-calibration of the deterrent signal generation system 130 may comprise monitoring the behavioral response of detected wildlife to emitted deterrent signals and determining whether the wildlife trajectory has changed away from a protected zone, which as used herein refers to a defined volume of airspace around the aircraft that the system 100 is configured to keep clear of detected wildlife and that may comprise the aircraft's current flight path, a buffer volume around the aircraft, and the airspace immediately ahead of the aircraft along its trajectory. In an example embodiment, if no trajectory change is detected within a monitoring interval, the self-calibration mechanism may escalate one or more output parameters, which may include: a frequency adjustment within a range of approximately ±5 to 20% of the current frequency; a pulse rate adjustment; an amplitude adjustment within safe limits; and a beam steering adjustment to improve directional targeting. The self-calibration mechanism may iterate the escalation sequence until the detected wildlife is redirected away from the aircraft's flight path or exits an engagement range, which as used herein refers to a range within which the deterrent signal generation system 130 may produce an effective deterrent acoustic effect on detected wildlife, as a function of speaker output power, atmospheric conditions, and species-specific deterrence sensitivity.

[0048] The following table presents example deterrent frequency ranges that have been used or tested for the common threat species addressed by the system, including Canada Geese, European Starlings, White-tailed Deer, Pigeons and Seagulls. This data corresponds to the performance results described in “Performance detection data working example” Section below.Frequency ranges for species-specific deterrentsSpecies Frequency Range NotesWhite-tailed Deer 18-35 kHz High sensitivity to ultrasonic bursts Canada Geese 2-6 kHz + Harmonics Directional sound more effective than broadbandEuropean Starling 3-8 kHz Pulsed modulation increases response Pigeon 1-5 kHz Rapid sweep improves avoidanceSeagull 2-7 kHz Directional targeting highly effective

[0049] The system selects deterrent frequencies based on species classification and dynamically adjusts frequency, amplitude, and modulation pattern in response to observed behavioral feedback. Species exhibiting lower sensitivity to broadband signals - such as Canada Geese - respond more reliably to directional sound delivery, while species such as European Starlings show heightened response to pulse modulation. For species such as Pigeons, rapid frequency sweeps across the operative range produce improved avoidance behavior. For Seagulls, directional targeting of the deterrent output is identified as a substantial factor in effectiveness.Flight Control Interface

[0050] In an embodiment, the system 100 may include a flight control interface 140 configured to provide alert information and suggest evasive maneuvers based on detected wildlife trajectories. The flight control interface 140 may receive the Al processor's output, including the species identification result, the predicted trajectory of the wildlife, and the predicted behavioral response of the identified species to the emitted deterrent signal, and may use that output to compute a specific evasive maneuver recommendation configured to direct the aircraft in a direction away from the predicted post-deterrence trajectory of the wildlife, increasing the separation distance between the aircraft and the wildlife. The flight control interface 140 may communicate this recommendation to the pilot through a visual and audio alert, or, where the aircraft permits direct integration, may transmit the evasive maneuver as a direct input to the aircraft's flight control system for automatic execution. The coordination between predicted wildlife response and specific flight maneuver recommendation may provide a safety outcome that approaches that have been described may not achieve.

[0051] In an embodiment, the predicted post-deterrence trajectory of the detected wildlife may be generated by the Al processor 120 using a species-specific behavioral response model trained on observed deterrence response data for each identified species. The behavioral response model may associate each species with one or more characteristic evasive movement patterns that thespecies exhibits when exposed to an effective deterrent stimulus — including the expected direction of evasive movement relative to the deterrent signal source, the characteristic altitude change, and the expected velocity and angular change of the evasive maneuver. In an example embodiment, a Canada Goose exposed to a directional deterrent signal at a range of approximately 50 meters may characteristically bank away from the signal source and descend; the system 100 may therefore compute an evasive maneuver recommendation that directs the aircraft to climb and turn in the direction opposite to the goose's predicted evasive bank. In another example embodiment, a European Starling flock exposed to a pulsed deterrent signal may characteristically scatter in multiple directions before reforming; the system 100 may compute a maneuver recommendation that increases altitude and reduces airspeed to allow the flock to disperse before the aircraft passes through the affected airspace. In an embodiment, where a definitive singlespecies identification may not be achievable and the Al processor 120 has generated a probability distribution over multiple candidate species, the behavioral response model may generate a weighted composite predicted trajectory reflecting the evasive movement characteristics of each candidate species weighted by its identification probability, and the evasive maneuver recommendation may be computed from the composite predicted trajectory. The behavioral response model may be updated by the data recording and learning module 150 as encounter data accumulates, so that the predicted post-deterrence trajectory for each species may be refined based on observed real-world deterrence outcomes.

[0052] In an embodiment, the flight control interface 140 may be further configured to provide direct inputs to the aircraft's flight control systems for automatic evasive maneuvering. In an example embodiment, where the aircraft's flight control architecture permits external control inputs, the flight control interface 140 may transmit computed evasive maneuver commands directly to the flight control system without requiring pilot action, enabling the system 100 to initiate an evasive response autonomously when a wildlife threat is detected and a deterrence response is initiated. The direct-input capability may operate in parallel with the pilot alert function, so that the pilot may simultaneously receive visual and audio notification of the threat and the system-initiated evasive response.

[0053] In an embodiment, the flight control interface 140 may comprise a pilot interface configured as a cockpit display or a heads-up display through which the pilot may receive visual and audio notifications of detected wildlife and suggested evasive maneuvers. In an exampleembodiment, the pilot interface may present the species identification result, the location and trajectory of detected wildlife relative to the aircraft, the emitted deterrent signal parameters, and the suggested evasive maneuver in a format accessible to the pilot during normal flight operations. The pilot interface may also provide confirmation of system status, deterrent signal emission, and self-calibration activity.Environmental Sensors

[0054] In an embodiment, the system 100 may include a plurality of environmental sensors 160 configured to measure wind speed, wind direction, atmospheric temperature, and atmospheric pressure, and to supply environmental data to the system for real-time recalibration of detection, deterrence, and flight control parameters. Environmental conditions may affect the propagation of acoustic deterrent signals, the reliability of sensor detection, and the accuracy of predicted evasive maneuvers. The environmental sensors 160 may provide the system with continuous environmental input so that signal parameters, detection thresholds, and maneuver calculations may be adjusted to reflect actual ambient conditions.

[0055] In an embodiment, the environmental data input to the system 100 may comprise dedicated environmental sensors 160 mounted to the aircraft and configured to measure environmental parameters directly. In another embodiment, the system 100 may include a data interface configured to receive environmental parameter data from pre-existing sensors already onboard the aircraft, such as avionics sensors or flight management system data feeds, in lieu of or in addition to dedicated environmental sensors 160. In an example embodiment, the data interface may be configured to receive environmental data in one or more standard avionics data formats, including ARINC 429, MIL-STD-1553, or any other suitable interface protocol supported by the host aircraft. This architecture may allow the system 100 to be installed on aircraft that already provide the required environmental data without requiring additional dedicated sensor hardware, reducing system weight and integration complexity.

[0056] In an embodiment, the system 100 may include environmental adaptability features comprising real-time recalibration for wind speed, wind direction, temperature, and atmospheric interference to maintain consistent detection and deterrence performance across all weather conditions. The environmental sensors 160 may supply updated ambient data to the Al processor 120 on a continuous basis, enabling the system 100 to adjust detection thresholds, deterrent signalpropagation models, and evasive maneuver calculations in real time as conditions change during flight.

[0057] In an embodiment, the system 100 may use real-time wind direction and wind speed data from the environmental data input to actively steer the deterrent signal output of the deterrent signal generation system 130 in a manner that accounts for acoustic propagation effects of the prevailing wind. As used herein, "wind-guided deterrence steering" refers to the adjustment of the beam direction, frequency, and amplitude of the deterrent signal output based on the current wind vector, so that the deterrent signal may be directed with the wind toward the detected wildlife to maximize acoustic energy delivery, and so that the projected evasive trajectory of the wildlife may be computed to take advantage of favorable wind patterns that guide the animal away from the protected zone. In an example embodiment, at an airfield where the prevailing wind is directed toward an open field adjacent to the runway, the system 100 may steer the deterrent signal output in the upwind direction relative to the detected wildlife, so that the acoustic energy is carried by the wind toward the animal, and may compute an evasive trajectory prediction in which the animal is expected to turn and flee in the downwind direction away from the protected airspace. In an embodiment, the Al processor 120 may model the acoustic propagation of the deterrent signal through the current wind field, including the effects of wind speed, wind direction, temperature gradient, and atmospheric pressure on sound propagation velocity and directional spread, and may adjust the beam steering, amplitude, and frequency of the deterrent signal output to account for the modeled propagation conditions, so that the delivered acoustic energy at the location of the detected wildlife may be maintained within the effective deterrence range regardless of wind conditions. This wind-aware deterrence capability may differentiate the system 100 from static scare systems that emit a fixed-direction or omnidirectional deterrent output without accounting for the acoustic propagation effects of ambient wind on the effective deterrence range and direction.

[0058] In an embodiment, the system 100 may be configured not only to disperse detected wildlife but to guide detected wildlife along a non-hazardous trajectory around one or more protected zones. As used herein, a “non-hazardous trajectory” refers to a path of travel that does not intersect a protected zone (such as a runway approach corridor, a turbine swept-area, an active rail right-of-way, or other defined exclusion region) and that does not direct the wildlife into a secondary hazard such as adjacent traffic, energized infrastructure, or another protected zone. Inan example embodiment, the Al processor 120 may compute a target trajectory for the detected wildlife based on the location and geometry of the protected zone, the current position, velocity, and heading of the detected wildlife, species-specific behavioral response data, and prevailing environmental conditions including wind direction, wind speed, and terrain features. The deterrent signal generation system 130 may then select beam direction, frequency, amplitude, and pulse pattern calculated to shape the movement of the detected wildlife toward the target trajectory, rather than to disperse the wildlife indiscriminately. As the encounter progresses, the Al processor 120 may iteratively update the selected deterrent parameters in response to observed wildlife movement, escalating, attenuating, or redirecting the deterrent output as needed to maintain the wildlife along the target trajectory until the wildlife clears the protected zone.

[0059] In a further embodiment, the iterative guidance process described above may establish a dynamic safe pathway, referred to herein as an “adaptive corridor,” along which detected wildlife are routed around one or more protected zones. The adaptive corridor may be computed in real time based on the location of the protected zone or zones, the location of one or more adjacent protected zones (such that the corridor does not route wildlife from one protected zone into another), terrain features, and the prevailing wind field, and may be reshaped as wildlife position and environmental conditions change. In an example embodiment, at a wind farm comprising a plurality of turbines, the adaptive corridor may route a detected flock between adjacent turbine swept-areas along a path consistent with the prevailing wind, with the deterrent signal output from one or more system installations cooperatively steering the flock along the corridor. The adaptive-corridor guidance described herein may differentiate the system 100 from scare cannons, sonic broadcast units, and static repellents that have been described, which may disperse wildlife in unpredictable directions but do not shape wildlife trajectories or establish defined routing pathways around protected zones.Data Recording and Learning Module

[0060] In an embodiment, the system 100 may include a data recording and learning module 150 configured to record wildlife encounter data, including species identification results, deterrent signals emitted, wildlife behavioral responses, and the effectiveness of evasive maneuvers, and to use the recorded data to update and improve the machine learning models of the Al processor 120. The data recording and learning module 150 may create a continuous improvement cycle in which each operational encounter may contribute to the refinement of subsequent identification accuracy,deterrent signal selection, and evasive maneuver prediction, so that the accumulated encounter history of the system 100 may inform all subsequent performance. In an example embodiment, recorded encounter data may be stored in an on-device encrypted log accessible to the Al processor 120, and model updates may be performed during scheduled maintenance intervals or, where on-aircraft processing capacity permits, in a continuous on-device learning mode. In another embodiment, the data recording and learning module 150 may be configured to transmit recorded encounter data to a remote fleet-learning server, enabling model updates derived from the combined encounter history of multiple deployed systems to be distributed back to individual units, so that each system in a deployed fleet may benefit from the collective encounter experience of the entire fleet.General Design Considerations

[0061] In an embodiment, the system 100 may be housed in a modular housing adaptable to multiple airframes, which as used herein refers to an enclosure or mounting assembly configured to integrate the sensors, Al processor, speaker array, and flight control interface of the system 100 into a single installable unit that may be adapted for attachment to a range of aircraft types. The modular housing may be constructed from lightweight composite materials, and the system 100 may be suitable for integration with ultralight aircraft, unmanned aerial vehicles (UAVs), electric vertical take-off and landing (eVTOL) vehicles, and conventional passenger aircraft.

[0062] In an embodiment, the system 100 may be configured for deployment on platforms other than aircraft. In an example embodiment, the system 100 may be deployed on one or more of: a wind turbine or wind farm installation, in which the system 100 may detect and deter birds and other wildlife to reduce the risk of collision with turbine blades; a marine vessel, in which the system 100 may emit deterrent signals to deter marine wildlife within a defined radius around the vessel, including sharks, to create safe zones for swimmers or divers; a landing area for eVTOL aircraft, in which the system 100 may detect and deter non-flying wildlife such as dogs or deer from the landing zone; and a motor vehicle, including a car, truck, or motorcycle, in which the system 100 may detect and deter wildlife from the path of the vehicle. In further embodiments, the system 100 may be deployed at additional fixed-infrastructure locations susceptible to wildlife interference, including: an airport or airfield perimeter or runway approach zone, in which the system 100 may detect and deter birds and other wildlife from incursion paths used by departing or arriving aircraft; an electrical substation or transmission infrastructure installation, in which thesystem 100 may deter birds, rodents, and other wildlife from contact with energized equipment to reduce the risk of outages, equipment damage, and wildlife mortality; a rail yard or rail right-of-way, in which the system 100 may deter wildlife from approaching active trackage to reduce trainwildlife collisions and associated service disruptions; a marine dock or port facility, in which the system 100 may deter marine wildlife and shorebirds from berthing areas, cargo handling zones, and adjacent waters; a solar installation, in which the system 100 may deter wildlife that may damage panels or interfere with site operations; an agricultural facility, including a feedlot, orchard, vineyard, aquaculture site, or crop field, in which the system 100 may deter wildlife that may damage crops, predate livestock, or contaminate produce; and a roadway or highway corridor, in which the system 100 may be deployed at known wildlife-crossing locations to deter wildlife from entering the roadway and to reduce vehicle-wildlife collisions. The platform-invariant detection-identification-deterrence core described herein may remain substantially unchanged across these additional deployment contexts, with the species-specific frequency parameters and behavioral response model of the Al processor 120 updated to reflect the wildlife threat profile relevant to each deployment environment.

[0063] In an embodiment, the architecture of the system 100 may be understood as comprising a platform-invariant core and one or more platform-specific interface elements, enabling deployment across the range of mobile and stationary platforms described above. The platforminvariant core may comprise the multi-sensor array, the Al processor 120, and the deterrent signal generation system 130, which together constitute the detection-identification-deterrence loop and may operate identically regardless of the deployment platform. The platform-specific interface elements may vary depending on the deployment context: in an airborne deployment, the platformspecific interface may comprise the flight control interface 140, which may provide evasive maneuver recommendations to a pilot or to an automatic flight control system; in a stationary deployment such as a wind turbine installation or an eVTOL landing zone, the flight control interface 140 may be omitted or replaced with a fixed-installation alert interface configured to notify an operator or trigger a physical deterrent response; in a marine deployment, the platformspecific interface may comprise a navigation interface configured to issue alerts or suggest course corrections to the vessel's navigation system; and in a motor vehicle deployment, the platformspecific interface may comprise a driver alert or vehicle navigation interface configured to warn the driver or suggest an evasive steering or braking response. In each deployment context, theplatform-invariant detection-identification-deterrence core may remain substantially unchanged, and the acoustic deterrent output may be adapted to the wildlife threat profile relevant to the deployment environment — including non-airborne species such as sharks in marine contexts or deer and dogs in ground vehicle and landing zone contexts — by updating the species-specific frequency parameters and behavioral response model of the Al processor 120.

[0064] In an embodiment, a stationary deployment of the system 100 at an eVTOL landing area, airport, airfield, or other fixed aviation infrastructure location may be configured to function as a wildlife hazard sensor node within a broader airspace management network. In such an embodiment, the data recording and learning module 150 may be configured to transmit wildlife detection event data — including species identification results, detected trajectories, system response records, and deterrence outcome data — to an urban air traffic management system or centralized airspace management platform, which as used herein refers to any system or network infrastructure configured to coordinate flight operations, situational awareness, or hazard data across multiple aircraft operating in a defined airspace. The transmission of detection event data from one or more stationary system deployments to a UATM platform may enable airspace-level wildlife hazard awareness, in which approaching wildlife threats identified at a fixed installation may be communicated to aircraft operating in the surrounding airspace before those aircraft enter the detection range of their own onboard systems, providing earlier warning and a longer deterrence response window than onboard sensing alone may provide. In an example embodiment, a plurality of stationary system deployments at multiple locations within a metropolitan airspace may together form a distributed wildlife hazard sensor network, in which each installation may contribute detection event data to a shared airspace-level situational awareness picture accessible to aircraft operators and air traffic management systems overseeing operations in the monitored airspace.

[0065] In an embodiment, the integrated subsystems of the system 100 may operate as a closed-loop architecture in which the multi-sensor array may provide detection input; the Al processor may perform identification, threat assessment, and deterrent selection; the deterrent signal generation system may deliver a species-specific deterrent signal and self-calibrate based on observed animal response; the flight control interface may translate the predicted postdeterrence wildlife trajectory into a specific evasive maneuver recommendation; where present, the environmental data input may supply real-time ambient data for recalibration of systemoutputs; and where present, the data recording and learning module may capture encounter outcome data to improve future performance. The following sections describe each of these subsystems in further detail.

[0066] Referring to FIG. 2, an illustrative image 200 illustrates the physical placement of system components on an example aircraft 210. The aircraft 210 shown is representative of an eVTOL or advanced air mobility vehicle, and the placement of components depicted is illustrative of one embodiment, actual placement may vary depending on the airframe.

[0067] The sensors 230 are mounted on the body of the aircraft 210 and are positioned to provide continuous coverage of the surrounding flight environment. The sensors 230 may include a thermal imaging camera and a high-definition optical camera of the multispectral sensing system 110, and may be configured to detect heat signatures, visual outlines, and movement patterns of wildlife in the vicinity of the aircraft 210. The sensors 230 may be strategically positioned on the aircraft to provide optimal coverage. The sensors can be mounted forward-facing or in a 360-degree array configuration for full flight awareness. For example, thermal sensors may be positioned on the wings for the broadest view, while high-definition cameras may be positioned on the nose or underside of the aircraft for detailed observation. The specific configuration may be customized according to the particular aircraft type and operational requirements. The sensors 230 are connected to an onboard computing module containing the Al processing unit. In the embodiment depicted, the sensors 230 are positioned on or near the nose of the aircraft 210 to maximize forward-facing detection range, though alternative configurations, including 360-degree arrays are within the scope of this disclosure.

[0068] The speaker array 240 may be mounted on the underside or forward section of the aircraft 210 and may constitute the sound emission elements of the deterrent signal generation system 130. The speaker array 240 may be configured to emit directional ultrasonic and infrasonic signals outward from the aircraft 210, as represented by the sound waves 220 depicted radiating from the aircraft 210 in FIG. 2. The sound waves 220 illustrate the directional propagation of the deterrent acoustic output toward detected wildlife in the flight path or surrounding environment. The directionality of the speaker array 240 allows the system to target specific animals or zones without generating omnidirectional noise that could interfere with aircraft communications or cause unintended disturbance.

[0069] The Al processor 250 is housed within the body of the aircraft 210 and is the onboard computational unit responsible for receiving sensor data from the sensors 230, processing the data through the machine learning and neural network algorithms described herein, and directing the output of the speaker array 240 with the appropriate deterrent signal parameters. The Al processor 250 corresponds to the artificial intelligence processor 120 of the system 100 described with reference to FIG. 1. The integration of the sensors 230, the speaker array 240, and the Al processor 250 within a single airframe-mounted package enables compact, lightweight installation suitable for a broad range of aircraft types, as further described in the Physical Implementation section below.METHOD FOR WILDLIFE DETECTION AND DETERRENCE

[0070] FIG. 3 illustrates the wildlife detection process flow of the system 100. The process may begin at step 310, where the system 100 initiates operation. In an embodiment, the system 100 may perform continuous monitoring at step 320, in which the multi-sensor array comprising the thermal imaging sensor 112 and the optical imaging sensor 114 may continuously scan the surrounding flight environment for anomalies. At step 330, the system 100 may evaluate whether an anomaly has been detected. If no anomaly is detected, the process may return to step 320 and continue monitoring. If an anomaly is detected at step 330, the process may advance in parallel to step 340, in which the system 100 may perform thermal analysis of the detected anomaly using input from the thermal imaging sensor 350, and to step 360, in which the system 100 may perform optical analysis of the detected anomaly using input from the optical imaging sensor 370, which as used herein refers to the HD camera providing high-definition visual data to the Al processor 120. In an embodiment, the thermal analysis at step 340 and the optical analysis at step 360 may proceed concurrently, with the thermal channel providing coarse detection cues and the optical channel providing fine-grained visual data for subsequent species differentiation. At step 380, the system 100 may perform data fusion, combining the thermal imaging data and the optical imaging data from the dual-sensor array into a unified sensor data set. The unified sensor data set may then be passed to the identification process at step 390, which serves as the entry point to the wildlife identification and deterrence process flow illustrated in FIG. 4.

[0071] FIG. 4 illustrates the wildlife identification and deterrence process flow of the system 100. The process may begin at step 410, receiving the unified sensor data set from the detection process of FIG. 3 via step 390. At step 420, the Al processor 430 may perform speciesidentification, running one or more machine learning models to identify the detected wildlife based on shape, thermal signature, visual outlines, and movement characteristics of the detected anomaly. In an embodiment, the Al processor 430, which as used herein refers to the onboard artificial intelligence processor 120 responsible for all real-time classification and decisionmaking, may perform threat assessment at step 440, determining the range, trajectory, and threat level of the identified wildlife. The Al processor 430 may further assign behavior scores at step 450, reflecting the proximity, trajectory, and behavioral characteristics of the identified wildlife relative to the aircraft. At step 460, the system 100 may determine whether multiple wildlife species have been identified. If only a single species has been identified, the process may advance to step 470, in which the deterrent signal generation system 130 may emit a single species-specific deterrent signal. If multiple species have been identified at step 460, the process may advance to step 480, in which the deterrent signal generation system 130 may emit multiple species-specific deterrent signals concurrently or in rapid alternation to address all identified species. In an embodiment, a self-calibration feedback loop may be provided at step 475, in which the speaker system of the deterrent signal generation system 130 may monitor the behavioral response of detected wildlife to the emitted deterrent signals and may adjust frequency, amplitude, pulse rate, and beam steering in real time based on observed behavioral response. Upon completion of deterrent signal emission and self-calibration, the process may advance to step 490, which serves as the entry point to the flight control integration process flow illustrated in FIG. 5.

[0072] FIG. 5 illustrates the flight control integration process flow of the system 100. The process may begin at step 510, receiving output from the deterrence process of FIG. 4 via step 490. At step 520, the system 100 may perform trajectory analysis, in which the Al processor 120 may analyze the predicted movement direction, speed, altitude change, and post-deterrence behavioral response of the identified wildlife to determine its expected trajectory relative to the aircraft. At step 530, the system 100 may calculate evasive maneuvers, computing one or more suggested flight path adjustments configured to direct the aircraft away from the predicted postdeterrence trajectory of the wildlife, thereby increasing the separation distance between the aircraft and the detected wildlife. The evasive maneuver calculation at step 530 may take into account the animal's predicted movement direction, speed, and altitude change, as well as the aircraft's current flight parameters, to generate actionable recommendations. At step 540, the system 100 may determine whether the aircraft is equipped for automatic flight control integration. If the aircraftdoes not support automatic flight control integration, the process may advance to step 560, in which the system 100 may generate a manual alert, and to step 565, in which the pilot interface 140 may present the alert information to the pilot through a cockpit display or heads-up display comprising visual and audio notifications of the detected wildlife, its species, location, and trajectory, the deterrent signals emitted, and the suggested evasive maneuver. If the aircraft supports automatic flight control integration at step 540, the process may advance to step 570, in which the system 100 may execute automatic flight control, and to step 575, in which the flight control system of the aircraft may receive the evasive maneuver as a direct input for automatic execution. In an embodiment, a feedback loop may be provided from the flight control system 575 back to the trajectory analysis step 520, enabling the system 100 to continuously update the predicted post-deterrence trajectory and recalculate suggested maneuvers as the aircraft responds and as the wildlife's actual post-deterrence behavior is observed. Following either the manual alert path or the automatic control path, the process may advance to step 580, in which the system 100 may perform environmental adaptation, adjusting detection thresholds, deterrent signal propagation models, and maneuver calculations based on real-time environmental data received from the environmental data input 585. In an example embodiment, the environmental data input 585 may comprise dedicated environmental sensors 160 or a data interface receiving environmental parameter data from pre-existing avionics sensors onboard the aircraft, as described above. At step 590, the system 100 may record encounter data, storing the species identification result, confidence level, deterrent signals emitted, wildlife behavioral response, evasive maneuver recommendations issued, pilot response or automatic control action taken, and environmental conditions for the encounter. The recorded encounter data may be provided to the data recording and learning module 150 for use in improving the accuracy of future trajectory analysis and deterrent signal selections through a performance improvement feedback loop.General Computing Framework

[0073] FIG. 8 illustrates a general computing framework 2000 representative of the hardware architecture on which the system 100 may be implemented. In an embodiment, the computing framework 2000 may represent one implementation of the onboard processing hardware of the Al processor 120, the data recording and learning module 150, and related system components. The computing framework 2000 is provided for illustrative purposes and does not limit the system 100 to any particular hardware configuration. In an example embodiment, the system 100 may beimplemented on hardware having a comparable architecture, including commercially available embedded computing platforms suitable for lightweight aircraft installation.

[0074] In an embodiment, the computing framework 2000 may comprise a processor 2010 configured to execute the machine learning models, species identification algorithms, threat assessment computations, and evasive maneuver calculations described herein. A system bus 2020 may interconnect the processor 2010 with system memory 2030, a storage media 2040, a network interface 2070, and an input / output interface 2060. The system memory 2030 may store instructions and data required for real-time operation of the system 100, including the machine learning model weights, species identification databases, and encounter processing buffers. The storage media 2040, which as used herein refers to one or more non-transitory computer-readable media configured to store data persistently, may be communicatively coupled to the system memory 2030 and may store recorded encounter data, updated machine learning models, and configuration parameters for the system 100. The input / output interface 2060 may be configured to receive sensor data from the multi-sensor array comprising the thermal imaging sensor 112 and the optical imaging sensor 114, and to transmit deterrent control signals to the deterrent signal generation system 130 and alert information to the flight control interface 140. In an embodiment, the processor 2010 and the system memory 2030 may be grouped within a module 2050, which as used herein refers to an integrated processing unit configured for compact airborne deployment.

[0075] In an embodiment, the network interface 2070 may be configured to communicatively couple the system 100 to a network 2080, which as used herein refers to any wired or wireless communications network, including local area networks, wide area networks, and cloud-based networks. The network interface 2070 may enable the data recording and learning module 150 to transmit recorded encounter data to a remote fleet-learning server and to receive updated machine learning model parameters from such a server, as described above. In an example embodiment, the network interface 2070 may support data rates of at least approximately 100 Mbps, with higher data rates of 1 Gbps or more supported where available, to enable efficient transmission of sensor data and model updates. The computing framework 2000 of FIG. 8 is further representative of the hardware architecture on which the non-transitory computer-readable medium may be implemented, wherein the storage media 2040 may constitute the non-transitory computer-readable medium storing instructions that, when executed by the processor 2010, cause the processor to perform the wildlife detection and deterrence method described herein.EXAMPLESExample 1 - Field Performance of the System

[0076] A working prototype of the system 100 was constructed and tested under real -world conditions, including roadside environments, airfield-adjacent areas, and controlled wildlife exposure scenarios. The following paragraphs describe performance data observed during prototype testing. The values reported are representative of observed prototype performance and are provided to illustrate the capabilities of the system; actual performance may vary depending on environmental conditions, sensor configuration, and species encountered.

[0077] Detection Range. In prototype testing, the system 100 demonstrated the following detection and classification ranges for representative threat species. For large mammals, including White-tailed deer, thermal detection was observed at ranges of approximately 300 to 800 meters, with positive species classification achieved at approximately 300 to 650 meters. For medium to large birds in flight, including Canada geese, detection was observed at approximately 400 to 800 meters, with species classification at approximately 250 to 500 meters. For small birds, including European starlings, detection was observed at approximately 400 to 800 meters, with species classification at approximately 200 to 350 meters. Classification range is shorter than detection range in each category because species classification requires sufficient optical resolution for shape and movement pattern analysis, which the thermal channel alone may not provide at maximum detection range.

[0078] Species Identification Accuracy. Based on aggregated test logs and observed encounters during prototype testing, the system 100 achieved the following species identification accuracy rates: large mammals, approximately 92 to 97%; medium birds, approximately 88 to 94%; small birds, approximately 80 to 90%. Identification accuracy was observed to improve with multi-frame tracking across consecutive sensor frames, fusion of thermal and visible spectrum data, motion pattern recognition applied to wingbeat frequency and flock geometry, and deployment of multiple units with overlapping detection fields.

[0079] Deterrence Effectiveness. In prototype testing, deterrence effectiveness was measured as the percentage of encounters resulting in behavioral redirection of the detected wildlife prior to the wildlife entering the defined collision-risk range. Observed deterrence success rates were approximately 90 to 98% for White-tailed deer, approximately 85 to 95% for Canada geese, and approximately 70 to 90% for European starlings and other flocking bird species. Observedbehavioral responses to the deterrent signals included directional avoidance in which the animal steered away from the protected zone, altitude gain in bird species, and retreat or hesitation in mammal species. Unlike approaches that have been described that emit static or random deterrent outputs, the system 100 links detection and deterrence in real time, targeting the deterrent signal at the identified species and adapting the signal parameters based on observed behavioral response, so that the system influences the directional trajectory of the wildlife rather than merely signaling its presence.Example 2 - Example Adaptive Deterrence Sequence.

[0080] The following is an illustrative example of the self-calibrating feedback loop in operation during a prototype test encounter. Upon detection and classification of a Canada goose at approximately 400 meters, the system 100 selected an initial deterrent signal of approximately 4 kHz pulse based on the Canada goose species profile. The initial signal produced no observed trajectory change in the detected wildlife. The self-calibration mechanism, having determined that no trajectory change occurred within the monitoring interval, escalated the deterrent output by shifting the frequency to approximately 6 kHz with added modulation, which produced a slight deviation in the goose's flight path but did not result in full redirection. The self-calibration mechanism escalated further by engaging directional beam steering to concentrate the deterrent energy in the direction of the detected wildlife and increasing amplitude, which produced full directional avoidance — the goose banked away from the aircraft's flight path and descended, consistent with the species-specific behavioral response model stored in the Al processor 120. The evasive maneuver recommendation provided to the flight control interface at this stage directed the aircraft to climb and turn away from the goose's predicted post-deterrence descent trajectory. This three-step escalation sequence, initial species-profile signal, frequency-and-modulation escalation, directional beam concentration, is representative of the system's biologically-responsive control loop and distinguishes the system 100 from approaches that have been described that emit a static or randomized deterrent output without behavioral feedback.***

[0081] Various modifications and variations of the described systems and methods of the invention will be apparent to those skilled in the art without departing from the scope and spirit of the invention. Although the invention has been described in connection with specificembodiments, it will be understood that it is capable of further modifications and that the invention as claimed should not be unduly limited to such specific embodiments. Indeed, various modifications of the described modes for carrying out the invention that are obvious to those skilled in the art are intended to be within the scope of the invention. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure come within known customary practice within the art to which the invention pertains and may be applied to the essential features herein before set forth.

Claims

CLAIMS1. A system for wildlife detection and deterrence for aircraft, comprising:a multi-sensor array comprising a thermal imaging sensor and an optical imaging sensor mounted to an aircraft;an onboard artificial intelligence processor configured to run one or more machine learning models that detect wildlife presence in real time and identify wildlife species based on data from the multi-sensor array;a deterrent signal generation system configured to emit signals to deter identified wildlife, wherein the emission system self-calibrates its output based on detected wildlife behavior; and a flight control interface configured to provide alert information and suggest evasive maneuvers based on detected wildlife trajectories.

2. The system of claim 1, wherein the multi-sensor array is configured in either a forwardfacing arrangement or a 360-degree array for full flight awareness.

3. The system of claim 1, wherein the artificial intelligence processor comprises convolutional neural networks trained to detect animal signatures and differentiate between wildlife species based on shape, thermal signature, visual outlines, and movement characteristics.

4. The system of claim 1, wherein the artificial intelligence processor is further configured to prioritize threats and assign behavior scores to determine response frequencies.

5. The system of claim 1, wherein the deterrent signal generation system comprises an array of speakers capable of emitting a range of frequencies, including ultrasonic and infrasonic frequencies, and includes feedback-driven calibration that adjusts volume, tone, frequency, and pattern based on observed animal reactions.

6. The system of claim 5, wherein the deterrent signal generation system uses either a stationary 360-degree array or a servo-controlled / gimbaled directional array.

7. The system of claim 6, wherein the deterrent signal generation system steers the direction of deterrent signal emission through electronic beam steering without mechanical actuation.

8. The system of claim 1, wherein the flight control interface is further configured to provide direct inputs to the aircraft’s flight control systems for automatic evasive maneuvering.

9. The system of claim 1, wherein the flight control interface comprises a pilot interface configured as a cockpit display or heads-up display through which the pilot receives visual and audio notifications of detected wildlife and suggested evasive maneuvers.

10. The system of claim 1, further comprising a data recording and learning module configured to record, for each wildlife encounter, species identification results, deterrent signals emitted, wildlife behavioral responses, and effectiveness of evasive maneuvers, and to use the recorded encounter data to update and improve the machine learning models of the artificial intelligence processor.

11. The system of claim 10, wherein the data recording and learning module is further configured to transmit recorded encounter data to a remote fleet-learning server, and to receive updated machine learning model parameters from the fleet-learning server derived from the combined encounter history of a plurality of deployed systems, so that the machine learning models of the artificial intelligence processor are improved based on the collective encounter experience of an entire fleet of systems.

12. The system of claim 1, wherein the artificial intelligence processor is further configured to: determine the range, trajectory, and threat level of detected wildlife; and predict the response of identified wildlife to deterrent signals.

13. The system of claim 1, further comprising an environmental data input configured to supply wind speed, wind direction, atmospheric temperature, and atmospheric pressure data to the system for real-time recalibration of detection, deterrence, and flight control parameters, wherein the environmental data input comprises one or more of (a) a plurality of dedicated environmentalsensors mounted to the aircraft and configured to measure the environmental parameters directly; or (b) a data interface configured to receive environmental parameter data from pre-existing sensors onboard the aircraft.

14. The system of claim 13, wherein the artificial intelligence processor is configured to: adjust beam direction, frequency, and amplitude of the deterrent signal output of the deterrent signal generation system based on the current wind vector to increase acoustic energy delivery to detected wildlife relative to delivery without wind-vector compensation; model acoustic propagation of the deterrent signal through the current wind field based on wind speed, wind direction, temperature, and atmospheric pressure; and compute a predicted evasive trajectory for detected wildlife that accounts for wind-influenced movement patterns, so that the suggested evasive maneuver or deterrent output is configured to guide detected wildlife in a direction away from the protected zone consistent with favorable wind patterns.

15. The system of claim 1, wherein the system comprises a modular housing adaptable to multiple airframes, constructed from lightweight composite materials suitable for integration with ultralight aircraft, unmanned aerial vehicles (UAVs), electric vertical take-off and landing (eVTOL) vehicles, and conventional passenger aircraft.

16. The system of claim 1, wherein the artificial intelligence processor is capable of identifying multiple wildlife species simultaneously and the deterrent signal generation system is capable of emitting multiple species-specific deterrent signals simultaneously.

17. The system of claim 1 further comprising a data fusion module configured to merge thermal imaging data and optical imaging data from the multi-sensor array into a unified sensor data set prior to species identification by the artificial intelligence processor.

18. The system of claim 1, wherein the deterrent signal generation system comprises an array of between one and more than six speakers positioned at one or more of: a bottom, a top, or a nose of the aircraft.

19. The system of claim 1, wherein the thermal imaging sensor serves as a primary detection sensor during night operation and in conditions of reduced ambient light, and the multi-sensor array is configured for all-weather, all-conditions detection operation.

20. The system of claim 1, wherein the deterrent signal generation system is configured to emit deterrent signals in species-specific frequency ranges.

21. The system of claim 1, wherein the self-calibration of the deterrent signal generation system comprises monitoring behavioral response of detected wildlife to emitted deterrent signals, determining whether the wildlife trajectory has changed away from the protected zone, and if no trajectory change is detected, escalating one or more output parameters comprising frequency adjustment within a range of ±5% to 20% of the current frequency, pulse rate adjustment, amplitude adjustment within safe limits, and beam steering adjustment, and iterating the escalation sequence until the wildlife is redirected or exits the engagement range.

22. The system of claim 1, wherein the artificial intelligence processor and deterrent signal generation system are cooperatively configured to guide detected wildlife along a non-hazardous trajectory by: (i) computing a target trajectory that routes the detected wildlife around a protected zone; (ii) selecting beam direction, frequency, amplitude, and pulse pattern of the deterrent signal calculated to shape the movement of the detected wildlife toward the target trajectory rather than to disperse the wildlife indiscriminately; and (iii) iteratively updating the selected deterrent parameters in response to observed wildlife movement to maintain the wildlife along the target trajectory, thereby creating a dynamic safe pathway around the protected zone.

23. The system of claim 22, wherein the target trajectory is computed based at least in part on current wind direction, wind speed, terrain features, and the location of one or more adjacent protected zones, such that the deterrent signal generation system establishes an adaptive corridor along which detected wildlife are routed.

24. The system of claim 1, wherein the system is configured for deployment on one or more of: a wind turbine or wind farm installation to prevent bird strikes with turbine blades, a marinevessel to deter marine wildlife within a deterrence radius around the marine vessel to create safe zones, a landing area for eVTOL aircraft to deter non-flying wildlife, and a motor vehicle comprising one of: a car, a truck, or a motorcycle to deter wildlife from the path of the vehicle.

25. A method for wildlife detection and deterrence for aircraft, comprising:detecting wildlife in the vicinity of an aircraft using a multi-sensor array comprising at least a thermal imaging sensor and an optical imaging sensor;running a machine learning algorithm on an onboard processor to detect wildlife presence in real time;fusing thermal imaging data and optical imaging data from the multi-sensor array into a unified sensor data set;identifying wildlife species using one or more machine learning models executed by the processor;differentiating between species based on shape, thermal signature, and movement characteristics;emitting species- specific ultrasonic and / or infrasonic signals to deter identified wildlife, wherein the emission system self-calibrates based on detected wildlife behavior; and providing alert information and suggested evasive maneuvers through a flight control interface based on detected wildlife trajectories.

26. The method of claim 25, further comprising assigning behavior scores and prioritizing threats based on proximity, trajectory, and species characteristics.

27. The method of claim 25, further comprising recording data about wildlife encounters and using recorded data to improve wildlife identification and deterrence performance.

28. The method of claim 25, wherein emitting species-specific sonic frequencies comprises pulsating through various frequency ranges and adjusting volume, tone, frequency, and pattern based on observed animal reactions.

29. The method of claim 25, wherein providing suggested evasive maneuvers comprises predicting the response of identified wildlife to deterrent signals and suggesting maneuvers configured to direct the aircraft away from the predicted post-deterrence trajectory of the wildlife, thereby increasing the separation distance between the aircraft and the detected wildlife.

30. The method of claim 25, further comprising providing direct inputs to the aircraft’s flight control systems for automatic evasive maneuvering.

31. The method of claim 25, further comprising performing real-time recalibration for environmental factors including wind speed, wind direction, temperature, and atmospheric pressure to maintain consistent detection and deterrence performance across varying weather conditions.

32. The method of claim 25 further comprising: monitoring behavioral response of detected wildlife to emitted deterrent signals, determining whether the wildlife trajectory has changed away from a protected zone defined as a volume of airspace that the method is configured to keep clear of detected wildlife, and if no trajectory change is detected, escalating one or more deterrent output parameters comprising frequency adjustment within a range of ±5% to 20% of the current frequency, pulse rate adjustment, amplitude adjustment within safe limits, beam steering adjustment, and iterating the escalation sequence until the wildlife is redirected or exits the engagement range.

33. The method of claim 25, further comprising guiding detected wildlife along a non-hazardous trajectory around a protected zone, the protected zone comprising a defined exclusion region to be kept clear of detected wildlife, by: computing a target trajectory that routes the detected wildlife around the protected zone; selecting beam direction, frequency, amplitude, and pulse pattern of the emitted deterrent signal calculated to shape the movement of the detected wildlife toward the target trajectory rather than to disperse the wildlife indiscriminately; and iteratively updating the selected deterrent parameters in response to observed wildlife movement to maintain the wildlife along the target trajectory, thereby creating a dynamic safe pathway around the protected zone.

34. The method of claim 25, wherein emitting species-specific deterrent signals comprises modelling an expected post-deterrence trajectory of the detected wildlife based on species-specific behavioral data and real-time behavioral feedback, and calculating suggested evasive maneuvers based on the modelled post-deterrence trajectory.

35. The method of claim 25 further comprising recording, for each wildlife encounter, the species identification result, confidence level, deterrent signals emitted, wildlife behavioral response, evasive maneuver recommendations issued, and environmental conditions, and using the recorded encounter data to improve accuracy of the future trajectory analysis and deterrent signal selections through a performance improvement feedback loop.

36. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform a method for wildlife detection and deterrence for aircraft, the method comprising:analyzing data from a multi-sensor array comprising at least a thermal imaging sensor and an optical imaging sensor to detect wildlife in the vicinity of an aircraft;running one or more machine learning models to identify wildlife species based on shape, thermal signature, visual outlines, and movement characteristics;fusing thermal imaging data and optical imaging data from the multi-sensor array into a unified sensor data set prior to species identification;differentiating between wildlife species and prioritizing threats based on proximity, trajectory, and behavior;selecting species-specific ultrasonic and / or infrasonic frequencies for deterring the identified wildlife;controlling a deterrent signal generation system to emit the selected frequencies with selfcalibration based on detected wildlife behavior; andproviding alert information and suggested evasive maneuvers through a flight control interface based on detected wildlife trajectories.

37. The non-transitory computer-readable medium of claim 36, wherein the method further comprises recording data about wildlife encounters and improving wildlife identification and deterrence performance based on the recorded data.

38. The non-transitory computer-readable medium of claim 36, wherein identifying wildlife species comprises determining a confidence level for the identification and selecting multiple possible species when the confidence level is below a threshold.

39. The non-transitory computer-readable medium of claim 36, wherein selecting speciesspecific frequencies comprises selecting multiple frequencies when multiple possible species are identified, and adjusting volume, tone, frequency, and pattern based on observed animal reactions.

40. The non-transitory computer-readable medium of claim 36, wherein providing suggested evasive maneuvers comprises predicting the response of identified wildlife to the selected sonic frequencies and suggesting maneuvers configured to direct the aircraft away from the predicted post-deterrence trajectory of the wildlife, thereby increasing the separation distance between the aircraft and the detected wildlife.

41. The non-transitory computer-readable medium of claim 36, wherein the method further comprises performing real-time recalibration for environmental factors including wind speed, wind direction, temperature, and atmospheric pressure to maintain consistent detection and deterrence performance across varying weather conditions.

42. The non-transitory computer readable medium of claim 36, wherein the method further comprises: monitoring behavioral response of detected wildlife to emitted deterrent signals, determining whether the wildlife trajectory has changed away from a protected zone defined as a volume of airspace to be kept clear of detected wildlife, and if no trajectory change is detected, escalating one or more deterrent output parameters comprising frequency adjustment of ±5% to 20%, pulse rate adjustment, amplitude adjustment within safe limits, beam steering adjustment, and iterating the escalation sequence until the wildlife is redirected or exits the engagement range.

43. The non-transitory computer readable medium of claim 36, wherein the method further comprises transmitting alert information to a pilot interface configured as a cockpit display or heads-up display, the alert information comprising visual and audio notifications of detected wildlife and suggested evasive maneuvers.

44. A mobile wildlife detection and deterrence system, comprising:a multi-sensor array comprising at least a thermal imaging sensor and an optical imaging sensor mounted to a mobile platform, the multi-sensor array configured to detect wildlife via thermal and optical imaging;an artificial intelligence processor configured to identify detected wildlife and determine species-specific deterrent responses;a deterrent signal generation system configured to emit ultrasonic or infrasonic deterrent signals toward detected wildlife based on the determined species-specific deterrent responses; and a flight or navigation interface configured to issue alerts or control signals to the vehicle in response to wildlife threat data.

45. A fixed infrastructure wildlife detection and deterrence system, comprising:a multi-sensor array comprising at least a thermal imaging sensor and an optical imaging sensor mounted to a fixed structure;an artificial intelligence processor configured to run one or more machine learning models that detect wildlife presence in real time and identify wildlife species based on data from the multisensor array;a directional deterrent emission system configured to emit ultrasonic and / or infrasonic signals to deter identified wildlife, wherein the emission system self-calibrates its output based on detected wildlife behavior; anda data logging module configured to record detection events and system responses.

46. The system of claim 1, wherein the artificial intelligence processor adjusts deterrent signals in real time based on observed wildlife behavior.

47. The system of claim 45, wherein the directional deterrent emission system emits speciesspecific deterrent frequencies.

48. The system of claim 45, further comprising:an environmental data input configured to supply wind speed, wind direction, atmospheric temperature, and atmospheric pressure data to the system for real-time recalibration of detection and deterrence parameters; and / ora networking interface configured to transmit recorded encounter data to, and receive updated machine learning model parameters from, a centralized server that aggregates encounter data across a plurality of deployed systems, so that the machine learning models of the artificial intelligence processor are improved based on the collective encounter experience of the deployed systems.

49. The system of claim 45, wherein the fixed structure comprises one or more of: an airport or airfield perimeter, a runway approach zone, a wind turbine or wind farm installation, an electrical substation or transmission infrastructure, a rail yard or rail right-of-way, a marine dock or port facility, a solar installation, an agricultural facility, and a roadway or highway corridor.

50. The system of claim 45, wherein the artificial intelligence processor and directional deterrent emission system are cooperatively configured to route detected wildlife along a non-hazardous trajectory around a protected zone associated with the fixed structure by selecting beam direction, frequency, amplitude, and pulse pattern of the deterrent signal calculated to shape movement of the detected wildlife toward the non-hazardous trajectory, and iteratively updating the selected deterrent parameters in response to observed wildlife movement so as to establish an adaptive corridor around the protected zone.

51. The system of claim 44, wherein the flight or navigation interface automatically inputs evasive maneuvers.

52. The system of claim 45, wherein the data logging module is further configured to transmit wildlife detection event data, including species identification results, detected trajectories, andsystem response records, to an urban air traffic management system or centralized airspace management platform, to enable airspace-level wildlife hazard awareness and coordination across multiple aircraft operating in the same airspace.

53. The system of claim 5, wherein the deterrent signal generation system comprises a phased-array speaker arrangement and is configured to steer a direction of deterrent signal emission through electronic beam steering without mechanical actuation, by adjusting relative phase and amplitude of signals emitted from a plurality of speakers of the array to produce constructive interference in a direction of the detected wildlife and destructive interference in non-target directions.