Next-Gen Hyper-Sensing AI Module
The AI VisionCube Kit is a revolutionary AI visual tracking system designed specifically for drones, integrating high-performance target recognition, dynamic tracking, and anti-jamming control. Through the fusion of deep learning algorithms and high-performance intelligent chips, drones equipped with this kit can achieve automatic target acquisition, stable tracking, and precise strikes in complex electromagnetic environments. This significantly reduces the operational threshold for pilots and greatly improves mission execution efficiency.
Introduction
The AI VisionCube Kit is a revolutionary AI visual tracking system designed specifically for drones, integrating high-performance target recognition, dynamic tracking, and anti-jamming control. Through the fusion of deep learning algorithms and high-performance intelligent chips, drones equipped with this kit can achieve automatic target acquisition, stable tracking, and precise strikes in complex electromagnetic environments. This significantly reduces the operational threshold for pilots and greatly improves mission execution efficiency.
1.2km Ultra-Long-Range Recognition
Powerful AI recognition capability, automatically captures targets.
Multi-target recognition
Simultaneously recognizes up to 50 targets within a 1.2 km range, including humans and vehicles.
Continuous tracking of moving targets
Utilizes trajectory prediction and feature memory dual-mechanism tracking, ensuring stable tracking even with fast movement or brief obstructions.
Adaptive Lock-On Tracking
The automatic adhesion feature around the crosshair increases target lock rate by over 80%, simplifying pilot operation.
PiP Architecture
Dual-stream PiP technology combines global situational awareness (main view) with 8x digital zoom tracking (sub-view), allowing real-time environmental analysis and precision strike decision-making.
Multi-Sensor Compatibility
Switchable visible/thermal sensors with 9650 mV/lux·s ultra-high sensitivity and uncooled thermal imaging. Ensures 24/7 target ID capability from -20℃ to 60℃, defeating camouflage/thermal masking.AI VisionCube Kit Recognition Distance Doubled
AI VisionCube Kit’s recognition distance is increased by over 80%, accuracy
improved by over 60%, and dynamic target tracking success rate increased by over 45%.
Lightweight Integration Design
Modular design (<200g) supports Betaflight/APM open-source flight controllers. Achieves AI capability deployment within 5 minutes via plug-and-play installation, compatible with industrial drones to FPV.
AI Vision Cube Kit's Super New Features
1.2km Ultra-Long-Range Recognition
Equipped with single/dual-camera collaborative algorithms and precision optical modules, automaticallyidentifies human/vehicle targets at 1.2km range. Achieves 85% higher recognition accuracy than conventional AI systems, enabling beyond-visual-range operations in complex environments.
Equipped with single/dual-camera collaborative algorithms and precision optical modules, automatically identifies human/vehicle targets at 1.2km range.
Multi-Target Recognition, Achieving Rapid Judgment
Using the YOLO series model, supporting recognition of various types and sizes of targets, with a maximum of 50 targets simultaneously.
Continuous Tracking of Moving Targets
Combined trajectory prediction and feature memory mechanisms ensure stable tracking even with fast movement or brief obstructions.
Automatic Adhesive Tracking and Lock-On
The crosshair automatically adheres to the nearest target, and the pilot only needs to move the crosshair near the target for automatic lock and tracking.
If you want Lock and track Target C：Move the crosshair to TargetC’s vicinity, and the system will automatically lock Target C
Crosshair automatically adheres and locks onto Target C The crosshair automatically locks onto Target C and starts tracking.
PiP (Picture-in-Picture) Function
Dual-stream PiP technology combines global situational awareness
(main view) with 8x digital zoom tracking (sub-view), allowing
real-time environmental analysis and precision strike decision-making.
Ultra-low Latency, Real-Ttime Image
Transmission is Right in Front of You
Full-Link <60ms Low Latency Transmission: From super-high-definition image capture
to the pilot’s glasses, the total delay is under 60ms, ensuring real-time decision-making is reliable.
Multi-Sensor Compatibility
Both AI VisionCube S and AI VisionCube D support thermal cameras,
adapting to more complex operational environments.
Single Lens + Thermal Lens,
Adapting to Diverse Environments
Combines thermal and wide-angle lenses for flexible target recognition in different environments.
A
B
Dual Lenses + Thermal,
Adapting to Diverse Environments
Combines thermal and wide-angle lenses for flexible target recognition in different environments.
A
B
C
Video Demonstration
Real product testing video, compressed. Due to venue limitations, the
product's features were not fully displayed. If you want to view more test video clips, please contact customer service.
DIY? No Problem
Customizable according to needs, supporting personalized modifications.
Supports popular brand firmware:
Compatible with Betaflight, APM firmware.
AI VisionCube Kit Technical Specifications
| Figure |  |
 |
 |
 |
|---|---|---|---|---|
| Module Name | AI VisionCube S | AI VisionCube D | AI VisionCube ST | AI VisionCube DT |
| Default Target Recognition Types | Human, Vehicle | Human, Vehicle | Human, Vehicle | Human, Vehicle |
| Target Detection Range | Vehicle: 450m, Human: 170m | Vehicle: 1200m, Human: 500m | Vehicle: 450m, Human: 170m | Vehicle: 1200m, Human: 500m |
| Minimum Target Tracking Pixels | 16x16 Pixels | 16x16 Pixels | 16x16 Pixels | 16x16 Pixels |
| Maximum Tracking Speed of Dynamic Targets | 60km/h | 60km/h | 60km/h | 60km/h |
| Intelligent Adhesive Tracking crosshair fuzzy locking for close-range adhesive tracking |
Support | Support | Support | Support |
| Supports trajectory prediction and feature memory tracking | Support | Support | Support | Support |
| Maximum Number of Targets Supported | 50 | 50 | 50 | 50 |
| Picture-in-Picture (PiP) Support | Support | Support | Support | Support |
| AI Image Processing Board | ||||
| Computing Power (TOPS) | 1 | 6 | 1 | 6 |
| Protocol | CRSF | CRSF | CRSF | CRSF |
| Supported Firmware | BetaFlight / APM /PX4 | BetaFlight / APM /PX4 | BetaFlight / APM /PX4 | BetaFlight / APM /PX4 |
| Input Voltage (V) | 9-16 | 9-16 | 9-16 | 9-16 |
| Dimension (LxWxH, mm) | 38*38*27.7 | 38*38*26.7 | 38*38*27.7 | 38*38*26.7 |
| Mounting Holes (mm) | 25.5*25.5 | 25.5*25.5 | 25.5*25.5 | 25.5*25.5 |
| Visible Camera 1 | ||||
| Sensor (Inches) | 1 / 2.8 | 1 / 2.6 | 1 / 2.8 | 1 / 2.6 |
| Focal Length (mm) | 4 | 3.9 | 4 | 3.9 |
| FOV | 69°(H)*42°(V) | 72°(H)*45°(V) | 69°(H)*42°(V) | 72°(H)*45°(V) |
| Low Light Performance | 7341 mV/lux·s | 9650 mV/lux·s | 7341 mV/lux·s | 9650 mV/lux·s |
| Video Input Format | 1920*1080@30hz | 1920*1080@60hz | 1920*1080@30hz | 1920*1080@60hz |
| Camera Dimension (LxWxH, mm) | 19*19*30 | 40.8*25*26 | 19*19*30 | 40.8*25*26 |
| Visible Camera 2 | ||||
| Sensor (Inches) | N/A | 1 / 2.6 | N/A | 1 / 2.6 |
| Focal Length (mm) | N/A | 12 | N/A | 12 |
| FOV | N/A | 26°(H)*15°(V) | N/A | 26°(H)*15°(V) |
| Low Light Performance | N/A | 9650 mV/lux·s | N/A | 9650 mV/lux·s |
| Video Input Format | N/A | 1920*1080@50h | N/A | 1920*1080@50h |
| Camera Dimension (LxWxH, mm) | N/A | 40.8*25*26 | N/A | 40.8*25*26 |
| Thermal Camera | ||||
| Pixel Pitch | N/A | N/A | 12um | 12um |
| Focal Length | N/A | N/A | 9.1mm | 9.1mm |
| FOV | N/A | N/A | 20.3°(H)×15.2°(V) | 20.3°(H)×15.2°(V) |
| Response Band | N/A | N/A | 8～14μm (LWIR) | 8～14μm (LWIR) |
| Video Input Format | N/A | N/A | 384*288@25hz | 384*288@25hz |
| Camera Dimension (LxWxH, mm) | N/A | N/A | 26*26*42.63 | 26*26*42.63 |
| Interface | N/A | N/A | USB | USB |
Application Scenario
When there is a need to track moving targets
The other party has a full-band jammer (strong interference environment)
In weak interference environments
When there is a need to track moving targets
The other party has a full-band jammer (strong interference environment)
In weak interference environments
Recommended Combination Solutions
AI Assistance Flight Solution can be achieved by adding the AI Vision Cube Kit to the FPV drone such as HS-X07M/HS-X10M/HS-P106M/HS-P134M/HS-P136M
Comparison Simulation of the Effects Before &
After Installing the AI Visual Locking &
Assistance System
(1) An FPV drone without the AI Assistance System, unable to approach the target
(2) The transmission image is interfered, leading to a crash
(1) The FPV drone with the AI Assistance System. With the AI module activated to lock and track the target
(2) The transmission image is interfered, but the target can still be locked and tracked.
Comprehensive Comparison of Solutions
-
Comprehensive Comparison of SolutionsComparison Table
- Anti-interference Capability
- Target Rate
- Response Capability
- Portability
- Remote Control Distance
- Load Capacity (with equivalent configuration)
-
Frequency Band Optimization SolutionSolution 1
- Weak
- Low
- Fast
- High
- Far
- High
-
AI Assistance(AI VisionCube Kit)
Flight SolutionSolution 2 - Medium
- Medium
- Medium
- High
- Far
- High
-
Fiber Optic Control
SolutionSolution 3 - Strong
- High
- Slow
- low
- Medium
- Medium
Note:
Solution 1: Frequency Band Optimization Solution
Solution 2: Solution of AI Vision Locking & Assistance System
Solution 3: Solution of Optical Fiber Transmission Control System