DADA Technology Co.,Ltd.

21/07/2026

The DA-057 is an ultra-compact dual-sensor AI tracking board based on the RK3588 platform, designed specifically for demanding industrial applications such as optoelectronic pods, border and coastal surveillance, and counter-UAV systems. It integrates one Camera Link Base infrared input and one SDI visible-light input, supporting collaborative dual-channel video processing. Equipped with a built-in 6 TOPS NPU, it performs real-time detection of targets—including humans, vehicles, ships, and aircraft—and outputs miss-distance data to enable automatic or manual lock-on tracking.

It features a rich set of interfaces—including SDI loop-out, RS422 serial ports, and Gigabit Ethernet (supporting RTSP streaming)—to facilitate system integration. Measuring just 85×55mm and weighing approximately 102g (including the heatsink), the board operates at under 10W, making it ideal for space-constrained mobile platforms. It supports H.265 hardware encoding, with 64GB of storage allowing for approximately 3 hours of video recording (at a 6Mbps bitrate). Comprehensive SDK and driver support are provided to help customers rapidly bring AI vision products to market.
Learn more: https://uavauto.com/product/trac-da057h-video-tracking-board/

16/07/2026

DS071 Drone AI Module — Industrial-Grade Performance, Lightweight Innovation
Positioning: An AI edge computing module designed specifically for FPV drones and mobile platforms; it equips devices with real-time object detection and tracking capabilities while minimizing weight.

Key Highlights
Ultra-Lightweight: Weighs only 21.6g with dimensions of 43×38 mm and a height of 14.5 mm (including B2B connectors and mounting standoffs), significantly reducing the drone's payload.

Industrial-Grade Reliability: Features industrial-grade chips; operating temperature range of -40°C to +65°C and storage temperature range of -55°C to +75°C, ensuring suitability for harsh outdoor environments.

Powerful Computing: Delivers 3 TOPS of computing power to meet real-time AI inference needs (standard support for detecting and tracking humans, vehicles, boats, and drones; custom algorithms available for other targets).

Low Power Consumption: Total board power consumption ≤4W, extending drone flight time.

Interface Configuration (Next-Gen Digital Interfaces)
Video Input: Dual MIPI interfaces, supporting multi-channel visual data acquisition.

Video Output: CVBS (analog output) + 100Mbps Ethernet (digital output), offering flexible compatibility with video transmission links.

Expansion Interfaces: USB 3.0 (high-speed data transfer), Serial Port, and RS422 (industrial control/telemetry), providing ample room for expansion.

Supporting Services
Ready-to-Use: We offer a range of compatible cameras with varying specifications (fields of view and resolutions) for the DS071, significantly shortening the component selection and validation cycle for customer projects.
Learn more: https://uavauto.com/product/ds071-compact-ai-video-tracking-board/

22/06/2026

Field testing of drone AI modules has revealed various issues and requirements. A critical challenge is that detection and tracking performance varies with operating distance when using a single fixed-focus lens. To address this, we designed the DSE072 AI module, which features dual MIPI inputs and supports a dual-lens setup (wide and narrow fields of view) to extend the effective operating range.
The AI ​​module utilizes a compact, square single-board design measuring 45 x 45 x 7.5 mm, weighing approximately 13.5 g, with a peak power consumption of no more than 4 W.
It is equipped with two MIPI interfaces, one DVP interface, and outputs including CVBS and 100 Mbps Ethernet, alongside USB 2.0, RS422, and UART ports. It supports high-frame-rate (120 fps) cameras to autonomously detect specific targets—such as people, vehicles, boats, and aircraft (including drones)—within the field of view, providing data on target type, location, and size.
For this AI module, we have customized a high-frame-rate, global-shutter monochrome camera featuring a MIPI interface, 1080P resolution, selectable frame rates (120/60 Hz), and auto-exposure support.

Wide Field of View (FOV) Mode: Operates at 120 Hz input using a standard 3.6 mm low-distortion lens. For a 1-meter-sized target, the operating range is 300–500 m, with a field of view (D×H×V) of 90°×80°×55°.

Narrow Field of View (FOV) Mode: Operates at 60 Hz input using a standard 8 mm low-distortion lens. For a 1-meter-sized target, the operating range is 600–1000 m, with a field of view (D×H×V) of 45°×40°×23°.
By switching between wide and narrow fields of view, the system effectively detects and tracks targets at varying distances, enhancing both accuracy and stability.
We have successfully integrated the module with Betaflight flight controller firmware; additionally, users can implement integration with other flight controllers via serial protocols to enable functions such as automated tracking and interception. https://uavauto.com/product/ds072-ultra-light-ai-video-tracking-board/

08/10/2025

A. Airborne AI integrates identification, tracking and attack; B. The ecology is complete, analog surveying and mapping are compatible with digital surveying and mapping, and the supporting APP can realize pointing, tracking and attack; realizing a closed loop of identification/tracking/attack; C. Image input compatibility: The video end is compatible with analog CVBS, USB HD camera, MIPI HD camera; D. Lens compatibility: FPV visible light, low-light night vision, infrared thermal imaging, dual light sources, long and short focal length lenses, and can even be adapted to millimeter wave radar and laser module guidance E. Wide platform adaptability; rotorcraft, vertical composite wing, cruise bomb and other platforms; F. Flight control hardware and software firmware are highly adaptable; mainstream BF and INAV firmware; open source APM and closed source commercial flight control platforms are also supported; G. Simple operation and easy to use; H. The whole stack of software and hardware is self-developed, and the underlying parameters are highly open.

10/09/2025

A. Airborne AI integrates identification, tracking and attack; B. The ecology is complete, analog surveying and mapping are compatible with digital surveying and mapping, and the supporting APP can realize pointing, tracking and attack; realizing a closed loop of identification/tracking/attack; C. Image input compatibility: The video end is compatible with analog CVBS, USB HD camera, MIPI HD camera; D. Lens compatibility: FPV visible light, low-light night vision, infrared thermal imaging, dual light sources, long and short focal length lenses, and can even be adapted to millimeter wave radar and laser module guidance E. Wide platform adaptability; rotorcraft, vertical composite wing, cruise bomb and other platforms; F. Flight control hardware and software firmware are highly adaptable; mainstream BF and INAV firmware; open source APM and closed source commercial flight control platforms are also supported; G. Simple operation and easy to use; H. The whole stack of software and hardware is self-developed, and the underlying parameters are highly open.

04/09/2025

01/09/2025

A. Airborne AI integrates identification, tracking and attack; B. The ecology is complete, analog surveying and mapping are compatible with digital surveying and mapping, and the supporting APP can realize pointing, tracking and attack; realizing a closed loop of identification/tracking/attack; C. Image input compatibility: The video end is compatible with analog CVBS, USB HD camera, MIPI HD camera; D. Lens compatibility: FPV visible light, low-light night vision, infrared thermal imaging, dual light sources, long and short focal length lenses, and can even be adapted to millimeter wave radar and laser module guidance E. Wide platform adaptability; rotorcraft, vertical composite wing, cruise bomb and other platforms; F. Flight control hardware and software firmware are highly adaptable; mainstream BF and INAV firmware; open source APM and closed source commercial flight control platforms are also supported; G. Simple operation and easy to use; H. The whole stack of software and hardware is self-developed, and the underlying parameters are highly open.