🤖 [B]Next-Generation Reference Design for Autonomous Robots: Dragonwing IQ10[/B] 🚀

[Erkan Teskancan]

The world of industrial automation and robotics is rapidly evolving. One of the newest and most ambitious steps in this evolution is the Dragonwing IQ10 Robotic Reference Design (RRD) presented by Qualcomm. This integrated platform aims to accelerate the transition from prototype to production by combining AI performance and deterministic control.

─────────────────────────

💡 Integrated Power in a Single Platform​

The Dragonwing IQ10 RRD combines computing, sensing, networking, and software components into a single hardware platform. This eliminates the complexity of heterogeneous processing and sensor interfaces for various applications such as industrial automation, autonomous mobile robots (AMR), and humanoid robots.

─────────────────────────

🧠 Peak AI Performance​

At the heart of the reference design, the Dragonwing IQ10 processor stands out with 18 Qualcomm Oryon central processing unit (CPU) cores, multi-core neural processing units (NPU), and a powerful graphics processing unit (GPU) architecture. This hardware configuration delivers an impressive artificial intelligence (AI) performance of 700 tera operations per second (TOPS), eliminating the need for external accelerators for on-device sensing, planning, and reasoning workloads.

─────────────────────────

👁️ Advanced Sensor Integration​

The platform reduces integration complexity by directly processing sensor data without the need for external bridging components. It supports up to 12 Gigabit Multimedia Serial Link 2 (GMSL2) cameras, along with LiDAR, Time-of-Flight (ToF), and Inertial Measurement Unit (IMU) sensors. This native integration reduces data latency between sensing and processing layers while ensuring synchronization in multi-modal data streams.

─────────────────────────

⚙️ Precise Control and Industrial Durability​

For precise motion control, the platform includes high-speed deterministic interfaces and industrial communication protocols such as PCIe, Time-Sensitive Networking (TSN), USB, Control Area Network (CAN), Ethernet, EtherCAT, and CAN Flexible Data-Rate (CAN-FD).

The hardware is housed in a forced air-cooled enclosed system and is rated to operate in a temperature range of -40 to 70 °C. It accepts 12V/24V power inputs, suitable for environmental and electrical constraints common in industrial deployment scenarios.

─────────────────────────

💻 Software Architecture and Ecosystem​

The accompanying software infrastructure uses a layered architecture. It includes on-device AI runtimes for low-latency decision-making, native Robot Operating System 2 (ROS2) support that separates hardware from application logic, and core platform services for sensing, planning, and actuation. Cloud-connected lifecycle management, fleet monitoring, and model deployment are handled via Qualcomm AI Hub. The integrated stack offers ready-to-use functional blocks for positioning, navigation, manipulation, and natural language interaction.

The system was unveiled at Computex 2026, held in Taipei, Taiwan, from June 2-5, 2026. Partners such as NEURA Robotics, Advantech, APLUX, Booster, Innodisk, MeiG, NEXCOM, Radxa, Thundercomm, and VinMotion are supporting early evaluation and ecosystem deployment.

─────────────────────────

📊 Competitive Analysis: Dragonwing IQ10 vs. Nvidia Jetson AGX Orin​

The Dragonwing IQ10 RRD enters a competitive market for high TOPS edge AI computing for robotics. The primary benchmark in this area has been set by the Nvidia Jetson AGX Orin Industrial module. The Jetson AGX Orin Industrial offers 275 TOPS of AI performance, operates in a temperature range of -40 to 85 °C, and features a 12-core Arm Cortex-A78AE CPU and a 2048-core Ampere architecture GPU.

Comparatively, the Dragonwing IQ10 reference design offers a higher computational ceiling with 700 TOPS and an 18-core architecture. However, its maximum specified operating temperature of 70 °C is narrower than the Jetson module's industrial limit of 85 °C. Additionally, while competing platforms typically require external carrier boards to break out high-density GMSL2 and EtherCAT interfaces, the Dragonwing design integrates 12 GMSL2 channels directly into a deployment-ready enclosure, minimizing propagation delay in multi-modal sensor fusion.

This new reference design represents a significant step forward for the future of autonomous systems and is poised to open new doors in the field of industrial robotics.