
Real-Time
Intelligence
Tiny Build,
Mighty Brian
The ultimate AI hardware platform
for legged robots, autonomous agents, and home controllers.
Introducing AirPath® V4 McQueen — A Development Kit for VLA-Powered Autonomy and On-Device RFM
Designed to enhance real-time inference performance,
AirPath® V4 McQueen is a compact and lightweight edge AI development kit, purpose-built for next-generation mobile robotics and on-device autonomy.
It seamlessly integrates language, vision, and robotic foundation models (RFM) into a single on-device AI stack — enabling real-time response without reliance on cloud connectivity.
Vision-Language-Action autonomous agents
Matter-compliant AI home hubs
AI-integrated kiosks and display controllers
Conversational concierge robots powered by Korean LLMs
It empowers developers to deploy next-gen AI agents that see, think, and act locally, redefining what’s possible in edge robotics.
The AirPath V4 McQueen platform is powered by the Qualcomm® Dragonwing™ QCS8550 and features
ROS2 and container-based heterogeneous hardware integration,
an open AI framework based on ONNX,
real-time 5G connectivity and dataset generation, and
on-device inference of lightweight Korean SLMs and Vision-Language-Action (VLA) models,
enabling immediate AI execution at the edge. This architecture simultaneously addresses data security, response latency, and network instability.
Furthermore, AirPath V4 reflects the development environment required for humanoid robots trained and simulated using NVIDIA Isaac-based GR00T foundation models and SDKs. It offers an open interface that supports diverse proprietary behavior models—alongside NVIDIA’s Robot Foundation Model (RFM)—trained across various real-world and virtual environments.
Real-Time Monitoring and Command Integration with AirPath® V4 + Flyinglet™
AirPath® V4 platform transforms raw data collected from robot-mounted cameras, LiDARs, and multi-modal sensors into structured, lightweight datasets in real time. This enables a standardized data schema for heterogeneous robot control and centralized orchestration—laying the groundwork for legacy system integration, data science workflows, and scalable AI services.
Flyinglet™ Control Cloud
The integrated Flyinglet™ dashboard provides real-time visualization of robot status, path tracking, mission planning, and AI vision alerts—across multiple robot platforms.
Multi-Robot Integration
Seamlessly connects diverse robots such as Boston Dynamics Spot and Unitree models, offering unified control across heterogeneous fleets.
Synchronized Sensor Data Ingestion & Inference
Synchronizes and visualizes data streams from vision, LiDAR, IMU, and other sensors, enabling edge-level intelligence and situational awareness.
On-Device AI Control with Ultra-Low Latency
Powered by the AirPath® V4 McQueen platform, all AI inference, decision-making, and motion control are executed locally with minimal latency.
Optimized for Real-World Deployment
Validated in harsh industrial settings such as semiconductor fabs, logistics warehouses, and smart factory automation scenarios.
The AirPath® V4 platform is a high-performance AI computing module designed for customizable deployment across robotics and industrial settings.
Due to its flexible architecture and project-specific configurations, detailed specifications are provided upon request.
If you’re interested in adopting AirPath® V4 or require further technical documentation, please reach out via the contact channels below:
Designed to deliver robust, scalable intelligence in mission-critical environments
where latency, connectivity, and operational independence matter
Unifying LLM, VLA, and RFM for autonomous intelligence on the edge
AirPath® V5 Stratos is a next-generation edge AI computing module powered by the Qualcomm™ Dragonwing IQ9.
Engineered for high-load environments, it delivers powerful GPU/NPU performance and integrates
LLM, Vision-Language-Action (VLA), Robotics Foundation Model (RFM), and control systems into a single on-device AI platform.
It is purpose-built for advanced social humanoids, quadruped robots, and high-reliability industrial systems
that demand real-time autonomous inference, control, and decision-making — without cloud dependency.