loading

TECHNOLOGY

What is Embedded AI?

Core technology that embeds “intelligence” into devices, enabling them to become smart and autonomous

  • Independent operation without reliance on cloud servers
  • Integrated AI and machine learning models directly within the hardware
  • No need to transmit AI computation data to data centers
  • Independent AI functionality and real-time response even without internet connectivity
  • AI models optimized for low power consumption and compact design

Smart Mobility (ITS, Intelligent Transport System)

Advanced AI Recognition Technology

  • Multi-Object Detection & Tracking Simultaneous detection and continuous tracking of various road objects such as vehicles, pedestrians, and motorcycles using technologies like DeepSORT
  • Edge-Based License Plate Recognition High-speed recognition and extraction of vehicle license plates directly on the edge board without transmitting video data to the cloud
  • Incident Detection Immediate identification of abnormal situations such as stopped vehicles, wrong-way driving, jaywalking, and road debris, with instant alerts sent to the control center

NPU-Based Edge Optimization Technology

  • Heterogeneous Computing Optimization Maximizing processing performance by efficiently distributing workloads across CPUs, GPUs, and NPUs (Neural Processing Units)
  • Deep Learning Model Optimization (Lightweighting) Deploying optimized AI models on edge devices through quantization (Int8) and pruning techniques
  • Multi-Stream Pipeline Processing Parallel processing technology capable of simultaneously decoding and analyzing 4 to 16 channels of high-resolution CCTV video (FHD/4K) on a single board

Defense AI Edge Computing

Standardization & Miniaturization Technology

  • SOSA-Aligned H/W Design Full compliance with VITA 65 OpenVPX and SOSA technical standards to ensure interoperability
  • High-Speed Signal Integrity High-speed data signal design and simulation for technologies such as PCIe Gen4/5 and 100GbE
  • SWaP-C Optimization (Size, Weight, Power & Cost) Design optimization that minimizes size, weight, and power consumption while maintaining performance for weapon systems such as drones, including VNX (VITA 74) form factor technologies

AI/NPU Optimization Technology: Tactical Edge Computing

  • Heterogeneous System Architecture Architecture implementation enabling efficient data exchange among CPU (control), GPU/NPU (AI inference), and FPGA (sensor preprocessing) on a single board
  • Mission-Critical RTOS Support Full NPU driver and BSP support for aviation operating systems and military-grade RTOS environments