WE ARE A
NEXT GEN
MOBILITY LAB

EDGE AI

MOBILITY AI

DEFENSE AI

Edge LLM-based Mobility Control

LLM-Based Car Control Transforming the Driving Experience with Natural Language Interaction This technology empowers drivers with intuitive voice commands, enabling effortless control over various vehicle functions, personalized entertainment, and convenient in-car purchases, ultimately creating a seamless and hands-free driving experience that prioritizes comfort and convenience.

Intelligent Voice Command System

Understand and execute any driver's verbal instructions

Inference Control:

Even in situations where communication is disrupted, complex devices such as cars, tanks, drones, and UAVs are controlled through their own inference capabilities.

Unmanned Robot Mission Execution:

Carries out missions autonomously through its own inference capabilities without communication.

Rapid Response:

Processes on the Edge for swift reactions without latency.

Large Multimodal Model Technology

Large Multimodal Model Technology Enhancing Driver Safety and Emergency Response with AI By leveraging vision-based driver behavior recognition, remote biometrics monitoring, intelligent emergency stop, and automated e-call systems, this technology significantly enhances driver safety, promptly detects and responds to critical situations, ensuring a secure and worry-free driving environment.

Vision-Based Situational Awareness:

Uses vision to perceive and assess the surrounding environment.

Remote Biometrics Monitoring:

Measure driver's heart rate and breathing patterns remotely.

Hybrid LMM:

Processes video streams at the central system for distances up to 400 km. If communication is lost, processing is handled on the edge.

400 km Long-Range HD Streaming:

Provides video transmission up to 400 km at Netflix HD quality and LMM decision-making via UAV without satellite assistance.

Red Queen Edge AI Terminal

Through quantization, we have significantly reduced the size of 8 billion parameters, enabling powerful LLM capabilities on our Red Queen Edge terminal.

Compact Size, Powerful AI Functionality:

Measuring only 15x20x15 cm, equipped with an Nvidia Nano and an 8-billion-parameter quantized LLM, this device is ready for versatile and rapid deployment across various applications.

Use Case: Unmanned Robot UAV:

Military Tier 3 drone. Capable of autonomously performing missions under GPS jamming and communication loss through Edge AI decision-making.

Use Case: Pedal Misapplication Warning Device:

Provides instant warnings in cases of pedal misapplication. Installation takes just 10 minutes, and it is available for purchase at 200,000 KRW (pre-order available).

Use Case: Driver and Pilot Biometrics:

Converts drivers' and pilots' heart rate and respiration data into EKG readings, enabling real-time risk detection and immediate autonomous control in critical situations.

Edge AI On-Device Technology

Edge AI On-Device technology enables artificial intelligence to run directly on local devices, such as smartphones, sensors, or autonomous vehicles, without relying on cloud computing. This allows for real-time data processing, faster response times, and reduced latency, making it ideal for applications where quick decisions are critical. By processing data locally, Edge AI also enhances privacy and security, as sensitive information doesn't need to be transmitted over networks.

Core Technologies

High-Performance Quantization Algorithm

Our advanced quantization algorithm achieves exceptional model compression while maintaining performance integrity. Through sophisticated parameter reduction and optimization, we've achieved:

Significant reduction in model parameters
Minimized execution time and power consumption
Enhanced computational efficiency
Optimized bandwidth utilization

Domain-Specific Fine-Tuning with Hybrid Quantization

The Red Queen Edge AI System employs a revolutionary hybrid approach combining Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT) technologies. This innovative solution delivers:

Optimized model size without compromising performance
Automatic Over-The-Air (OTA) updates for real-time model improvements
Environment-specific model optimization
Automated hyperparameter optimization

Automated Performance Optimization

Our system incorporates sophisticated automation for optimal performance:

Automated hyperparameter tuning
Self-optimizing learning rate adjustment
Dynamic batch size optimization
Intelligent epoch management

Quantifiable Performance Improvements

Through our advanced FP16 and INT4 quantization techniques, the Red Queen Edge AI System delivers:

Significant memory footprint reduction
3x speed improvement in processing
Enhanced fine-tuning capabilities
Optimized on-device execution

Edge Device Optimization

Specifically designed for edge deployment, our system achieves:

Minimized memory requirements
Accelerated inference speed
Efficient on-device execution
Reliable performance in resource-constrained environments

Features of Red Queen Edge AI Drone

At Blue Intelligence, we are proud to introduce the Red Queen Edge AI Drone-a revolutionary system that integrates cutting-edge Large Language Models (LLMs) and Edge AI technology into unmanned aerial vehicles. Designed for exceptional autonomy and precision, the Red Queen system empowers drones to perform complex missions with minimal human intervention, even in challenging environments with limited communication.

Long-Range UAV (VTOL Vertical Takeoff and Landing, 800 km Range):

Dimensions: 2.3 x 4.6 m, Weight: 4.5 kg, Payload Capacity: 10 kg, Maximum Altitude: 6.5 km.

Up to 400 km HD Video Streaming:

Utilizes advanced MIMO communication technology to deliver real-time video streaming up to 400 km at 1–3 Mbps (Netflix HD quality).

Edge AI and Cloud LLM Hybrid Functionality:

To enable video streaming up to 400 km, the UAV operates at an altitude of 1–3 km. If communication is lost, control is managed through Edge AI’s autonomous decision-making capabilities.

Advanced Target Recognition:

Equipped with sophisticated AI algorithms, our drones accurately identify and distinguish between various targets. This contextual understanding minimizes errors and enhances mission precision. By analyzing patterns and behaviors, the drone can predict target movements and adjust its strategy proactively.

WE ARE ANEXT GEN MOBILITY LAB