MOBILUS

Mobilus Inc., a company specializing in autonomous mobility solutions, has been selected for a Regional Innovation Cluster Collaborative R&D Project by Korea's Ministry of Science and ICT.

The project is officially titled "Development of an AI-Based Autonomous Driving & Autonomous Operation Tractor Controller with Ultra-Low Latency Remote Supervision." The project spans 33 months from April 2026 to December 2028, with a total budget of KRW 2.93 billion (KRW 2.3 billion government-funded).

The project is led by Mobilus in consortium with the Gyeongbuk IT Convergence Industry Technology Institute and Kyungpook National University Industry-Academic Cooperation Foundation.

■ Background and Objectives

 Korea's agricultural sector faces a growing labor shortage driven by rural population decline and an aging workforce, accelerating the need for smart farming technologies. At the same time, AI-only autonomous systems have inherent limitations in handling edge cases in unstructured farmland environments, making the integration of real-time remote intervention a critical industry requirement.

Against this backdrop, the project aims to develop an integrated system combining RTK-GNSS, IMU, Radar, Camera, AI, and 5G ultra-low latency video communication technologies to deliver:

▲an autonomous driving and autonomous operation tractor controller

▲an ultra-low latency remote control system

▲a unified AI training and deployment platform

■ Key Technical Deliverables

 On the hardware side, the project will develop an Automotive-grade, high-performance Edge AI controller combining the processor with Dual NPUs, integrated with a 5G network module, high-resolution cameras, and radar sensors.

 On the software side, the team will advance algorithms targeting automated farmland recognition accuracy above 90%, path-following error below 7 cm, and emergency stop time within 2 seconds.

 For remote control capabilities, the project targets ultra-low latency video streaming of under 200ms and a 98% success rate for transitioning from emergency stop to remote control intervention, built on a WebRTC-based unified teleoperation platform.

 On the AI infrastructure side, a MLOps environment and an integrated data collection, processing, and labeling management server will be established, targeting quantized model performance of 95%+ for Object Detection and 90%+ for Semantic Segmentation.

CEO ByongHo Cho stated, "The integration of autonomous driving technology with ultra-low latency remote control is the key to ensuring the safety and continuity of AI systems in real-world agricultural environments," adding, "Through this project, we aim to become the leading provider of a Physical AI-based autonomous operation platform in the global agricultural machinery market."