698.INTELLIGENT CONTROL OF KUKA ROBOTIC SYSTEMS BASED ON AI-DRIVEN HUMAN MOTION TRACKING

Abstract

Industrial robot control, often restricted by proprietary systems like KUKA Robot Language, presents a challenge for advanced scientific research and intuitive programming. This paper introduces a novel, low-cost framework for Intelligent Control of KUKA Robotic Systems using AI-Driven Human Motion Tracking to facili-tate kinesthetic teaching. The system integrates the MediaPipe Hands Deep Learning model for real-time 3D hand landmark tracking with a custom PyOpenShowVar/KUKAVARPROXY control middleware, enabling soft real-time command transmission to a KUKA KR 16-2. The framework achieved 97.3% accuracy for discrete gesture commands and used a Direct Landmark Differencing approach to provide intuitive, simultaneous control over 3D joint-space movement. While exhibiting 200ms soft real-time overhead, the performance is highly suitable for path teaching and signif-icantly lowers the technical barrier for human-robot collaboration and flexible manufacturing.