<ul><li>A real-time AI-powered robotic system implemented in Unity3D simulates a card-sorting task, using computer vision to pick and place cards with an industrial robotic arm.</li><li>The system uses a custom-trained YOLO-based object detector for card detection, with the Unity Runtime Inference Engine enabling real-time perception and control.</li><li>The AI Builder Unity asset and Digital Twin Toolkit for Unity are utilized for training and integrating a Python-based AI vision training pipeline into Unity.</li><li>The application connects to the real robot via ROS running on a Raspberry Pi, allowing the digital twin to control operations in a synthetic environment before replicating them on the physical robot.</li></ul>

AI-powered Robotics: Pick and Place Demo with Runtime Interference and ROS in Unity

Discover more