Robot perception

Programming language : Python

System OS : Windows, Linux

Libraries used: Pytorch, Tensorflow, OpenCV, Networking, Pandas, Numpy, Pandas, Matplotlib, etc

Tags: vision, image, 2D & 3D, deep learning, instance segmentation, transformer, robotics, system integration, TCP/IP, GRPC, etc

Tasks

• Designed vision sensor system and training dataset
• Prototyped AI models and post-processing algorithms
• Researched vision AI models for robot pick&place and x-ray based pork deboning
• Deployed AI models to diverse operation sites
• Automated deployment process for faster installation

Difficulties

• Complexity in system integration of robot, vision, and AI
• Low latency required for real time performance
• Countless sales product in fullfillment center
• Insufficient training and validation datasets

Accomplishment

• Piece picking solution launched after sucessful demo
• Piece picking solution presented at KOREA MAT 2023 & Automation World 2023 (trade shows)
• Reduction in image data & labels cost
• Automated AI setup process to reduce installation time