Task Graph

Task graphs represent the structure of complex tasks by depicting subtasks and their dependencies, aiming to improve task planning, execution, and understanding. Current research focuses on learning task graphs from various data sources, including videos, text transcripts, and knowledge graphs, employing techniques like graph neural networks, transformers, and reinforcement learning to optimize graph generation and inference. These advancements have implications for diverse fields, enhancing intelligent agent design, automated code repair, efficient knowledge graph utilization, and improved human-computer interaction in areas such as robotic task planning and instructional video analysis.