Richard Su

Beijing Academy of Artificial Intelligence

Papers

Total Citations

H-Index

About

Richard Su is a leading researcher at the intersection of computer vision, embodied AI, and 3D scene understanding. His primary contributions lie in developing automated pipelines for creating high-fidelity, semantically rich 3D environments that bridge the gap between simulation and reality. Su’s most notable work, "MetaScenes," introduces a groundbreaking framework for the automated replica creation of real-world 3D scans, addressing a critical bottleneck in embodied AI research: the need for diverse, high-quality scenes to support robust skill acquisition and sim-to-real transfer. By enabling precise replication of real-world object diversity, MetaScenes directly enhances the generalization capabilities of AI agents, with early work already garnering 2 citations since its 2025 publication. Su’s research is pivotal for advancing scalable training environments, reducing the manual labor traditionally required for scene creation, and accelerating progress in robotics and autonomous systems. His work is widely recognized for its potential to democratize access to high-quality 3D data, making him a key figure in the future of embodied intelligence and synthetic data generation.