Ziang Cao

Papers

Total Citations

H-Index

About

Ziang Cao is a robotics researcher whose work spans visual navigation, sim-to-real transfer, and data-efficient robot learning. His most cited paper, "Orchard Vision Navigation Line Extraction Based on YOLOv8-Trunk Detection" (2024, 15 citations), introduces a specialized detection model that enables autonomous orchard robots to identify tree trunks and extract navigation lines, addressing a critical challenge in agricultural robotics. In "What Went Wrong? Closing the Sim-to-Real Gap via Differentiable Causal Discovery" (2023, 4 citations), Cao tackles the persistent problem of simulation-to-reality discrepancies by proposing a causal discovery framework that identifies and corrects dynamical mismatches, making simulated training more reliable for real-world deployment. His work "EquiBot: SIM(3)-Equivariant Diffusion Policy for Generalizable and Data Efficient Learning" (2024, 4 citations) advances imitation learning by incorporating symmetry-aware diffusion policies, enabling robots to learn manipulation tasks from limited demonstrations while generalizing across diverse environments. Cao's research demonstrates a clear trajectory from perception-driven navigation to fundamental learning challenges, with each contribution addressing practical barriers in deploying autonomous systems. His work on causal discovery and equivariant policies represents important steps toward more robust and generalizable robot learning.