Yitian Shi

Papers

Total Citations

H-Index

About

Yitian Shi is a rising researcher at the intersection of robotics, machine learning, and uncertainty quantification, with a primary focus on advancing robotic manipulation through generative modeling and probabilistic reasoning. Shi’s most cited work, “Diffusion models for robotic manipulation: a survey” (2025, 16 citations), provides a comprehensive synthesis of how diffusion generative models—originally successful in image and video generation—are being adapted to solve complex manipulation tasks, establishing a foundational resource for the field. In “vMF-Contact: Uncertainty-Aware Evidential Learning for Probabilistic Contact-Grasp in Noisy Clutter” (2025, 3 citations), Shi introduces a novel evidential learning framework that distinguishes between aleatoric uncertainty (from data noise) and epistemic uncertainty (from out-of-distribution objects), significantly improving grasp reliability in cluttered, sensor-noisy environments. Complementing this, “Uncertainty-driven Exploration Strategies for Online Grasp Learning” (2024, 3 citations) pioneers an online learning approach that leverages uncertainty estimates to guide exploration when adapting to unseen objects and novel bin configurations—a critical step toward real-world robotic deployment. Through these contributions, Shi is shaping how robots learn to grasp and manipulate objects under uncertainty, with clear impact for autonomous systems and industrial automation.