Zihao Liang

Papers

Total Citations

H-Index

About

Zihao Liang is a researcher advancing the field of inverse optimal control, with a focus on learning human preferences and objectives from partial trajectory data. His work addresses a fundamental challenge in robotics and autonomous systems: how to infer an agent's underlying goals from only short, incomplete demonstrations of behavior. Liang's key contributions include developing methods that can recover an unknown objective function—parameterized as a weighted sum of features—from demonstration segments, rather than requiring full optimal trajectories. His 2022 paper, "An Iterative Method for Inverse Optimal Control," introduces a novel iterative framework that progressively refines the learned objective using small data snippets provided at each step, enabling more efficient and practical learning from real-world demonstrations. His foundational 2020 work, "Inverse Optimal Control from Demonstration Segments," established the theoretical basis for this approach, showing how to extract meaningful reward structures from partial optimal trajectories. With combined citations of 8, Liang's research is gaining traction in the learning-from-demonstration and optimal control communities, offering a scalable solution for teaching robots complex tasks through human demonstration.