Ziyan Xiong

Papers

Total Citations

H-Index

About

Ziyan Xiong is a researcher advancing the frontiers of reinforcement learning (RL) and robotics, with a focus on bridging the gap between simulation and real-world deployment. Their key contributions lie in developing data-efficient, model-based RL algorithms—specifically, offline world models that can be finetuned directly in physical environments. In their notable 2023 work, "Finetuning Offline World Models in the Real World," Xiong tackles the critical challenge of RL’s notorious data inefficiency, which often makes real-robot training impractical. By leveraging offline RL to pre-train world models from static datasets, then finetuning them with minimal real-world interaction, they demonstrate a path to drastically reduce the hours or days of robot training typically required. This approach has already garnered attention, with the paper accumulating 2 citations in its early stages, signaling growing interest from the RL and robotics communities. Xiong’s work is especially impactful for students and researchers seeking to deploy RL in resource-constrained, real-world settings, offering a pragmatic solution to one of the field’s most persistent bottlenecks. Their research promises to accelerate the adoption of intelligent robots in manufacturing, healthcare, and beyond.