Kara Liu

Papers

Total Citations

123

H-Index

About

Kara Liu is a robotics and machine learning researcher whose work sits at the intersection of visual perception, planning, and robotic manipulation. Her research addresses one of the central challenges in modern robotics: enabling robots to reason about and interact with complex, real-world objects that resist traditional analytical modeling. Liu's most influential contribution, "Learning Robotic Manipulation through Visual Planning and Acting" (2019), has accumulated over 90 citations and proposes a framework that allows robots to plan manipulation tasks directly from visual input, bypassing the need for hand-crafted physical models. This work has proven particularly relevant for domestic and industrial settings where object variability makes rigid-body assumptions impractical. Building on this foundation, her 2020 paper "Hallucinative Topological Memory for Zero-Shot Visual Planning" advances the field further by tackling zero-shot generalization — enabling agents to plan toward unseen goals without additional training. By moving beyond latent-space planning toward richer topological representations, Liu addresses persistent quality limitations in prior visual planning approaches. Collectively, her research has garnered over 120 citations, establishing her as a meaningful contributor to data-driven robot learning and a researcher worth following as autonomous manipulation systems continue to mature.