Ge Yang

Papers

Total Citations

H-Index

About

Ge Yang is a robotics and machine learning researcher whose work spans legged locomotion, reinforcement learning, and autonomous systems. His most notable contribution, "Neural Volumetric Memory for Visual Locomotion Control" (2023), addresses one of robotics' most challenging frontiers: enabling legged robots to navigate difficult terrains using only a single forward-facing depth camera. By developing a neural volumetric memory architecture to handle partial observability, Yang's work meaningfully advances the autonomy of robots operating in unstructured, real-world environments — a critical step toward expanding robotic reach beyond controlled settings. The paper has already garnered 27 citations since its publication, reflecting strong community interest in its approach. Yang has also explored the theoretical underpinnings of multi-goal reinforcement learning through his work on "Bilinear Value Networks" (2022), which investigates how Q-value function generalization can improve data efficiency in off-policy learning — a fundamental challenge in building more sample-efficient AI agents. Taken together, his research sits at a rich intersection of perception, memory, and decision-making, making meaningful contributions to both the practical deployment of autonomous robots and the foundational algorithms that power them.