Forrest Meng

Papers

Total Citations

H-Index

About

Forrest Meng focuses on the intersection of robotics and human-robot interaction, with a particular emphasis on online reward learning. His key research area addresses a fundamental challenge: enabling robots to learn a human’s preferences in real time during a single interaction, despite noisy or suboptimal human demonstrations. Meng’s major contribution is the development of **StROL** (Stabilized and Robust Online Learning from Humans), a framework designed to stabilize the approximate learning rules that robots must use when processing human feedback on the fly. By tackling the instability that plagues fast, approximate learning algorithms, StROL allows robots to more reliably infer a human’s reward function even when the human’s behavior is imperfect. This work has garnered early attention, with the 2024 version accumulating 4 citations and the 2023 version 2 citations, signaling growing interest in robust, real-time human-robot teaching. Meng’s research is particularly notable for addressing a practical bottleneck in deploying assistive or collaborative robots: the need for them to adapt quickly and safely to individual users without requiring extensive offline training data.