Jonathan Spencer

Papers

Total Citations

H-Index

About

Jonathan Spencer is a leading researcher in human-robot interaction and imitation learning, focusing on how robots can efficiently learn from natural, seamless human feedback. His work bridges the gap between explicit instruction and implicit behavioral cues, enabling more scalable and practical robot learning systems. Spencer’s most influential paper, “Learning from Interventions” (2020, 42 citations), introduces a framework where human-robot interaction serves as both explicit and implicit feedback, overcoming key limitations of traditional imitation learning that relies solely on off-policy demonstrations. He further advanced the field with “Expert Intervention Learning” (2021, 27 citations) and “Feedback in Imitation Learning: The Three Regimes of Covariate Shift” (2021, 17 citations), which systematically characterizes how interactive feedback can address the covariate shift problem—a critical challenge where conditioning on past actions leads to performance divergence in real-world deployment. Spencer’s work has been instrumental in making robot learning more robust and practical, with direct implications for assistive robotics, autonomous systems, and human-robot collaboration. His research is essential reading for anyone interested in scalable, interactive approaches to robot skill acquisition.