Hong Jun Jeon

Stanford University, South China University of Technology

Papers

Total Citations

H-Index

About

Hong Jun Jeon’s research sits at the intersection of reward learning, human-robot interaction, and assistive robotics, with a focus on enabling machines to infer and act upon human intent. His most influential work, “Reward-rational (implicit) choice: A unifying formalism for reward learning” (2020, 56 citations), provides a foundational framework for understanding how reward functions can be learned from diverse forms of human behavior—moving beyond simple demonstrations to capture implicit choices and preferences. This formalism has become a key reference for researchers tackling the challenge of specifying correct reward functions in complex tasks. In assistive robotics, Jeon has advanced shared autonomy systems that help people with disabilities perform everyday activities. His work on “Shared Autonomy with Learned Latent Actions” (2020) and “Learning latent actions to control assistive robots” (2021) introduces methods for robots to interpret coarse and fine-grained human motions, enabling more fluid collaboration during tasks like eating or reaching. Earlier, his exploration of “Configuration Space Metrics” (2018) challenged default Euclidean assumptions in robot motion planning, offering new distance measures that improve task-constrained manipulation. Through these contributions, Jeon is shaping how robots learn from and assist humans in real-world settings.