Yao-Hung Hubert Tsai

About

Yao-Hung Hubert Tsai is a rising researcher at the intersection of embodied AI, safe reinforcement learning, and self-supervised robotics. His work focuses on enabling mobile robots and agents to operate autonomously and safely in unstructured real-world environments—without requiring exhaustive human annotation or pre-mapped spaces. In his highly cited work "SAFER," Tsai introduces a collision avoidance system that combines focused trajectory search with reinforcement learning to correct operator commands in real time, significantly improving safety for dynamic navigation. This paper has garnered 6 citations and represents a practical step toward deployable, risk-aware robotic systems. In parallel, Tsai’s research on "Self-Supervised Object Goal Navigation with In-Situ Finetuning" tackles a critical bottleneck in household robotics: the ability to locate target objects without pre-annotated semantic maps. By enabling robots to finetune their navigation policies on the fly using only self-supervised signals, Tsai reduces the gap between simulated training and real-world deployment. His contributions are particularly notable for their emphasis on real-robot validation, a step often omitted in the field. With a growing citation footprint and a focus on scalable, annotation-free learning, Tsai is shaping the next generation of autonomous agents that can safely and intelligently navigate human spaces.

Research Focus

Key Achievements

Top Papers

1. 1
   SAFER: Safe Collision Avoidance Using Focused and Efficient Trajectory Search with Reinforcement Learning

   6 citations · 2023
2. 2
   Self-Supervised Object Goal Navigation with In-Situ Finetuning

   3 citations · 2023

Key Collaborators

Contact & Links