Yuchen Song
Yuchen Song is a PhD student in Computer Science at the University of California, Irvine, advised by Professor Unnat Jain. He previously earned a Master's degree from UC San Diego under Professor Xiaolong Wang and interned at Salesforce AI Research. He is best known for his research on intelligent robots, including the GLIDE framework for learning from infeasible demonstrations and work on loco-manipulation and 3D-spatial multimodal memory.
“GLIDE solves this by inferring task-specific failure modes and converting them into executable guardrails that filter operator commands.”
Source→“"After refinement, GLIDE raises data-collection success from 0–10% to 70–90% across the three tasks"”
Source→“"GLIDE discovers emergent guardrails that go beyond domain-expert hardcoded ones, revealing task phases that are hard to specify a priori"”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.