About
I am a PhD student at the School of Computing, National University of Singapore, advised by Prof. Harold Soh.
My research focuses on closing the deployment gap of pre-trained robot policies: how can we safely and flexibly use increasingly powerful learned policies in fast-changing, human-centered environments? To address this, I study inference-time steering methods that allow robot policies to adapt their behavior during execution, without retraining the underlying policy. My work explores how generative and streaming robot policies can be guided by dynamic constraints, human preferences, and structured predictions of future consequences.
Before joining NUS, I completed a B.A. in Engineering and an M.Eng. at the University of Cambridge, followed by an M.Res. at Imperial College London, where I was fortunate to be supervised by Prof. Nicole Salomons. I received First Class Honours / Distinction across my previous degrees.
Selected Publications
View All →Guided Streaming Stochastic Interpolant Policy
Puming Jiang†, Meiyi Wang†, Lin Kelvin, Ce Hao, Harold Soh
Proceedings of Robotics: Science and Systems (RSS)
GSSIP enables inference-time steering for streaming robot policies, where action generation is aligned with execution rather than planned as a fixed open-loop chunk. The framework derives guidance mechanisms from a stochastic interpolant formulation, allowing robots to adapt online to changing values (e.g. obstacle) while maintaining low control latency.
News
Our work GSSIP has been accepted by R:SS 2026 🎉
Starting my PhD at the National University of Singapore