Peiqi Liu | 刘沛淇

I am currently an early intern at LiberAI, working on world models for robotics. I am also a senior undergraduate student majoring in computer science at Peking University, advised by Prof. Hao Dong.

Previously, I was a visiting student at MIT CSAIL, where I worked in the groups led by Prof. Leslie Pack Kaelbling and Prof. Josh Tenenbaum, advised by Jiayuan Mao. I also spent a semester on exchange at the University of California, Berkeley, and interned with the BAIR.

Email  /  Github /  LinkedIn /  Book a Meeting

Research

My long-term research goal is to build robots that can acquire new skills as efficiently as humans, generalize across diverse tasks, and perform everyday physical labor. I am particularly interested in exploring novel algorithms and representations that improve the efficiency of robot skill acquisition and adaptation, enabling one-shot and zero-shot learning.

Now: I am growing into a full-stack robotics researcher, working across algorithms and hardware.

BiDexAffordance is a collaborative affordance-driven framework that learns object-centric, physics-grounded bimanual affordance maps to efficiently generate robust and generalizable bimanual dexterous grasps across diverse and unseen objects.

A framework for lifelong experience abstraction and planning that enables agents to learn and adapt continuously across different environments and tasks.

We propose a multi-object demand-driven navigation benchmark and train an coarse-to-fine attribute-based exploration agent to solve this task.

Experience

Template, Last updated: Jun 2026 © Peiqi Liu