Mingshuo Xiao

Mingshuo Xiao

Embodied AI / VLA / Real-Robot Deployment

Tsinghua University

Research Interests

LeRobot / OpenPI
Real-Robot Rollout
Robot Systems Debugging
Modeling & Simulation

About

I am Mingshuo Xiao, an incoming direct PhD student at Tsinghua University with a background in mathematical sciences and engineering physics.

My current focus is Embodied AI, Vision-Language-Action models, robot imitation learning, and real-robot deployment. I care about the full loop from teleoperation data, dataset schema, model training, inference integration, rollout evaluation, to failure analysis.

The main project I want interviewers to inspect first is SO-ARM101 VLA real-robot deployment. I built a pickup-and-putdown pipeline around LeRobot/OpenPI, ACT, Diffusion Policy, SmolVLA, and pi0/pi0.5 deployment validation. I also have hands-on experience in dual-arm manipulation prototypes, RoboCup humanoid robot field deployment, STM32 robot control, and scientific simulation.

Fast entry for interviewers:

Interview Focus

What I can contribute quickly

  • LeRobot/OpenPI: data schema, training scripts, inference integration, and rollout debugging.
  • Real-robot rollout: success-rate tracking, stage pass rate, latency profiling, and failure taxonomy.
  • Robot systems debugging: Linux/Git workflow, ROS2 interfaces, network/IP setup, finite-state machines, and hardware issue diagnosis.
  • Modeling & simulation: engineering-physics background for physical-system modeling, parameter scans, and error analysis.

Recent Updates

2026-05

Launched my public embodied AI portfolio and personal homepage.

2026-05

Built a public SO-ARM101 project repository for real-robot imitation learning and VLA deployment notes.

2026-05

Joined THMOS in RoboCup China humanoid Small group and won silver / second place.

2026-09

Incoming direct PhD student at Tsinghua University.