About me

★ I am currently looking for academic and industry positions. Feel free to reach out if you are interested in chatting or collaborating. I am always open to new ideas and opportunities.

I am a Ph.D. Candidate in RoboLAND at the University of Southern California (USC), advised by Dr. Feifei Qian, where I aim to transform legged robots into environment-aware co-scientists for planetary and Earth field exploration. I developed direct-drive robots that turns legs and bodys as robust active force sensors, enabling real-time estimation of the physical world during walking and adaptive planning and navigation beyond the limits of vision or LiDAR. My work contributes to NASA projects such as LASSIE, which deploys legged robots for science discovery, and TRUSSES, which uses legged robots as scouts and rescuers to assist large rovers overcome sandy entrapments.

I also had the fortune to work with and be advised by Dr. Gaurav S. Sukhatme, where I focused on building embodied proactive agents for human-robot collaboration. I developed frameworks for foundation models to better understand and reason about complex task dependencies between humans and robots, utilizing the combination of fast-slow agents to allow robots to proactively provide suggestions and instructions to humans in real-time instead of waiting for humans to ask for help.

News

2025-8 Successfully completed our second field testing expedition at White Sands National Park, New Mexico, demonstrating the LASSIE&Trussess robot's capabilities in challenging terrain and field operations workflow.
ws251 ws252 ws253 ws254 ws255
2025-6 Robotics: Science and Systems event:
  • RoboLAND participated in RSS 2025!
  • I presented my paper at RSS - "Adaptive Locomotion on Mud through Proprioceptive Sensing of Substrate Properties"
  • Our lab presented a lab tour and a live robot demo at RSS.
RSS 2025 Conference
2025-6 Our paper "Characterizing Robot-Ground Interactions for Autonomous Lunar Construction" has been accepted to the 19th International Symposium on Experimental Robotics (ISER'25)
2025-5 I passed the qualifying exam, now become a Ph.D. candidate!
2025-4 Our paper "Adaptive Locomotion on Mud through Proprioceptive Sensing of Substrate Properties" is accepted to Robotic Systems and Science (RSS)!

Research

Research illustration

My work sits at the intersection of embodied intelligence, active sensing, and human–robot collaboration. I build robots that don't just move through the world—they feel through interaction, learn from active manipulation, and team with humans. Across three directions, I connect foundation models, planning under uncertainty, and proprioceptive-inspired sensing into a unified vision of proactive, adaptive, and collaborative embodied agents.

1) Foundation Models & Embodied Agents
I design embodied, proactive agents that integrate foundation models to interacts with humans. These agents coordinate fast–slow (reactive–deliberative) behaviors, anticipate human needs, and produce real‑time guidance, instructions, and plans—enabling robots to act as collaborative teammates rather than passive tools.

2) Robot Planning & Navigation
I develop planning systems that go beyond vision/LiDAR by leveraging proprioception and contact to navigate uncertain, deformable terrain. Treating locomotion as information‑gathering enables risk estimation and online adaptation under partial observability, advancing safe exploration in planetary and field science settings.

3) Embodied Tactile Sensing
I engineer direct‑drive legged robots that transform locomotion into embodied force sensing—legs functioning as force sensors to probe and infer substrate properties (e.g., resistance, shear strength). This proprioceptive‑informed understanding feeds directly into planning and navigation and supports NASA‑aligned efforts such as LASSIE (legged robots as co‑scientists) and TRUSSES (robot teaming to free rovers from sandy entrapments).

I'm currently on the job market and open to opportunities in both academia and industry.

Foundational models & embodied agents

Adaptive Communication Support for Human-Robot Collaboration

Submitted to ACM Transactions on Human-Robot Interaction (THRI)

Workshop on Multimodal AI Collaborations @ AAAI 2025

foundation model embodied adaptive agent human-robot collaboration

Shipeng Liu, FNU Shrutika, Boshen Zhang, Zhehui Huang, Gaurav S. Sukhatme, and Feifei Qian

Project Page Code Arxiv PDF
  • - Submitted to ACM Transactions on Human-Robot Interaction (THRI)
  • - Workshop on Multimodal AI Collaborations @ AAAI 2025
  • - Novel LLM-powered framework with DAG architecture for adaptive communication
  • - Dual-module system: Coordinator (strategic) + Manager (tactical) for human-robot teaming
  • - Dynamic transition between passive and active interaction modes based on task complexity

Towards Real-time Adaptation of Embodied Agent in Human-Robot Collaboration

Language Models for Planning Workshop @ AAAI 2025

real-time agents fast-slow reasoning finetuned-foundation model

Shipeng Liu†, Boshen Zhang†, Zhehui Huang, Feifei Qian, and Gaurav S. Sukhatme

Project Page Code Arxiv PDF
  • - Submitted to Language Models for Planning Workshop @ AAAI 2025
  • - Novel fine-grained benchmark with 22 layouts to assess proactive adaptability and temporal responsiveness
  • - MonTA framework using hierarchical LLM approach (fast monitor + slow adapters)
  • - Real-time adaptation capabilities enabling proactive decision-making in collaborative scenarios
  • - 75% reasonability in human expert evaluations of adaptation plans and instructions
  • - Significant performance improvements in low-teaming-fluency scenarios

Discrete Multi-Agent Path Finding via Discrete Categorical Diffusion with Constraint Projections

diffusion for planning categorical discrete diffusion multi-agent path finding

Shipeng Liu et al.

Paper (in preparation)
  • - Novel approach to discrete multi-agent path finding using categorical diffusion models
  • - Constraint projection methods for ensuring feasible solutions
  • - Discrete optimization techniques for multi-agent coordination

Robot planning & navigation

Scout-Rover cooperation: online terrain strength mapping and traversal risk estimation for planetary-analog explorations

LPSC 2025

terrain-mapping risk-aware multi-rover navigation

Shipeng Liu, J. Diego Caporale, William Hoganson, Xingjue Liao, Yifeng Zhang, Shivangi Misra, Neha Peddinti, Rachel Holladay, Wilson Hu, Ethan Fulcher, Uland Wong, Daniel E. Koditschek, Cynthia Sung, and Feifei Qian

Paper
  • - In submission
  • - Scout-rover cooperation for online terrain strength mapping and traversal risk estimation
  • - Planetary-analog exploration applications
  • - Multi-institutional collaboration across multiple institutions

Safe Active Navigation and Exploration for Planetary Environments Using Proprioceptive Measurements

Multi-Objective Optimization Workshop @ RSS 25

safe navigation potential-field

Shipeng Liu†, Matthew Jiang†, and Feifei Qian

Paper
  • - Multi-Objective Optimization Workshop @ RSS 25
  • - Safe active navigation and exploration using proprioceptive measurements
  • - Joint work towards robust planetary navigation policies

Build direct-drive robots with embodied tactile sensing

Bio-inspired tail oscillation enables fast crawling on deformable granular terrains

Arxiv

bio-inspired robot design tail oscillation granular media

Shipeng Liu, Meghana Sagare, Shubham Patil, and Feifei Qian

arXiv
  • - Bio-inspired tail oscillation for fast crawling on deformable granular terrains
  • - Demonstrates effective locomotion strategies leveraging body-tail interactions

Adaptive Locomotion on Mud through Proprioceptive Sensing of Substrate Properties

RSS (Robotics: Science and Systems), 2025

proprioceptive sensing adaptive locomotion granular media

Shipeng Liu, Jiaze Tang, Siyuan Meng, and Feifei Qian

Project Page Code Arxiv PDF
  • - RSS (Robotics: Science and Systems), 2025
  • - Novel proprioceptive sensing method to estimate mud properties through actuator signals
  • - RFT-based approach for characterizing substrate strength and force responses
  • - Flipper-driven robot design for adaptive locomotion on varying muddy terrains
  • - Real-time adaptation prevents locomotion failures in complex, deformable natural environments

Modeling Experts' Sampling Strategy to Balance Multiple Objectives During Scientific Explorations

ACM/IEEE International Conference on Human-Robot Interaction (HRI 2024)

multi-objective decision making human-robot teaming field robotics 🏆 Best Paper Finalist (5%)

Shipeng Liu, Cristina G. Wilson, Zachary I. Lee, and Feifei Qian

Link PDF Web interface 1 Web interface 2
  • - Proceedings of the 2024 ACM/IEEE International Conference on Human-Robot Interaction (HRI '24)
  • - 🏆 Best Paper Finalist (5%)
  • - Developed models to understand expert sampling strategies in scientific exploration
  • - Created web interfaces for interactive exploration of sampling strategies
Cover image for flipper-mud interactions project

Adaptation of Flipper-Mud Interactions Enables Effective Terrestrial Locomotion on Muddy Substrates

IEEE International Conference on Robotics and Automation (ICRA), 2024 & IEEE Robotics and Automation Letters (RAL)

bio-inspired robots granular media adaptive locomotion

Shipeng Liu, Boyuan Huang, and Feifei Qian

Project Page Code Arxiv PDF
  • - IEEE International Conference on Robotics and Automation (ICRA), 2024 & IEEE Robotics and Automation Letters (RAL)
  • - Novel force modeling approach measuring horizontal (shear) and vertical (extraction) forces
  • - Mudskipper-inspired robot design for studying terrestrial locomotion on muddy substrates
  • - Identified two distinct failure mechanisms: high water content (slippage) and low water content (entrapment)
  • - Non-monotonic performance dependence on mud water content with optimal performance at 25%-26%
  • - Adaptation strategies increased robot speed by more than 200%
Cover image for understanding human dynamic sampling objectives project

Understanding Human Dynamic Sampling Objectives to Enable Robot-assisted Scientific Decision-Making

ACM Transactions on Human-Robot Interaction (THRI)

human decision-making modelling human-robot teaming field robotics

Shipeng Liu, Cristina G. Wilson, Bhaskar Krishnamachari, and Feifei Qian

Link PDF Web interface
  • - ACM Transactions on Human-Robot Interaction (THRI)
  • - Presented at ICRA Workshop, May 2022
  • - Presented at American Geophysical Union Fall Meeting, New Orleans, LA, 2021
  • - Developed frameworks for understanding human sampling objectives in scientific exploration

Other Publications

A Trajectory-Based Control Strategy with Vehicle Cooperation and Absolute Transit Priority at an Isolated Intersection

A Trajectory-Based Control Strategy with Vehicle Cooperation and Absolute Transit Priority at an Isolated Intersection

Journal of Advanced Transportation, 2024

autonomous driving vehicle cooperation

Zhen Zhang, Jintao Lai, Fangkai Wang, Xiaoguang Yang, Shipeng Liu, Mingyu Zhang

Link
  • - Journal of Advanced Transportation, 2024
  • - Developed trajectory-based control strategies for intersection management
  • - Implemented vehicle cooperation and transit priority systems
Characterizing Robot-Ground Interactions for Autonomous Lunar Construction

Characterizing Robot-Ground Interactions for Autonomous Lunar Construction

19th International Symposium on Experimental Robotics (ISER'25)

direct-drive rover design lunar field testing robot-ground interaction

Rachel Holladay, Shivangi Misra, Mason Mitchell, Akshay Ram Panyam, Erica Pauline Santos, J. Diego Caporale, Shipeng Liu, Yifeng Zhang, John Ruck, Douglas Jerolmack, Feifei Qian, Mark Yim, and Cynthia Sung

Paper
  • - ISER'25: Characterizing robot-ground interactions for autonomous lunar construction

Meetscript: Designing transcript-based interactions to support active participation in group video meetings

ACM Conference on Computer-Supported Cooperative Work & Social Computing (CSCW 2023)

ai aided group meeting tools ai for human human communication

Xinyue Chen, Shuo Li, Shipeng Liu, Robin Fowle, and Xu Wang

DOI Video
  • - CSCW 2023
  • - Designs transcript-based interactions to foster active participation

Media

My research on legged robots for space exploration has been featured in various media outlets worldwide:

Universe Today: Walking Moon Robots Possibly More Reliable than Lunar Rovers Universe Today
Read Article
OPB: How a dog-like robot is training for space exploration on Mount Hood Oregon Public Broadcasting
Read Article
KGW 8 - Portland, OR: Researchers using a four-legged robot on Mount Hood to help them land on the moon KGW 8 Portland
Read Article
CBC Kids: A new kind of rover: Dog-like robot trained to explore the moon CBC Kids
Read Article
Koin 6 - Portland, OR: Robot dog training to walk on the moon on Mt. Hood Koin 6 Portland
Watch Video
KATU 2 - Portland, OR: NASA tests walking robot on Mount Hood for space exploration with universities KATU 2 Portland
Read Article
CBS News - Los Angeles, CA: NASA and USC Robotics Team Up to Strengthen Space Exploration Program CBS News Los Angeles
Read Article
CBS Chicago: Robotic dog testing out surface for future exploration on the moon CBS Chicago
Watch Video
IEEE Spectrum: Video Friday: LASSIE On the Moon IEEE Spectrum
Read Article
Futurism: NASA quietly training robot dog to navigate landscape of the moon Futurism
Read Article
Reuters: Scientists train a robot to walk on the moon Reuters
Watch Video
NPR Weekend Edition: A robot dog is training on Earth to be able to go to space one day NPR Weekend Edition
Read Article
BBC: Robot dog trains to walk on Moon in Oregon trials BBC
Watch Video
CBC: Meet Spirit, a robot being trained to walk on the moon CBC
Watch Video
7News Australian: Robot dog trained to walk on the moon 7News Australian
Watch Video

Awards & Honors

MHI Scholar Award Finalist USC
USC Top-off Fellowship USC
CURVE PhD Mentor Award USC
Best Paper Finalist HRI 2024
WiSE Travel Award USC
Provost's PhD Graduate School Fellowship USC
Outstanding Graduate Student Tongji University

Teaching Experience

Fall 2024
Linear Algebra Teaching Assistant
Spring 2024
Machine Learning Teaching Assistant
Fall 2023
Robotic Mobility Teaching Assistant
Spring 2023
Machine Learning Teaching Assistant
Fall 2022
Robotic Mobility Teaching Assistant
Fall 2018
Open Source Hardware and Programming Teaching Assistant

Projects

Build direct-drive robots and implementation of robot control

turtle robot
turtle robot 2
turtle robot 3
operation interface

Build direct-drive robots

Use gearless brushless motors to build legged robots from scratch including 3d cad design, hardware design, and software control.

Implemented with motor control, can communication, and robot inverse kinematics controller and integrated them using ROS2 operation system.

From the proprioceptive information of each motor, we treat each motor as a torque sensor, and use it to estimate the external force.

Combining the estimated force information, and the changes of the physical world/terrain (from encoders and vision), we can understand the physical properties of world (terrain, object, etc.).

Skills:

3D cad design Hardware/Electronic integration and debugging Robot inverse kinematics controller ROS2 operation system

Github repo:

lassie-traveler lassie-turtle

Robot Planning & Navigation

Robot Planning & Navigation

Building interface based on foxglove for robot mapping,planning and navigation visualization.

Use Gaussian Process for mapping the terrain properties based on the proprioceptive information and develop potential field planning for robot navigation.

Integrate the entire workflow into a ros2 project to allow human to specify the planning and navigation goals

Conduct real-world field testing in NASA Ames Research Center, WhiteSands National Park, and USC campus.

Skills:

Foxglove Planning and potential field based navigation ROS2 operation system

Github repo:

SpiritHighLevel

Operation interface

Human-robot interaction interface

Build web interface for human-robot interaction using react, javascript and flask

Build agents to discuss with human and help human in lauguage about the high level samplign planning and navigation goal.

Skills:

React, Javascript, Flask LLM-powered agents RAG (Retrieval-Augmented Generation)

Github repo:

lassie-decisionmaking

Autonomous driving simulation

CARLA simulation 1
CARLA simulation 2
CARLA simulation 3
CARLA simulation 3
CARLA simulation 3
placeholder image

Autonomous driving simulation generaltion

Build a scenario generator for autonomous driving simulation based on CARLA

Build a decision making testing case zoo for autonomous driving decision making testing

open-source in github

CARLA ROS2 Autonomous driving
DI-drive Case Zoo CVSC (Carla Vehicle Simulator)