Ruixuan Liu
Ruixuan Liu is a robotics researcher trained at Carnegie Mellon University who connects generative AI with robots that build with LEGO bricks. He earned a bachelor's degree in electrical and computer engineering there and began a PhD at the Robotics Institute in 2021, advised by Changliu Liu in the Intelligent Control Lab. In 2024, he led the author list of StableLego, a stability analysis for brick assemblies, and in 2025 he co-authored BrickGPT, whose paper on generating physically stable brick structures from text won the ICCV Best Paper Award (Marr Prize). His 2026 article Prompt-to-Product in IEEE Robotics and Automation Magazine adds a bimanual robot system that builds the designs. He defended his thesis, Prompt-to-Product: Towards Robots that Imagine, Reason, and Build, in October 2026 and is now an alumnus of the Intelligent Control Lab.
Timeline
-
2021
A robotics PhD at Carnegie Mellon
He earns a bachelor's degree in electrical and computer engineering with a robotics minor at Carnegie Mellon, and works in Sebastian Scherer's AirLab on sensor fusion and 3D reconstruction for structural inspection. In 2021, he enters the Robotics Institute's PhD program, advised by Changliu Liu in the Intelligent Control Lab.
-
2022
Handover demo for President Biden
In January 2022, he and Rui Chen demonstrate human-robot handover on a FANUC industrial robot at Mill 19 in Pittsburgh during President Biden's visit. A camera-based digital twin tracks the human and the robot, and a safety controller moves the robot away when the human gets close.
-
2024
The StableLego stability tool
He leads the author list of StableLego, published in IEEE Robotics and Automation Letters in 2024 with Kangle Deng, Ziwei Wang and Changliu Liu. The tool analyzes the structural stability of block-stacking brick assemblies, and StableLego provides more than 50,000 assembly structures with stability inferences.
-
May 2025
First prize at ICRA 2025
In May 2025, his team wins the $60k first prize in the What Bimanuals Can Do competition at the IEEE International Conference on Robotics & Automation (ICRA).
-
2025
BrickGPT: bricks from text
He is a co-author of BrickGPT, which trains an autoregressive language model to predict the next brick to add and uses physics-aware rollback to prune unstable predictions. The ICCV 2025 paper comes with StableText2Brick, a dataset of over 47,000 brick structures made from more than 28,000 unique 3D objects.
-
2025
A Marr Prize at ICCV
The BrickGPT paper, which he co-authors with Ava Pun, Kangle Deng, Deva Ramanan, Changliu Liu and Jun-Yan Zhu, wins the Best Paper Award (Marr Prize) at ICCV in 2025.
-
2026
The Prompt-to-Product system
Prompt-to-Product appears in IEEE Robotics and Automation Magazine in 2026, with him and Philip Huang as equal-contribution first authors. The system turns a natural-language prompt into a physically buildable LEGO brick design and builds it with a bimanual robotic system, and a user study shows that it lowers the barrier and reduces manual effort in creating assembly products. In March 2026, he also co-authors BrickSim, the first real-time physics-based simulator for interlocking brick assemblies.
-
October 2026
Thesis defense at Carnegie Mellon
In October 2026, he defends his PhD thesis, Prompt-to-Product: Towards Robots that Imagine, Reason, and Build, at the Robotics Institute. Changliu Liu chairs the committee, which includes Oliver Kroemer, Jean Oh, Jun-Yan Zhu and Ken Goldberg of UC Berkeley. He is now an alumnus of the Intelligent Control Lab.
Career
- From 2021Carnegie Mellon University
Education
- 2026Carnegie Mellon UniversityBachelor's degree in Electrical and Computer Engineering
Sources and credits
Photo: We have no published photo of Ruixuan Liu yet. Send us a free-licensed photo.
- Ruixuan Liu
- Student Ruixuan
- HRI Demo for President Biden
- Unleashing Creativity with AI and Robots
- RI PhD Thesis Defense - Ruixuan Liu
- Generating Physically Stable and Buildable Brick Structures from Text
- ICCV 2025 Open Access Repository: Generating Physically Stable and Buildable Brick Structures from Text
- The AI Tool That Could Make Manufacturing Faster and More Efficient — by Using Lego Bricks
- Alumni Ruixuan


