Hi there!
I am a Research Scientist working at the intersection of robotics and machine learning to equip embodied agents with greater understanding of the world for better decision making. I am excited about developing 3D representations and using them in decision making algorithms to enable robust, general-purpose embodied agents, especially as data scales. I am currently with the Large Behavior Models research team at the Toyota Research Institute.
I have a PhD in Aeronautics and Astronautics from MIT. During my graduate studies, I conducted research with the Robust Robotics Group at MIT’s Computer Science and Artificial Intelligence Laboratory, advised by Dr. Nicholas Roy. My thesis focused on navigation and estimation in GPS-denied environments of size, weight, and power constrained vehicles. Before that, I obtained my B.S. in Mechanical Engineering from the University of California, San Diego.
Below are a few selected publications. For more, check out my Research page and my Google Scholar. You can email me at the dot katherine dot liu at gmail dot com.
Recent News
February 2025: One paper accepted to CVPR 2025 (details to follow)!
January 2025: OmniShape: Zero-Shot Multi-Hypothesis Shape and Pose Estimation in the Real World accepted to ICRA 2025 — see you in Atlanta!
September 2024: View-Invariant Policy Learning via Zero-Shot Novel View Synthesis accepted to CoRL 2024 and SE(3) Equivariant Ray Embeddings for Implicit Multi-View Depth Estimation accepted to NeurIPS
July 2024: Zero-Shot Multi-Object Scene Completion accepted to ECCV 2024
June 2024: DiffusionNOCS accepted to IROS 2024
May 2024: Recursive Field Networks for Cross-Modal Multi-Scene Representations accepted to SIGGRAPH 2024
July 2023: NeO 360 accepted to ICCV 2023 NeO 360: Neural Fields for Sparse View Synthesis of Outdoor Scenes
February 2023: Two papers accepted to CVPR 2023:
CARTO: Category and Joint Agnostic Reconstruction of ARTiculated Objects and
Multi-Object Manipulation via Object-Centric Neural Scattering FunctionsSeptember 2022: Our paper ROAD: Learning an Implicit Recursive Octree Auto-Decoder to Efficiently Encode 3D Shapes has been accepted to CoRL 2022. [twitter thread]
June 2022: I joined the Toyota Research Institute as a Machine Learning Research Scientist.
Selected Publications
-
ReFiNe: Recursive Field Networks for Cross-modal Multi-scene Representation. S Zakharov, K Liu, A Gaidon, R Ambrus. To appear at SIGGRAPH 2024. [pdf][site]
ReFiNe can encode multiple objects represented as neural fields via a recursive formulation that allows for a single latent vector to be expanded into an octree representing the object.
-
-
Learned Sampling Distributions for Efficient Planning in Hybrid Geometric and Object-Level Representations. K Liu꙳, M Stadler꙳, N Roy. ICRA 2020. [pdf][video][read more]
We enable more efficient navigation by training a neural network to predict a sampling distribution from object-level maps and occupancy maps.
-
Robust Object-Based SLAM for High-Speed Autonomous Navigation. K Ok꙳, K Liu꙳, K Frey, JP How, N Roy. ICRA 2019. [pdf][read more]
Robust Object-Based SLAM for High-Speed Autonomous Navigation (ROSHAN) improves object-level SLAM with ellipsoid landmarks for autonomous navigation by incorporating texture and semantic measurements.
-
Search and Rescue Under the Forest Canopy Using Multiple UAVs. Y Tian, K Liu, K Ok, L Tran, D Allen, N Roy, J P How. ISER 2019 and IJRR 2020. [pdf][video]
We describe a system for a real-world deployment of a multi-agent system of quadrotors operating in a GPS-denied scenario. This project was a collaboration with NASA and one of the recipients of the AUVSI Humanitarian Award in 2019.
-
Deep Inference for Covariance Estimation: Learning Gaussian Noise Models for State Estimation. K Liu꙳, K Ok꙳, W Vega-Brown, N Roy. ICRA 2018. [pdf][video]
Deep Inference for Covariance Estimation (DICE) uses a deep convolutional neural network to predict noise models for high dimensional, complex sensors.
