5 papers
CODA: Depth-Aligned Scene Completion and Object Decomposition from a Single RGB-D Image
Dongwon Son, Junhyek Han, Yoontae Cho +5
Robots operating safely in cluttered everyday environments often need to infer scene geometry from partial observations. Methods that detect objects in 2D and reconstruct them inde…
TAM: Torque Adaptation Module for Robust Motion Transfer in Manipulation
Dongwon Son, Florian Shkurti, Jason Lee +3
A policy tuned for one robot often behaves differently on another, whether due to the sim-to-real gap, unknown payloads, or the differing dynamics of two instances of the same robo…
NeuralSVCD for Efficient Swept Volume Collision Detection
Dongwon Son, Hojin Jung, Beomjoon Kim
Robot manipulation in unstructured environments requires efficient and reliable Swept Volume Collision Detection (SVCD) for safe motion planning. Traditional discrete methods poten…
DEF-oriCORN: efficient 3D scene understanding for robust language-directed manipulation without demonstrations
Dongwon Son, Sanghyeon Son, Jaehyung Kim +1
We present DEF-oriCORN, a framework for language-directed manipulation tasks. By leveraging a novel object-based scene representation and diffusion-model-based state estimation alg…
An intuitive multi-frequency feature representation for SO(3)-equivariant networks
Dongwon Son, Jaehyung Kim, Sanghyeon Son +1
The usage of 3D vision algorithms, such as shape reconstruction, remains limited because they require inputs to be at a fixed canonical rotation. Recently, a simple equivariant net…