5 papers
BooST: Bridging Semantics and Motions for Efficient Skill Transfer
Jusuk Lee, Daesol Cho, Jonghun Shin +4
Skill abstraction---the process of learning reusable and temporally extended behaviors---has emerged as a key paradigm for improving sample efficiency and generalization in robot l…
: A Scalable 3D Interaction-Trace World Model
Seungjae Lee, Yoonkyo Jung, Jusuk Lee +6
World models that capture how actions induce physical change enable scalable robot learning without reliance on embodiment-specific action labels. Pixel-space video models provide…
FLAG: Flow Policy MaxEnt-RL by Latent Augmented Guidance
Sungha Kim, Gawon Lee, Jusuk Lee +3
Maximum entropy reinforcement learning (MaxEnt-RL) enables robust exploration, yet practical implementations often restrict policies to simple Gaussians. While recent approaches in…
DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation
Jusuk Lee, Seungjae Lee, Jonghun Shin +6
Robot manipulation critically depends on perception that preserves the action-relevant aspects of a scene. Yet most robot learning pipelines are built upon visual encoders pre-trai…
Periodic Skill Discovery
Jonghae Park, Daesol Cho, Jusuk Lee +3
Unsupervised skill discovery in reinforcement learning (RL) aims to learn diverse behaviors without relying on external rewards. However, current methods often overlook the periodi…