13 papers
EvoSpatial: Experience-Driven Skill and Tool Evolution for Spatial Reasoning Agents
Bin Yu, Shijie Lian, Zhaolong Shen +10
Spatial reasoning agents can inspect tool outputs and repair errors, but reusing these lessons requires preserving both the computation and the conditions that make it valid. We in…
SIEVE: Structure-Aware Data Selection for Imitation Learning with VLA Models
Changti Wu, Bin Yu, Zhaolong Shen +6
Vision-Language-Action (VLA) models are typically trained by imitation learning on large-scale robot demonstration datasets, but more data does not necessarily yield better policie…
Human-as-Humanoid: Enabling Zero-Shot Humanoid Learning from Ego-Exo Human Videos with Human-Aligned Embodiments
Xiaopeng Lin, Ruoqi Yang, Shijie Lian +14
Vision-language-action (VLA) models across robot embodiments require high-quality observation--action supervision to learn deployable action distributions, yet scaling such robot d…
RoboSemanticBench: Diagnosing Semantic Grounding in Action Prediction for VLA Models
Bin Yu, Yao Zhang, Haishan Liu +9
Vision-language-action (VLA) models are built on the premise that semantic understanding from pretrained language or vision-language backbones should guide robot action prediction.…
PhysBrain 1.0 Technical Report
Shijie Lian, Bin Yu, Xiaopeng Lin +10
Vision-language-action models have advanced rapidly, but robot trajectories alone provide limited coverage for learning broad physical understanding. PhysBrain 1.0 studies a comple…
FrameSkip: Learning from Fewer but More Informative Frames in VLA Training
Bin Yu, Shijie Lian, Xiaopeng Lin +8
Vision-Language-Action (VLA) policies are commonly trained from dense robot demonstration trajectories, often collected through teleoperation, by sampling every recorded frame as i…
IntentVLA: Short-Horizon Intent Modeling for Aliased Robot Manipulation
Shijie Lian, Bin Yu, Xiaopeng Lin +8
Robot imitation data are often multimodal: similar visual-language observations may be followed by different action chunks because human demonstrators act with different short-hori…
3D-Mix for VLA: A Plug-and-Play Module for Integrating VGGT-based 3D Information into Vision-Language-Action Models
Bin Yu, Shijie Lian, Xiaopeng Lin +8
Vision-Language-Action (VLA) models leverage Multimodal Large Language Models (MLLMs) for robotic control, but recent studies reveal that MLLMs exhibit limited spatial intelligence…
PEARL: Personalized Streaming Video Understanding Model
Yuanhong Zheng, Ruichuan An, Xiaopeng Lin +10
Human cognition of new concepts is inherently a streaming process: we continuously recognize new objects or identities and update our memories over time. However, current multimoda…
ScalSelect: Scalable Training-Free Multimodal Data Selection for Efficient Visual Instruction Tuning
Changti Wu, Jiahuai Mao, Yuzhuo Miao +6
Large-scale Visual Instruction Tuning (VIT) has become a key paradigm for advancing the performance of vision-language models (VLMs) across various multimodal tasks. However, train…
AD-MIR: Bridging the Gap from Perception to Persuasion in Advertising Video Understanding via Structured Reasoning
Binxiao Xu, Junyu Feng, Xiaopeng Lin +7
Multimodal understanding of advertising videos is essential for interpreting the intricate relationship between visual storytelling and abstract persuasion strategies. However, des…
TwinBrainVLA: Unleashing the Potential of Generalist VLMs for Embodied Tasks via Asymmetric Mixture-of-Transformers
Bin Yu, Shijie Lian, Xiaopeng Lin +8
The fundamental premise of Vision-Language-Action (VLA) models is to harness the extensive general capabilities of pre-trained Vision-Language Models (VLMs) for generalized embodie…
LangForce: Bayesian Decomposition of Vision Language Action Models via Latent Action Queries
Shijie Lian, Bin Yu, Xiaopeng Lin +6
Vision-Language-Action (VLA) models have shown promise in robot manipulation but often struggle to generalize to new instructions or complex multi-task scenarios. We identify a cri…