AI for Auto-Research: Roadmap & User Guide
Paper • 2605.18661 • Published • 71
papers
-
-
StableVLA: Towards Robust Vision-Language-Action Models without Extra Data
Paper • 2605.18287 • Published • 15 -
MixSD: Mixed Contextual Self-Distillation for Knowledge Injection
Paper • 2605.16865 • Published • 10 -
MolmoPoint: Better Pointing for VLMs with Grounding Tokens
Paper • 2603.28069 • Published • 9 -
VersaViT: Enhancing MLLM Vision Backbones via Task-Guided Optimization
Paper • 2602.09934 • Published • 1 -
Flash-WAM: Modality-Aware Distillation for World Action Models
Paper • 2606.05254 • Published • 7 -
Latent Reasoning with Normalizing Flows
Paper • 2606.06447 • Published • 8 -
Discrete-WAM: Unified Discrete Vision-Action Token Editing for World-Policy Learning
Paper • 2606.05645 • Published • 2 -
Dream.exe: Can Video Generation Models Dream Executable Robot Manipulation?
Paper • 2606.04811 • Published • 17 -
Cosmos 3: Omnimodal World Models for Physical AI
Paper • 2606.02800 • Published • 142 -
Qwen-Image-Flash: Beyond Objective Design
Paper • 2606.03746 • Published • 37 -
Filter, Then Reweight: Rethinking Optimization Granularity in On-Policy Distillation
Paper • 2606.02684 • Published • 17 -
GRAIL: Generating Humanoid Loco-Manipulation from 3D Assets and Video Priors
Paper • 2606.05160 • Published • 9 -
Unlocking Feature Learning in Gated Delta Networks at Scale
Paper • 2606.04048 • Published • 2 -
SpatialAct: Probing Spatial Reasoning-to-Action Capabilities of VLM Agents in 3D Scenes
Paper • 2605.31148 • Published • 3 -
On the Scaling of PEFT: Towards Million Personal Models of Trillion Parameters
Paper • 2606.02437 • Published • 241 -
PhysBrain 1.0 Technical Report
Paper • 2605.15298 • Published • 145 -
Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments
Paper • 2605.30280 • Published • 146 -
LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding
Paper • 2605.27365 • Published • 146 -
RLDX-1 Technical Report
Paper • 2605.03269 • Published • 128 -
Robots Need More than VLA and World Models
Paper • 2606.06556 • Published • 31 -
World Pilot: Steering Vision-Language-Action Models with World-Action Priors
Paper • 2606.12403 • Published • 27 -
Reason, Then Re-reason: Cross-view Revisiting Improves Spatial Reasoning
Paper • 2606.11683 • Published • 30 -
On Subquadratic Architectures: From Applications to Principles
Paper • 2606.12364 • Published • 24 -
Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models
Paper • 2606.11324 • Published • 172 -
World Model Self-Distillation: Training World Models to Solve General Tasks
Paper • 2606.12072 • Published • 15 -
MiniMax Sparse Attention
Paper • 2606.13392 • Published • 154 -
WEAVER, Better, Faster, Longer: An Effective World Model for Robotic Manipulation
Paper • 2606.13672 • Published • 4 -
Hy-Embodied-0.5-VLA: From Vision-Language-Action Models to a Real-World Robot Learning Stack
Paper • 2606.14409 • Published • 15 -
μ_0: A Scalable 3D Interaction-Trace World Model
Paper • 2606.13769 • Published • 11 -
World Tracing: Generative Pixel-Aligned Geometry Beyond the Visible
Paper • 2606.13652 • Published • 16 -
World Value Models for Robotic Manipulation
Paper • 2606.24742 • Published • 7 -
EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies
Paper • 2606.20092 • Published • 6 -
From Foundation to Application: Improving VLA Models in Practice
Paper • 2607.06403 • Published • 20 -
GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation
Paper • 2607.02642 • Published • 40 -
OpenCoF: Learning to Reason Through Video Generation
Paper • 2607.08763 • Published • 29 -
Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routing
Paper • 2607.07953 • Published • 16 -
RoboDojo: A Unified Sim-and-Real Benchmark for Comprehensive Evaluation of Generalist Robot Manipulation Policies
Paper • 2607.04434 • Published • 15 -
Dual Latent Memory in Vision-Language-Action Models for Robotic Manipulation
Paper • 2607.07608 • Published • 57 -
Let RGB Be the Language of Vision
Paper • 2607.12450 • Published • 14 -
GigaWorld-Policy-0.5: A Faster and Stronger WAM Empowered by AutoResearch
Paper • 2607.13960 • Published • 28 -
SPEAR: A Simulator for Photorealistic Embodied AI Research
Paper • 2607.06701 • Published • 8 -
Loop the Loopies!
Paper • 2607.16051 • Published • 77 -
Xiaomi-Robotics-1: Scaling Vision-Language-Action Models with over 100K Hours of Real-World Trajectories
Paper • 2607.15330 • Published • 73 -
On-Policy Delta Distillation
Paper • 2607.15161 • Published • 38 -
Understanding Reasoning from Pretraining to Post-Training
Paper • 2607.16097 • Published • 29 -
See like a Robot: Robot-Centric Pointmaps for Vision-Language-Action Models
Paper • 2607.11498 • Published • 7 -
RynnBrain 1.1: Towards More Capable and Generalizable Embodied Foundation Model
Paper • 2607.17977 • Published • 198 -
Group Entropy-Controlled Policy Optimization
Paper • 2607.16850 • Published • 29 -
Open-AoE: An Open Egocentric Manipulation Dataset and Toolchain for Embodied Learning
Paper • 2607.14183 • Published • 68 -
Distilled Reinforcement Learning for LLM Post-training
Paper • 2607.17247 • Published • 9 -
JoyNexus: Service-Oriented Multi-Tenant Post-Training for VLA Models
Paper • 2607.16074 • Published • 8 -
ABot-World-0: Infinite Interactive World Rollout on a Single Desktop GPU
Paper • 2607.19191 • Published • 309 -
Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing
Paper • 2607.19064 • Published • 77 -
AlayaWorld: Interactive Long-Horizon World Modeling -- Full Technical Report
Paper • 2607.18367 • Published • 60 -
H^2SD: Hybrid Hindsight Self-Distillation
Paper • 2607.18955 • Published • 7 -
Appearance Pointers -- Multimodal Region Control of Diffusion Transformers
Paper • 2607.19344 • Published • 4 -
Masked Visual Actions for Unified World Modeling
Paper • 2607.19343 • Published • 9 -
SeededGrasp: Language-Guided Grasping in Complex Scenes with Multiple Embodiments
Paper • 2607.20207 • Published • 4 -
Self-Supervised Learning of Structured Dynamics from Videos
Paper • 2607.21576 • Published • 19 -
Visual Contrastive Self-Distillation
Paper • 2607.21556 • Published • 52 -
SANA-Video 2.0: Hybrid Linear Attention with Attention Residuals for Efficient Video Generation
Paper • 2607.21553 • Published • 39 -
TableVerse: A Large-scale Tabletop Dataset with Real-world Grounded Layouts for Generalizable Manipulation
Paper • 2607.21017 • Published • 7 -
Molt: A Scalable PyTorch-Native Training Framework for Agentic Reinforcement Learning
Paper • 2607.21653 • Published • 32 -
Scaling Native Multimodal Pre-Training From Scratch
Paper • 2607.22043 • Published • 28 -
Kimi K3: Open Frontier Intelligence
Paper • 2607.24653 • Published • 486 -
Progress Reward Modeling for Robotic Learning: A Comprehensive Survey
Paper • 2607.21655 • Published • 193 -
Data Pyramid for Embodied Manipulation
Paper • 2607.24744 • Published • 36 -
UltraViT: Latency-Optimized On-device Vision Encoder for Large Vision-Language Models
Paper • 2607.23373 • Published • 7 -
TurboVLA: Real-Time Vision-Language-Action Model at 32 Hz on an RTX 4090 with <1 GB VRAM
Paper • 2607.27205 • Published • 139 -
StatePlay: State-Aware Game World Models for Mechanics-Consistent Generation
Paper • 2607.26754 • Published • 18 -
PhiZero: A World Model Built Around Physical Language
Paper • 2607.28624 • Published • 167 -
Multi-Head Attention Residuals
Paper • 2607.27230 • Published • 11 -
Chimera: Designing and Chinchilla-Scaling Hybrid Visual Diffusion Transformers
Paper • 2607.28611 • Published • 21 -
Mental World Modeling
Paper • 2607.27201 • Published • 103 -
RL^2-VLA: Adaptive RL Latent Compositional Steering with Test-Time Scaling for Vision-Language-Action Models
Paper • 2607.26991 • Published • 8 -
ODEWorld: A Continuous Predictive Architecture via Physical-Time Flow
Paper • 2607.27924 • Published • 14 -
N_0-TWAM: Scaling Tactile-Native World-Action Model for Contact-Rich Manipulation
Paper • 2607.23783 • Published • 47 -
N_0-VTLA: Scaling Vision-Tactile-Language-Action Model with Latent Tactile Tokens
Paper • 2607.23782 • Published • 76 -
DiffusionGemma Technical Report
Paper • 2608.00146 • Published • 33 -
Towards Physics of Multimodal Pretraining: Knowledge Flow, Modality Synergy, Early Unification, and Recipes
Paper • 2608.05000 • Published • 52 -
Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data
Paper • 2608.02580 • Published • 22 -
BridgeVLA++: A Data-Efficient, Generalizable, and Memory-Augmented Vision-Language-Action Framework for 3D Manipulation
Paper • 2608.05042 • Published • 7