SP-VLA | arxiv 2025.10.03 | Paper Reading
SP-VLA: A JOINT MODEL SCHEDULING AND TOKEN PRUNING APPROACH FOR VLA MODEL ACCELERATION
Read more
Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success
OpenVLA | CoRL 2024 | Paper Reading
OpenVLA: An Open-Source Vision-Language-Action Model
这篇文章基于Prismatic-7B VLM提出了一个统一的VLA模型架构。
FiSVLA | arxiv 2025.6.02 | Paper Reading
Fast-in-Slow: A Dual-System Foundation Model Unifying Fast Manipulation within Slow Reasoning
这篇文章将双系统架构中的系统1嵌入到系统2中从而得到更好的结果。
CoA-TAM | NIPS 2025 | Paper Reading
Chain-of-Action: Trajectory Autoregressive Modeling for Robotic Manipulation
这篇文章的核心思想是将原先的顺序轨迹改为逆向的预测。
BridgeVLA | NIPS 2025 | Paper Reading
BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language Models
这篇文章基于OpenVLA将嵌入空间投影到共享空间从而进行更好学习。
SimpleVLA | arxiv 2025.9.11 | Paper Reading
SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning
这篇文章通过对GRPO改进为DAPO之后引入VLA达到很好的性能。
VLA-Adapter | AAAI 2026(oral) | Paper Reading
VLA-ADAPTER: AN EFFECTIVE PARADIGM FOR TINY-SCALE VISION-LANGUAGE-ACTION MODEL
这篇文章通过设计了一个bridge Attenion block以及广泛的实验尝试验证了小模型也能够有很好的性能。
RDP | RSS 2025 | Paper Reading
Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation
这篇文章通过软件与硬件两套设计实现了模型面对需要精细操作任务时的快速相应能力，其中的motivation很有意义。
RoboTwin 2.0 | arxiv 2025.8.27 | Paper Reading
RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation
