MANIPULATION
FineVLA：面向可操控视觉-语言-动作策略的细粒度指令对齐
Xintong Hu, Xuhong Huang, Jinyu Zhang, Yutong Yao, Yuchong Sun, Qiuyue Wang, Mingsheng Li, Sicheng Xie, Yitao Liu, Junhao Chen, Yixuan Chen, Yingming Zheng, Shuai Bai, Tao Yu
- 发表年份
- 2026
- 访问权限
- 开放获取
摘要
本文提出FineVLA框架，通过构建包含47,159条细粒度轨迹的数据集和专用VLM标注器，实现机器人动作执行中细粒度指令（如手臂朝向、接触区域）与目标级指令的混合训练。实验表明细粒度监督不牺牲任务成功率，且与原始指令互补，在仿真和真实双臂操作中均取得显著提升。
关键词
Vision-Language-Actionfine-grained instructionsteerable policyrobot manipulationdata annotation
相关论文
MANIPULATION
📊 7,533 引用
Real-Time Obstacle Avoidance for Manipulators and Mobile Robots
Oussama Khatib
1986
MANIPULATION
📊 6,720 引用
A Mathematical Introduction to Robotic Manipulation
Richard M. Murray, Zexiang Li, Shankar Sastry
2017
MANIPULATION
📊 3,821 引用
Robot dynamics and control
Mark W. Spong
1989
MANIPULATION
📊 3,499 引用
A tutorial on visual servo control
Seth Hutchinson, Gregory D. Hager, Peter Corke
1996