ENCP: Episode-Normalized Conformal Prediction for Vision-and-Language Navigation
This research provides a model-agnostic framework for improving safety in VLN agents by better estimating uncertainty during navigation.
This research provides a model-agnostic framework for improving safety in VLN agents by better estimating uncertainty during navigation.
MIT团队发布TANGO模型,让Unitree G1人形机器人能零样本执行语言指令在杂乱环境中导航,无需真实世界训练数据。
做航空视觉语言导航的看这个:DreamFly在OpenFly上SR和SPL都超过现有方法,导航误差最低,还能判断何时到达。
BioVLN专门为实验室机器人导航设计,把仪器分成三个区域来评估,比普通方法更安全,成功率能上80%+。
CoFL-S用预测流场代替离散动作,在Habitat基准上比动作token和动作块方法都强,还能直接跑真实机器人,做语言导航的可以看看。