We introduce uncertainty-aware object instance segmentation (UncOS) and demonstrate its usefulness for embodied interactive segmentation. To deal with uncertainty in robot perception, we propose a method for generating a hypothesis distribution of object segmentation. We obtain a set of region-factored segmentation hypotheses together with confidence estimates by making multiple queries of large pre-trained models. This process can produce segmentation results that achieve state-of-the-art performance on unseen object segmentation problems. The output can also serve as input to a belief-driven process for selecting robot actions to perturb the scene to reduce ambiguity. We demonstrate the effectiveness of this method in real-robot experiments. Website: https://sites.google.com/view/embodied-uncertain-seg
EmbodiedOcc++: Boosting Embodied 3D Occupancy Prediction with Plane Regularization and Uncertainty Sampler
State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University（多媒体信息处理国家重点实验室，计算机学院，北京大学）；School of electronic science and engineering, Nanjing University（电子科学与工程学院，南京大学）；Berkeley Artificial Intelligence Research Lab, Department of EECS, University of California, Berkeley（伯克利人工智能研究实验室，电子工程与计算机科学系，加州大学伯克利分校）
UniPart: Towards Zero-shot Language-Grounded 3D Part Segmentation for Embodied Interaction
NLPR, Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所模式识别国家重点实验室）；Peking University（北京大学）；Tsinghua University（清华大学）；Galbot Inc.（北京银河通用机器人有限公司）；Shanghai Jiao Tong University（上海交通大学）；Beijing Academy of Artificial Intelligence（北京智源人工智能研究院）
Segment-to-Act: Label-Noise-Robust Action-Prompted Video Segmentation Towards Embodied Intelligence
School of Artificial Intelligence and Robotics and the National Engineering Research Center of Robot Visual Perception and Control Technology, Hunan University, China（人工智能与机器人学院及机器人视觉感知与控制技术国家工程研究中心，湖南大学，中国）；Institute for Anthropomatics and Robotics, Karlsruhe Institute of Technology, Germany（人机学与机器人研究所，卡尔斯鲁厄理工学院，德国）
Evaluating Uncertainty Estimation Methods on 3D Semantic Segmentation of Point Clouds
Hochschule Bonn Rhein Sieg（波恩莱茵锡根应用技术大学）
深度学习模型广泛用于各种安全关键应用。因此，这些模型除了要准确外，还需要高度可靠。实现这一目标的一种方法是量化不确定性。UQ的贝叶斯方法已被广泛研究用于图像上应用的深度学习模型，但对于3D模态（如常用于机器人和自治系统的点云）的研究较少。在这项工作中，我们评估了三种不确定性量化方法，即深度集成，MC-Dropout和MC-DropConnect在DarkNet 21 Seg 3D语义分割模型上，并全面分析了各种参数的影响，如集成或正向传递中的模型数量，以及丢弃概率值，对任务性能和不确定性估计质量的影响。我们发现，深度集成在性能和不确定性指标上都优于其他方法。深度集成在mIOU方面优于其他方法2.4%，在准确性方面优于其他方法1.3%，同时为决策提供可靠的不确定性。
Underwater Embodied Intelligence for Autonomous Robots: A Constraint-Coupled Perspective on Planning, Control, and Deployment
Department of Mechanical Engineering, The University of Hong Kong（香港大学机械工程系）；Tsinghua Shenzhen International Graduate School, Tsinghua University（清华大学深圳国际研究生学院）；Department of Data and Systems Engineering, The University of Hong Kong（香港大学数据与系统工程系）；Department of Data Science, New Jersey Institute of Technology（新泽西理工学院数据科学系）
