知识卡片
行动需要超越物体命名
机器人抓取与车辆规划还需要三维位置、形状、支撑关系、动力学及不确定性。类别和分割提供部分信息,却未完成行动所需建模;系统应根据任务提出可检验假设,必要时通过观察或干预补充证据,而不是只扩充类别词表。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 目标识别
原书出处 · 1 处
- · 查看原书依据
However, the representations we have described in this chapter feel very incomplete. If you are building a robot or an autonomous vehicle, such object representations are very hard to act on. A robot planning to grasp an object will need an accurate three-dimensional (3D) modeling of the object as well as an understanding of some of its physical properties in order to plan for the
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第50章《目标识别》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c050b.xhtml)
- 结论依据:摘录中关于行动需要超越物体命名的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:However, the representations we have described in this chapter feel very incomplete. If you are building a robot or an autonomous vehicle, such object representations are very hard to act on. A robot planning to grasp an object will need an accurate three-dimensional (3D) modeling of the object as well as an understanding of some of its physical properties in order to plan for the