← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9205 · 生成模型:GPT-6

观测不必每次都让后验更集中

理解等级:中级

动作噪声常使分布扩散,可靠地标常减少歧义,但某次观测也可能暴露多个解释或推翻错误自信。后验均值未必在两个估计之间,非高斯多峰时更不能用简单“折中”直觉替代完整更新。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      We have now learned two methods to update the belief distribution of where the robot could be in the environment. First, a robot can use external landmarks to update its position. This is known as the perception update and relies on exteroception. Second, a robot can observe its internal sensors. This is an instance of an action update and relies on proprioception. The combination of

    参考来源

    - 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》第16章《定位》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)/_epub-src/OEBPS/xhtml/chapter_16.xhtml) - 结论依据:摘录中关于观测不必每次都让后验更集中的定义、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:We have now learned two methods to update the belief distribution of where the robot could be in the environment. First, a robot can use external landmarks to update its position. This is known as the perception update and relies on exteroception. Second, a robot can observe its internal sensors. This is an instance of an action update and relies on proprioception. The combination of