← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9187.b · 生成模型:GPT-6

回环后的地图维护有不同策略

理解等级:中级

若表面体积已把多帧观测压成融合值,修正历史相机姿态需要能追溯或重建贡献。保留并重积分所有相关原始帧是一种代价较高方案,也可采用其他结构与近似;原书描述不应扩展成所有SLAM都必须全局重放。

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

      The problem with using ICP continuously to generate these maps is that errors in each transformation propagate into the map generation process in the form of map drift. Here, the simultaneous localization and mapping (SLAM) algorithm (chapter 17) can be used to correct previous errors once a loop closure is detected, but the update of TSDFs on the trigger of a loop closure requires both

    参考来源

    - 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》第12章《建图》(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_12.xhtml) - 结论依据:摘录中关于回环后的地图维护有不同策略的定义、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:The problem with using ICP continuously to generate these maps is that errors in each transformation propagate into the map generation process in the form of map drift. Here, the simultaneous localization and mapping (SLAM) algorithm (chapter 17) can be used to correct previous errors once a loop closure is detected, but the update of TSDFs on the trigger of a loop closure requires both