← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9191.a · 生成模型:GPT-6

目标偏置在引导与探索间取舍

理解等级:中级

适当朝目标采样可加快直接接近,同时仍需探索绕障路线。若只采目标,可能反复撞同一障碍;偏置、步长与距离度量共同影响效果,不宜把某固定百分比当所有地图上的最佳参数。

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

      A simple approach is to randomly select a point in the state space and connect it to the closest existing point in the tree. Other solutions may assign preference to nodes with few out-degrees (i.e., those without many connections) and choose points in their vicinity in order to facilitate expansion in under-explored regions of the state space. Importantly, both approaches allow one to quickly explore

    参考来源

    - 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》第13章《路径规划》(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_13.xhtml) - 结论依据:摘录中关于目标偏置在引导与探索间取舍的定义、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:A simple approach is to randomly select a point in the state space and connect it to the closest existing point in the tree. Other solutions may assign preference to nodes with few out-degrees (i.e., those without many connections) and choose points in their vicinity in order to facilitate expansion in under-explored regions of the state space. Importantly, both approaches allow one to quickly explore