A high-altitude archive for robotics builds, world-model notes, mountain routes, and the practical details that become useful when the weather changes.
Base entry paths
Choose a route through the station: research shelves for papers, field routes for mountain notes, or the logbook when you want the whole trail.
Best first shelf
Begin with the densest research shelf, then follow the active reading routes into notes that connect papers, builds, and field work.
Robotics experiments, build notes, and implementation details kept in plain reading order.
Open selected notesSnow, routes, elevation, and field observations from the part of life that happens outside the lab.
Follow mountain notesEmbodied systems, hardware stacks, and the debugging notes that make them repeatable.
Open robotics pathThe full archive stays intact, but it now sits behind the person and projects that give it shape.
Scan the archiveHosted papers across robotics, world models, JEPA, and embodied AI with searchable notes.
Browse libraryRecommended reading
Curated approach
A guided path through the strongest material: source shelf, lead note, then the route with the most signal.
Research shelf
58 papers form the densest source layer before the notes branch outward.
Open paper indexStation pick
Reading Notes: World Models + JEPA Read date: 2026 06 01 Scope: 14 downloaded papers in this folder. I read the abstracts, introductions, method sections, result/conclusion section...
Read World Models + JEPA Reading NotesStrongest route
5 notes currently define the strongest active path.
Enter 学习 routeNewest note
SLAM 中的 Pose Graph Optimization (PGO) 算法详解 SLAM 中的 Pose Graph Optimization (PGO) 算法详解 从直觉到数学，从公式到动画 —— 一份可交互的图优化教程 目录 为什么需要 PGO？ 位姿图的数学结构 误差函数与信息矩阵 线性化与雅可比 ⭐ 完整推导：从 F(x) 到 HΔx=-b ⭐...
Latest by publication time, kept separate from pinned notes so the newest dispatch stays easy to find.
Field notes and research
Topic paths
The strongest active path is surfaced first so returning readers can enter the densest route without scanning the whole archive.
Selected reading
Curated lead
Reading Notes: World Models + JEPA Read date: 2026 06 01 Scope: 14 downloaded papers in this folder. I read the abstracts, introductions, method sections, result/conclusion section...
Start with the most intentional note, then branch into the route that gives it context.
Supporting signals
2 morePinned note
Course 1：神经网络与深度学习 学习目标 ：从零开始理解神经网络的核心原理，掌握前向传播与反向传播的数学推导，能够用纯 NumPy 搭建一个任意深度的神经网络。 目录 第一章：深度学习引言 第一章深度学习引言 第二章：神经网络编程基础——逻辑回归 第二章神经网络编程基础逻辑回归 第三章：浅层神经网络 第三章浅层神经网络 第四章：深层神经网络 第四章深层...
Pinned note
第一次搭建个人博客 关于我 我是 Badger，一个对 AI、机器人和户外探索充满热情的工程师。平时主要研究 Pytorch、Mujoco、URDF、ROS 这些机器人相关的技术栈，同时也在探索 AI 工程化的各种可能性。 搭建这个博客的初衷很简单——我需要一个属于自己的地方，记录学习过程中的思考和踩过的坑。 为什么要自己搭博客 用过 CSDN、知乎，也试过...
Primary Archive
Start with the strongest active route, then branch into the newest signal when you want a current entry point.
