This lab works on phones, but writing code is much easier on a laptop.
Localization & SLAM · Advanced SLAM and Research · M8-L1 · Intermediate
The map of the corridor is ready, and it has a diagonal smear across the middle where the cleaning staff walked, and a small blob where someone waited for the lift. A planner using this map would refuse to go through either.
You rebuild the map so that only the structure remains, then learn what this kind of test can and cannot remove.
You have passed every stage. Now connect it back to the theory and move on:
Plain Python with math and, for matrices, slamkit (matmul, transpose, solve, inverse, identity). The data arrive as the dictionary D. There is no NumPy, random, json or copy in the browser runtime. Your code runs locally and never leaves this page.
Stage 1 turns the log-odds argument of M8-T1 into counting: a cell that is hit repeatedly and rarely seen through is structure. It is graded on hidden rooms with walkers and a person who stops, by the number of ghost cells and the fraction of the real walls kept. Stage 2 is an experiment on the two thresholds and on the standing time, ending with the limit of the method.
Every dynamic-aware mapper rests on this evidence, however elaborate its machinery. Knowing what the evidence can and cannot separate tells you when you need semantics or change detection.
Educational simplification: known robot poses, a 2D lidar without dropouts, people as discs, a 0.2 m grid, sampled ray traversal and hit/pass counting instead of a full log-odds update. Real mappers trace rays exactly in 3D, include pose uncertainty and use visibility checks or learned masks.
Get notified when new robotics labs are released. One email per new lab, nothing else.