This lab works on phones, but writing code is much easier on a laptop.
Localization & SLAM · Advanced SLAM and Research · M8-L3 · Advanced
The delivery robot localises perfectly in the open lobby and slides along the long corridor to the loading bay: the walls look the same everywhere, and the position estimate wanders by metres.
You fix the fusion so that it knows which source to believe in which direction, and you make it say so.
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 implements the information-form fusion of M8-T4: add the information of the sources, weight each estimate by it, and read degeneracy from the eigenvalues of the scan-match information matrix. It is graded on a corridor and a room whose true path stays in the simulator. Stage 2 tunes the one parameter that sets the exchange rate between the sources.
Hard switches between sensors cause jumps and hide the cause of drift. Fusing by information hands over smoothly, and the eigenvalues are the monitor that tells an operator when the pose rests on one source only.
Educational simplification: translation only with a fixed heading; independent steps; an exact information matrix for the scan match; odometry with a 5% scale error and Gaussian noise. Real systems fuse in 3D with poses on manifolds, approximate information from registration and estimate the odometry noise.
Get notified when new robotics labs are released. One email per new lab, nothing else.