This lab works on phones, but writing code is much easier on a laptop.
Localization & SLAM · LiDAR SLAM · M5-L2 · Advanced
The lidar odometry works in the lab and slides in the corridor. Its pose jumps forward and back along the hallway while the walls look perfect in the map.
You find out why, build the check that would have warned the system, and then explore what else makes registration fail.
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 degeneracy analysis of M5-T4 into code: assemble the normal matrix of a point-to-line problem, read its smallest translation eigenvalue, and refuse to invent motion in a direction the scan cannot measure. It is graded on a room and a corridor whose true correction stays in the simulator. Stage 2 is an experiment on the two other classic causes of bad registration: the cost function and the starting error.
A registration that always returns an answer is dangerous. The useful ones also return how much to trust it.
Educational simplification: straight, perfectly known walls, one linearised step with rotation about the origin, 2D instead of 6-DOF, and a fixed threshold on the eigenvalue. Production systems analyse the 6×6 matrix, scale the directions and combine the result with the IMU.
Get notified when new robotics labs are released. One email per new lab, nothing else.