This lab works on phones, but writing code is much easier on a laptop.
Localization & SLAM · Advanced SLAM and Research · M8-L2 · Advanced
Three students claim to have the best SLAM on the same path. One reports 8 cm, one 2 m, one 0.3 m, and each used a different script. You are asked to settle it.
You write the evaluation yourself, then find out which questions the metrics answer and which ranking is fair.
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 metrics of M8-T8: ATE after a closed-form alignment (with an optional scale) and the translational RPE over an interval, graded against reference values for three runs generated in different frames and scales. Stage 2 is the discussion of fairness as an experiment: metric, interval and alignment.
Most disagreements between reported SLAM results come from the evaluation protocol rather than the algorithm. Being able to reproduce the numbers is the first step to trusting them.
Educational simplification: planar poses; index-aligned samples with no time association; the alignment uses positions only; one RPE interval at a time. Real evaluation tools handle 3D, time stamps, sub-trajectory alignment and several intervals.
Get notified when new robotics labs are released. One email per new lab, nothing else.