Each lecture opens as slides in the browser, with its demos built in. Lectures are added here as they are posted.
-
Lecture05
Uncertainty & Kalman Filter
Why a sensor reading is a distribution, how odometry's uncertainty grows as the robot drives, and how a Kalman filter fuses prediction and measurement.
More lectures will appear here as they are converted.
Every demo opens as its own short page and runs the same Python script used in class, in the browser, with no install.
One beam at a wall
A LiDAR beam aimed at a wall that does not move. The readings fill in a bell curve: the spread stays put while the mean gets sharper as 1/√n.
Odometry's uncertainty
Drive a simulated TurtleBot and watch its 2σ position ellipse grow, propagated step by step from wheel noise.
A Kalman filter on a wall
A 1D Kalman filter on the distance to a wall. Turn LiDAR off, change r and q, and see how the estimate and its ±2σ band respond.
What wrong noise values do
Three Kalman filters on one run, with the sensor or motion noise set too small, as measured, or too large. See jitter, lag, and whether the ±2σ band tells the truth.