Kalman Filter Learning Path¶
A path from estimation intuition to tested numerical kernels and a timestamped camera-fusion experiment. Advanced lessons identify the additional work needed for full inertial or visual odometry systems.
Learning Path¶
Level 1: First intuition¶
- 01 - What is Estimation? - Understanding guessing and improving guesses
- 02 - Noisy Measurements - Why sensors aren't perfect
- 03 - Combining Information - Using multiple clues
Level 2: Arithmetic and uncertainty¶
- 04 - Averages and Uncertainty - Mean and variance
- 05 - Weighted Averages - Trusting some measurements more
- 06 - Prediction - Guessing the future from the past
Level 3: Algebra and code¶
- 07 - The Kalman Filter Idea - Putting it all together
- 08 - One Dimensional Kalman Filter - First real implementation
- 09 - Tracking a Moving Object - Position and velocity
Level 4: Undergraduate (College)¶
- 10 - Matrix Form - Mathematical notation
- 11 - Multidimensional Kalman Filter - Multiple states
- 12 — Process noise and measurement noise
- 13 — A complete Kalman update, with numbers
Level 5: Graduate (Advanced)¶
- 14 — Extended Kalman filter: linearize the right function
- 15 — Unscented Kalman filter: propagate a set of representative points
- 16 — Error-state filtering: small corrections around a nominal pose
- 17 — MSCKF: use a feature track without keeping every landmark
Level 6: Numerical methods and integration¶
- 18 — Kalman smoothing: use later information to revise an earlier estimate
- 19 — Outlier rejection: test a residual against its uncertainty
- 20 — Numerical implementation: preserve the meaning of covariance
- 21 — Integrate an estimator into a robot observation pipeline
How to Use This Guide¶
- Start at your level - Don't skip ahead too fast
- Do the exercises - Each chapter has hands-on problems
- Code along - Implementation examples in Python
- Build intuition first - Understand "why" before "how"
Prerequisites by Level¶
- Level 1-2: Basic arithmetic
- Level 3: Algebra, basic probability
- Level 4: Linear algebra, probability theory
- Level 5: Calculus, statistics
- Level 6: Advanced mathematics, programming
Quick Reference¶
Once you've learned the basics, use these for quick lookup: - Kalman Filter Equations Cheat Sheet - Common mistakes and diagnostic checks - Executable implementation templates
Let's begin the journey!
Primary theory reference: Särkkä, Bayesian Filtering and Smoothing (author-provided 2013 edition), chapters 4–5 and 8–9. New examples in this course are original teaching fixtures. Source/API checks and executable tests: 2026-10-18.