Navigation & State Estimation — Learning Plan
Study order, prerequisites, and checkpoints for estimator theory and diagnosis.
01 — Dead-Reckoning & Wheel Odometry
Derive differential-drive motion updates directly from encoder measurements.
02 — The Kalman Filter & EKF
Build intuition for prediction, covariance propagation, and nonlinear updates.
03 — Measurement Models & Mahalanobis Gating
Model sensors, innovations, and gating logic for robust estimator updates.
04 — IMU Fusion & Gyroscope Integration
Fuse gyro information, bias handling, and orientation constraints coherently.
05 — Failure Mode Diagnosis
Map covariance behavior and log symptoms to likely estimator faults.
06 — Ceres Bell-Curve Solver: Internals, Failure Modes & Diagnosis
Inspect the line-sensor optimizer internals and diagnose edge-case failures.
Exercise 01 — Dead-Reckoning & Wheel Odometry
Work through encoder math and pose propagation by hand.
Exercise 02 — Kalman Filter: 1D Problems
Compute 1D Kalman predict and update steps with explicit numbers.
Exercise 03 — Measurement Models & Mahalanobis Gating
Work through line constraints, innovations, and measurement rejection by hand.
Exercise 04 — IMU Fusion & Gyroscope Integration
Reason about gyro bias, theta fusion, and when IMU corrections should be trusted.
Exercise 05 — Failure Mode Diagnosis
Read estimator symptoms like RCA cases and separate slip, delocalization, and collision paths.