Module 11 · Phase 4: Localization and autonomy
Sensor Fusion & State Estimation
Configure EKF-based localization with robot_localization, tune covariances, reject bad measurements and decide between filters and factor graphs. Not every SLAM system is a Kalman filter, and the guides say which are not.
This overview is open. 5 of 5 guides are written; the others show Coming soon until they are complete and tested.
Questions this module answers
- How do I configure an EKF for robot localization?
- How do I choose process noise and measurement covariance?
- How do I identify a bad measurement and handle localization jumps?
- When should I use an EKF and when a factor graph, and how do I keep localizing when a sensor fails?
What it covers
- Wheel/IMU, GNSS/IMU, camera/IMU, LiDAR/IMU, GNSS+IMU+LiDAR fusion
- EKF, UKF, error-state EKF, factor graphs
- Loosely vs tightly coupled
- Covariance tuning, outlier rejection, delayed measurements, sensor health
Guides and design decisions
A. Engineering Guides (4)
B. Design & Selection (1)
Diagnostics, tools and challenges
C. Troubleshooting (2)
- Localization covariance grows unexpectedlyReported uncertainty rises with no obvious sensor loss.
- GPS position jumpsThe reported position leaps by metres with no robot motion.
D. Interactive tools (2)
- EKF Tuning PlaygroundChange process and measurement noise and watch prediction, update, state and covariance with noisy and delayed measurements.
- Sensor Fusion Architecture BuilderPick IMU, LiDAR, camera, GNSS, wheel odometry and radar; see fusion architectures, data flow, sync needs and failure considerations.
E. Design challenges (2)
- Challenge 2: Outdoor-to-Indoor Autonomous NavigationGNSS outdoors to LiDAR-inertial or visual-inertial indoors: sensor and frame architecture, sync, calibration, estimation, handover and recovery.
- Challenge 5: Humanoid Locomotion IntegrationSimulated humanoid: model, actuators, IMU and joint sensing, state estimation, gait, whole-body control, MPC and validation.
Practical deliverables
- robot_localization examples
- State-vector examples
- Fusion architecture diagrams
- Validation procedure
Read first: existing FixTheRobot lessons and labs
This module builds on these instead of repeating them.
- EKF-SLAM SLAM · Lesson
- Filtering-Based SLAM SLAM · Lesson
- Simple Filtering and Kalman Filter Intuition Perception · Lesson
- Uncertainty and Conflicting Observations Perception · Lesson