FixTheRobot Explore Robotics / Systems Engineering / Tools

D · Interactive tool · Module 11

Sensor Fusion Architecture Builder

Select the sensors your robot has; see the fusion structures that fit, how information flows, what must be synchronised and calibrated, and how each can fail.

Preliminary engineering guidance. Results depend on the assumptions listed below and on simplified models. Verify with manufacturer datasheets and a test on your own hardware before committing to a design.

Sensors you have

Assumptions and limitations

  • The architectures are standard structures (local EKF, global EKF, two-stage EKF, tightly or loosely coupled LiDAR-inertial and visual-inertial odometry, radar ego-velocity, parallel estimators with a supervisor). The list is not exhaustive and the order only reflects how many of your sensors each one uses.
  • Named software is mentioned only for facts confirmed from official sources: the FAST-LIO, LIO-SAM, VINS-Fusion and KISS-ICP READMEs (fetched 2026-11-04) and the robot_localization parameter file in its repository. Everything else is general architecture knowledge.
  • “Loosely coupled” means one estimator’s output is fed to another as a measurement; “tightly coupled” means raw measurements are fused in one estimator. Neither is universally better: tight coupling is usually more accurate and more sensitive to calibration and timing errors.
  • Not every SLAM or odometry system uses a Kalman filter. FAST-LIO uses an iterated extended Kalman filter (per its README); LIO-SAM uses smoothing and mapping on a factor graph (per its README).
  • No performance figures are given; the diagrams show information flow, not timing. Module 11 explains EKF configuration and tuning; Module 12 shows a complete outdoor-to-indoor design.

References