Page 345 - Autonomous Mobile Robots

9 Map Building and

SLAM Algorithms

José A. Castellanos, José Neira, and
Juan D. Tardós

CONTENTS

9.1 Introduction ............................................................ 335
9.2 SLAM Using the Extended Kalman Filter ........................... 339
9.2.1 Initialization ................................................... 340
9.2.2 Vehicle Motion: The EKF Prediction Step................... 341
9.2.3 Data Association .............................................. 342
9.2.4 Map Update: The EKF Estimation Step...................... 344
9.2.5 Adding Newly Observed Features............................ 344
9.2.6 Consistency of EKF–SLAM .................................. 345
9.3 Data Association in SLAM............................................ 346
9.3.1 Individual Compatibility Nearest Neighbor.................. 346
9.3.2 Joint Compatibility ............................................ 347
9.3.3 Relocation ..................................................... 350
9.3.4 Locality ........................................................ 354
9.4 Mapping Large Environments......................................... 358
9.4.1 Building Independent Local Maps ........................... 359
9.4.2 Local Map Joining ............................................ 359
9.4.3 Matching and Fusion after Map Joining ..................... 361
9.4.4 Closing a Large Loop ......................................... 361
9.4.5 Multi-robot SLAM ............................................ 365
9.5 Conclusions ............................................................ 366
Appendix: Transformations in 2D ........................................... 367
Acknowledgment ............................................................. 368
References .................................................................... 368
Biographies ................................................................... 371

335

© 2006 by Taylor & Francis Group, LLC

FRANKL: “dk6033_c009” — 2006/3/31 — 16:43 — page 335 — #5