No description available.

Simultaneous localisation and mapping (SLAM) is an often used positioning approach in GPS denied indoor environments. This thesis presents a novel method of combining SLAM with autonomous/aided target detection and recognition (ATD/R), which is beneficial for both methods. The method uses physical objects that are recognisable by ATR as unambiguous features in SLAM, while SLAM provides the ATR with better position estimates. The intended application is to improve the positioning of a first responder moving through an indoor environment, where the map offers localisation and simultaneously helps locate people, furniture and potentially dangerous objects like gas cannisters. The developed algorithm, dubbed ATR-SLAM, uses existing methods from different fields such as EKF-SLAM and ATR based on rectangle estimation. Landmarks in the form of 3D point features based on NARF are used in conjunction with identified objects and 3D object models are used to replace landmarks when the same information is used. This leads to a more compact map representation with fewer landmarks, which partly compensates for the introduced cost of the ATR. Experiments performed using point clouds generated from a time-of-flight laser scanner show that ATR-SLAM produces more consistent maps and more robust loop closures than EKF-SLAM using only NARF landmarks.

No description available.