Feature Extraction for Autonomous Navigation

This work describes techniques using sonar sensors for environmental feature detection and identiication. By detecting common features in indoor environments and using them as landmarks , a robot can navigate reliably, recovering its pose when necessary. Results using a multiple hypothesis testing procedure for feature localization and identiication show that accurate feature information can be acquired with adequate sonar models and conngurations. In addition, a method that associates sonar conng-uration with the precision of feature extraction is discussed, as well as its utility for guiding an active sonar sensor. Future goals are to improve pose information when necessary to satisfy navigation constraints, and to use the navigational knowledge acquired to optimize the path generated by an incremental motion planner.