In the transiting stage of an unmanned undersea vehicle (UUV) mission, it is of interest to minimize platform localization error with minimal processing. Earlier work [1] derived a simultaneous localization and map building (SLAM) -inspired estimator of platform location and velocity, dubbed "velocity -over -ground" (VOG)-SLAM, that provides virtually identical performance in transit scenarios as conventional SLAM. The method lends itself to simple real-time operation as map building is not required. The "VOG" simplification was devised based on (a) the observation that the second measurement of a persistent contact was required for potential performance improvement in SLAM and (b) the intuitive idea that SLAM is providing velocity information since contact measurements can only be relative to the platform. We provide here a direct argument by arguing its optimality properties via connection to the maximum likelihood estimator (MLE). In addition, techniques for sonar data processing, measurement generation and data association methodologies to determine proper assignments between measurements and persistent bottom features are discussed. These further extend concepts found in [2]. The process can be currently completed before the next ping arrives suggesting near real-time SLAM performance in complex undersea environments

Chapter Contents:

• 7.1 Introduction
• 7.2 Development of MLE and relation to Kalman update
• 7.2.1 MLE development
• 7.2.2 VOG-SLAM Kalman update
• 7.3 Sonar data processing and measurement generation
• 7.3.1 Detector
• 7.3.2 Connected component clustering and labeling
• 7.3.3 Mapping detections to UTM frame
• 7.3.3.1 u⃗ → u⃗β
• 7.3.3.2 u⃗β → u⃗NED
• 7.3.3.3 u⃗NED → u⃗UTM
• 7.3.3.4 u⃗UTM → v⃗
• 7.3.4 UTM uncertainty region associated with each cluster
• 7.3.5 SLAM measurement and error covariance generation
• 7.4 Data association to determine persistent features
• 7.5 Performance example
• 7.6 Conclusions and future work
• References

Inspec keywords: data compression; SLAM (robots); autonomous underwater vehicles; sensor fusion

Other keywords: MLE; maximum likelihood estimator; VOG-SLAM; measurement generation; UUV; platform localization error; velocity information; unmanned undersea vehicle; reduced complexity SLAM techniques; data compression; data association; transiting stage; velocity-over-ground SLAM; sonar data processing

Subjects: Mobile robots; Transducers and sensing devices; Marine system control; Telerobotics