A multi-stage visual odometry system and failure detection mechanism for a monocular visual SLAM implementation in monotonously textured underwater scenes

EMANUEL TRABES; MARIO A. JORDÁN

Cordoba

Workshop; 2015 XVI Workshop on Information Processing and Control; 2015

A multi stage visual odometry system with a failuredetection mechanism is developed to measure movement andcorrect unreliable estimations in poorly textured underwaterenvironments. The system has only one camera as the sensor,and is composed of a primary movement estimation based offeatures presented in the scene and two backup system. Thefirst backup is based on phase correlation to estimate the AUVmovement. The second backup movement estimation mechanismis based on a high order prediction of the movement using thelast good movement estimations. Failure detection mechanismare constructed to detect unreliable measurements in each of thestages and appropriate responses are developed in all cases. Afull SLAM implementation is developed and a real case study ispresented in which the good properties of the system are shown