A CHAINED-NODES-BASED METHOD TO COMPLEMENT SLAM NAVIGATION IN SELF-SIMILAR UNDERWATER ENVIRONMENTS

MARIO A. JORDÁN; EMANUEL TRABES

Ciudad de Buenos Aires

Congreso; Congreso Argentino de Control Automatico 2016 (AADECA 16); 2016

Asociación Argentina de Control Automático

This work deals with a node-based monocamera visual system forunmanned vehicle navigation over an extensive rather feature-poor underwater environment with the goal of bridging two nonconnected, well-known SLAM-mapped regions. Its methodology comprises an exploration and revisiting phases. The startegy applies intermittently a visual-odometry-based guidance on a feature-poor stretch and a subsequent control strategy for loop closing over a, to somedegree, feature-rich terrain. A case study developed on a benchmark shows the feasibility of the method.