作者: Md Jahidul Islam , Junaed Sattar
DOI: 10.1109/ICRA.2017.7989516
关键词:
摘要: We present a novel algorithm for an autonomous underwater robot to visually detect and follow its companion human diver. Using both spatial-domain frequency-domain features pertaining swimming patterns, we devise the position direction of Our is unique in way that it allows detection arbitrary motion directions, addition keeping track diver's through image sequence over time. A Hidden Markov Model (HMM)-based approach prunes search-space all potential trajectories relying on intensities spatial-domain. signature subsequently detected non-overlapping subwindows exhibiting patterns. The pruning step ensures efficient computation by avoiding exponentially large search-spaces, whereas us accurately. Experimental validation proposed presented datasets collected from open-water closed-water environments.