Dynamic collision avoidance scheme for unmanned surface vehicles under complex shallow sea Environments.

In this paper, collision avoidance (CA) scheme for unmanned surface vehicles (USVs) under complex shallow sea environments which contain static sea surface (Static), undersea obstacles (UO) and surface island with reefs (IR) is proposed. The main challenge of this paper is that CA objects not only include surface obstacles, UO and IR, but more importantly, they will be affected by dynamic water level. It is difficult, if not impossible, to employ conventional CA schemes for USVs. The salient features of the proposed scheme are: 1) A UO model which incorporates time-varying water level is developed. 2) The smoothing edge (SE) method of island of IR that is based on fuzzy Gauss mixture model is enhanced by introducing fuzzy membership to Gaussian mixture model. 3) A real-time collision probability (RTCP) algorithm under IR with time-varying water level which obeys the Poisson and Gauss distribution is developed. By combining the UO-FMM and the SE-RTCP, the proposed CA scheme is able to handle uncertainties in complex shallow sea environments. Comparative studies with the state of the art demonstrate that the proposed CA scheme is superior in terms of safety and yaw cost consumption. • An underwater obstacle (UO) model which incorporates time-varying water level is developed. • The smoothing edge (SE) method is enhanced by introducing fuzzy membership to Gaussian mixture model. • A real-time collision probability (RTCP) algorithm which obeys the Poisson and Gaussian distribution is developed. [ABSTRACT FROM AUTHOR]