A time dimension-added multiple obstacles avoidance approach for unmanned surface vehicles

The capability of avoiding collisions in an environment with obstacles is a necessary function for an autonomous unmanned surface vehicle (USV). This paper reports the preliminary research results of a novel multiple obstacles avoidance approach for USVs. This method adds the time dimension in a time-varying environment containing multiple obstacles with arbitrary known motion patterns. The approach presented is a path searching-based algorithm, the three-dimensional multiple obstacle avoidance (TDMOA) algorithm, which applies a novel multi-objective optimization and interval programming method, and can plan a suboptimal collision-free path in seconds. A unique property of the approach is the compliance with existing marine collision regulations in the planning process. The approach first incorporates navigational safety and maritime operational rules to establish an obstacle risk domain model. The time dimension is then introduced to a two-dimensional plane to establish the three-dimensional risk model, and then the improved D* Lite algorithm searches a collision-free path in compliance with marine collision regulations. Several complex multi-obstacle collision scenarios are simulated in experiments to verify the validity of the new TDMOA algorithm and results indicate good real-time performance and safe navigation paths for USVs.