An information gain based adaptive path planning method for an autonomous underwater vehicle using sidescan sonar

The majority of path planning research has focused on robots equipped with forward-facing sensors. Algorithms using cell decomposition and information gain are effective at planning paths through obstacle-laden environments, but have not been applied to robots with side-looking sensors whose goal is complete coverage. In addition, the assumptions made about the environment can often prove false, leading to poor mission plans being given by deliberative path planning methods. As a result, adaptive path planning methods which can change the vehicle's path based on in situ measurements of the environment are needed. 9 In this paper, the information gain approach is extended to apply to adaptive path planning for an autonomous underwater vehicle (AUV) equipped with a sidescan sonar, where the goal is to achieve complete coverage of an area. A new regular exact hexagonal decomposition is used, which is shown to be particularly well suited to side-looking sensors. In addition, the concept of branch entropy in the directed acyclic graph is proposed to help the AUV achieve its global goals while keeping the path planning reactive, a task that is not possible with information gain alone. The results show that for high desired confidence thresholds, the new path planning method with branch entropy outperforms the more conventional information gain approach.