Comparison of GBNN Path Planning with Different Map Partitioning Approaches

Autonomous underwater vehicles (AUVs) usually work without any commands from remote control and finish tasks on their own. Path planning is necessary and very important. The Glasius bioinspired neural network (GBNN) model has already been applied to path planning of AUVs. Generally, the model divides the environment space into many rectangular grids. However, the triangular and hexagonal grids can also be adopted to divide the environmental space. To compare among the three different map partitions, analysis and simulations are performed. The path length, steering and speed varies are compared while the receptive range of AUV keeps the same. Analysis shows that the triangular grid may be the worst and the hexagonal grid may be the best. However, in the simulations, there is a path oscillation problem in the hexagonal partition, which makes the path longer and zigzag. On the other hand, the triangular gird also has some advantages, such as minimum turning angle, while the rectangular grid can make a balance.