Modification of the R* Search Algorithm
Daniel John Curtis · The UNSW Canberra at ADFA Journal of Undergraduate Engineering Research · 2014
Autonomous Underwater Vehicles (AUVs) have the potential to replace or augment existing manned or remotely piloted underwater vehicles in the completion of hazardous and tedious tasks. However to do so, a robust method of navigation and collision avoidance is required before these vehicles can operate fully autonomously. This thesis determined from the state of art of path planning, that the randomized A* search algorithm (R*) developed by Anthony Stenz and Maxim Likhachev was suitable for modification to better fit the unique requirements of the AUV being developed at ADFA. The algorithm was modified to incorporate a kinodynamic successor node generator, that can be described as a combination of the motion primatives method and a heuristic dynamic model based planner method. The path planner was then tested against multiple obstacle maps to test it’s robustness and optimization. The path planner is able to find a path where one exists however requires excessive amounts of computational effort to calculate a path. This however is rectifiable with smarter programming and is worth further development as the path planner is designed with an operational vehicle in mind.