How to obtain optimal path between start and goal pose using pathPlannerRRT() and plan()?

I am currently working on path planning of a vehicle for an automatic parking system.

I am currently using pathPlannerRRT() and plan() to generate a path between start and goal pose. The problem that i am facing is, with the same start and goal pose each time i re-run my program, i am getting a different path. Different results each time I re-run indicates that the path being generated is not optimal. Sometimes it is close to optimal, sometimes its very wavy and sub optimal.

How can i better control the path being generated and how can i ensure that the path being generated is optimal?

This is the part of code where i am configuring the path planner:

motion_planner = pathPlannerRRT(cstmp,'MinIterations',2000,'ConnectionDistance',5,'MinTurningRadius',4,'ConnectionMethod','Reeds-Shepp','ApproximateSearch',false);
path = plan(motion_planner,curr_pose,goal_pose);

Answer 1

[Qu Cao]

Please set the random seed at the beginning to get consistent results across different runs:

https://www.mathworks.com/help/matlab/ref/rng.html

rng(1);
motion_planner = pathPlannerRRT(cstmp,'MinIterations',2000,'ConnectionDistance',5,'MinTurningRadius',4,'ConnectionMethod','Reeds-Shepp','ApproximateSearch',false);
path = plan(motion_planner,curr_pose,goal_pose);

[Tarun Santosh]

By setting a random seed, the program is giving repeatable results but if possible could you explain how this is working? Why setting a random seed is making the results from pathPlannerRRT repeatable?

[Qu Cao]

A rapidly exploring random tree (RRT) is an algorithm designed to efficiently search nonconvex, high-dimensional spaces by randomly building a space-filling tree.

rng controls the generation of random numbers.