Testing performance of my global planner

I implemented my own global planner algorithm as a c++ program. To measure its performance, I implemented A* in c++ also.

When I compare them, I found that my algorithm is faster than A* and generates the same path length as A*.

I wrote my algorithm as ROS global planner plugin, and I can move a robot using it.

However, in ROS, when I compare my planner and navfn (A* based) I found that my algorithm is slower than navfn and produces longer paths.

why there is a difference between A* c++ implementation and navfn (A* version)?

Thanks,

Asked by Eman.m on 2021-08-23 11:21:04 UTC

Comments

Answers

As far as I know, NavFn uses dijkstra algorithm, not A*. This old discussion #q28366 has some interesting tidbits.

Asked by Mike Scheutzow on 2021-08-24 19:32:29 UTC

Comments

maybe simply because your global planner implementation and navfn's implementation have a slight difference in any way. A single variable declaration will really affect the run time or the results of the code. And we can't expect your code to be EXACTLY the same with navfn implementation.

Asked by mgrallos on 2023-05-04 02:01:50 UTC