Dear Eman Abdullah,

I suggest you to see links on topic.

https://www.academia.edu/15933495/BENCHMARKING_ROBOT_PATH_PLANNING

https://eric-heiden.com/publication/2021-benchmr-ral-icra/2021-benchmr-ral-icra.pdf

https://pdfs.semanticscholar.org/d62d/2f16cd0c4d85cb6c6fc220c08ee4221b7196.pdf

Article Hierarchical Path-Planning for Mobile Robots Using a Skeleto...

Article BENCHMARKING ROBOT PATH PLANNING

Article Benchmarks for Grid-Based Pathfinding

Best regards

Dear Eman Abdullah

These codes and resources might help, have a look:

1. https://github.com/AtsushiSakai/PythonRobotics

2. https://github.com/perfectly-balanced/PathBench

3. https://github.com/NKU-MobFly-Robotics/local-planning-benchmark

4. https://github.com/robot-motion/bench-mr

5. https://github.com/AtsushiSakai/PythonRobotics/blob/master/docs/modules/path_planning.rst

6. https://github.com/neka-nat/cupoch

7. https://github.com/RuslanAgishev/motion_planning

Kind Regards

Qamar Ul Islam

I can provide more information later: path planning algorithms fall into the category of np hard problems (and related to traveling salesman problems) it is possible to determine the statistical properties of a one algorithm (measured by the distribution of path length) and to determine the statistical properties of other algorithms using the same statistical properties and thereby determine which algorithm (on average) has the best result and the likelihood that one algorithm will be able to find the best path when compared to another, with a fixed limit on number of trials. Then if some other algorithm is developed the same statistical analysis can be used for comparison. you might find some of my research gate publications to be help, perhaps "benchmarking combnatorial problems and algorithms for their solution" Without performing statistical analysis the comparative results when testing two algorithms may be unreliable due to the statistical properties of the two algorithms. In other words one may have the better outcome solely based on chance. . . . . Let me know if you would like to discuss. You can use almost any path planning map to perform initial analysis and comparison and then later repeat comparisons using maps that have specific features or challenges.

Thank you all for you help.

Mr. Barry Fox , do you mean that providing benchmarking grid maps for path planning problem is not useful in deciding which algorithm is better than the other?

Please correct me if I misunderstood.

Actually, I generated 2d grid maps with randomly assigned obstacles, start, and goal. Then, I will do statistical analysis to compare the algorithms performance.

However, my supervisor wants a benchmark grid maps to test the performance of the algorithms, and they must be from reliable source ( ISI papers). But, I am not sure if such benchmarks exist and reliable.

You mentioned that "You can use almost any path planning map to perform initial analysis and comparison and then later repeat comparisons using maps that have specific features or challenges." This what I did; initial comparison with random maps, now I need the maps that " have specific features or challenges", but I don't know where can I find them.

Please I need resources that discuss this issue because I need to include it in my MA thesis.

I apologize I did not intend to confuse you. Yes, you should use known problems from reliable sources. However, Individual problems may be simple for some algorithms and difficult for others. So it is important to use the same problems for all algorithms. I will send you some examples of how each algorithm may perform differently for each problem.