Feature/comparison platform #32

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open

Xuester wants to merge 32 commits into JmcRobbie:master from matthomas15:feature/comparison_platform

.gitignore

-Original file line number
+Diff line change
@@ -1 +1,3 @@
-    *.pyc
+    *.pyc
+    .vscode

nova_rover_demos/pathfinding/a_star.py

            
                      Original file line number
                      Diff line number
                      Diff line change
                  
    @@ -1,3 +1,7 @@
  
    import numpy as np

    from itertools import product

    from math import sqrt, inf

    from pathfinding.heuristic import euclidean_cost

    import sys

    import os

    sys.path.append(os.path.dirname(os.path.abspath(__file__)) + "/../")

    @@ -6,10 +10,6 @@
  
    except:

        raise

    from pathfinding.heuristic import euclidean_cost

    from math import sqrt, inf

    from itertools import product

    import numpy as np

    def reconstruct_path_to_destination(prev, end):

        """

    @@ -25,6 +25,8 @@ def reconstruct_path_to_destination(prev, end):
  
        return path

    # A* Search

    def get_successors(node, grid):

        """

        The neighbors of a cell (node) in the grid are the 8-surrounding cells.

    @@ -35,7 +37,7 @@ def get_successors(node, grid):
  
        n_rows = len(grid)

        n_cols = len(grid[0])

        for dx, dy in product([-1,0,1],[-1,0,1]):

        for dx, dy in product([-1, 0, 1], [-1, 0, 1]):

            # skip the current node itself

            if (dx == 0 and dy == 0):

                continue

    @@ -49,17 +51,20 @@ def get_successors(node, grid):
  
                # put infinite penalty on successors that would take us off the edge of the grid

                cost = inf

            successors.append( ((x, y), cost) )

            successors.append(((x, y), cost))

        return successors

    def node_with_min_fscore(open_set, f_cost): # open_set is a set (of cell) and f_cost is a dict (with cells as keys)

    # open_set is a set (of cell) and f_cost is a dict (with cells as keys)

    def node_with_min_fscore(open_set, f_cost):

        """

        Find the cell in open set with the smallest f score.

        """

        f_cost_open = dict([a for a in f_cost.items() if a[0] in open_set])

        return min(f_cost_open, key=f_cost_open.get)

    def a_star_search(grid, start, end, heuristic_cost=euclidean_cost):

        """

        Implementation of A Star over a 2D grid. Returns a list of waypoints

    @@ -108,7 +113,7 @@ def a_star_search(grid, start, end, heuristic_cost=euclidean_cost):
  
                if neighbor in closed_set:

                    continue

                curr_g_score =  g_cost[curr] + cost

                curr_g_score = g_cost[curr] + cost

                # add neighbor to newly discovered nodes

                if neighbor not in open_set:

                    open_set.add(neighbor)

    @@ -122,4 +127,4 @@ def a_star_search(grid, start, end, heuristic_cost=euclidean_cost):
  
                f_cost[neighbor] = g_cost[neighbor] + heuristic_cost(neighbor, end)

        # if we get to this point, it's not possible to reach the end destination

        return []
      
        return []

nova_rover_demos/pathfinding/comparison_platform/README.md

-Original file line number
+Diff line change
@@ -0,0 +1,103 @@
+    # Path Finding Algorithm Comparison Platform
+    ## Introduction
+    This is software platform aimed at comparing the relative performance of various path planning/route finding algorithms. This software tool benchmarks the processing time, peak memory usage and distance of the path found and presents them in a visual manner. We have designed our path finding software in such a way so easily a range of different algorithms can be added to the platform and their performance can be benchmarked.
+    Currently in the repository there are three classes of pathfinding algorithms are present as default:
+. A* and variants (Heuristic based family)
+. Pledge (Wall follower family)
+. RRT and variants (Probabilistic family)
+    ## Usage
+    Our goal is to create a scalable easy to use function. To use the comparison platform you simply need to:
+. Import your algorithm(s) into the comparison platform
+. Select it inside the `run.py` file
+    ### 1. Importing your Algorithm(s)
+    **Step 1:** You can import the file or folder of the algorithm inside the lib folder.
+    <img alt = "Lib folder" src="img/lib-folder.png" width="250">
+    **Step 2:** Ensure you have fixed any relative imports inside your files. This step is only applicable if you have multiple files are they're importing functions/classes from each other.
+    <img alt = "Relative import" src="img/relative-import.png" width="600">
+    Because our a-star variant file imports a class from a-star and both are under the lib folder, we import in the manner shown in the image above.
+    **Step 3:** Write a wrapper class (If applicable). Currently the comparison platform is designed in a way that all functions accept 3 arguments in the order:
+. List of obstacles (A list of (x, y) tuples)
+. A starting coordinate (A (x, y) format tuple)
+. A goal coordinate (A (x, y) format tuple)
+    If your algorithm accepts other out formats, you can call the main algorithm with all the necessary arguments inside the wrapper function.
+    <img alt = "Wrapper class" src="img/wrapper-demo.png" width="400">
+    We have put together a very [short guide](wrapper.md) on writing the wrapper functions and you can find [here](wrapper.md). It has a template code for you to get started as well.
+    ### 2. Configure `run.py`
+    **Step 4:** Import algorithm or wrapper function in the `run.py` file.
+    <img alt = "Run.py file" src="img/run-file.png" width="200">
+    Importing the files should be done as the following (Remember to do relative importing):
+    <img alt = "Importing algorithms" src="img/runpy-import.png" width="600">
+    **Step 5:** List your algorithm / wrapper function in the algorithm list.
+    <img alt = "Listing algorithms" src="img/algorithm-list.png" width="600">
+    **Step 6:** Run `run.py` file. After completing the easy steps above simply run the following command:
+    ```python
+    python run.py
+    ```
+    You can also select the number of times the functions will be tested. For example:
+    ```python
+    python run.py 10
+    ```
+    **Step 7 (Optional):** Configure how dense you want the environment to be.
+    <img alt = "Map density" src="img/map-density.png" width="600">
+    ### Check algorithm is working correctly
+    Even if no errors are thrown during runtime, you should check the results folder. In the results folder a visual map (PDF file) with the planned route should be automatically generated with the name of your algorithm/wrapper function. Please review this figure and see if everything is okay.
+    You can also check the CSV generated inside the results folder to cross check things.
+    ## Results
+    The main power of this comparison platform lies within the extensive results it generated. The comparison platform gives you visualization of the following key performance factors:
+    - Runtime
+    - Memory Comparison
+    - Average path length
+    Upon running the program you should see a couple of figures which give you a visual comparison of the algorithms key metrics.
+    <img alt = "Result sample 1" src="img/results-fig-1.png" width="350"> <img alt = "Result sample 2" src="img/results-fig-2.png" width="350">
+    Furthermore inside the results folder you can also view the map of the environment and the path generated by each of the algorithms in separate PDF file. Sample:
+    <img alt = "Map sample" src="img/map-sample.png" width="500" style="padding-bottom: 15px;">
+    You have also have access to the raw numbers inside a CSV file inside the results folder.
+    <img alt = "CSV sample" src="img/csv-sample.png" width="600">

nova_rover_demos/pathfinding/comparison_platform/README.md.txt

Empty file.

nova_rover_demos/pathfinding/comparison_platform/img/algorithm-list.png

Sorry, something went wrong. Reload?