Warning
This page is located in archive.

T1d-plan - Grid-based path planning

The main task is to implement a simple grid-based planning approach.

Deadline 19. October 2019 23:59 PST
Points 3
Label in BRUTE t1d-plan
Files to submit archive with GridPlanner.py
Resources T1d-plan package

Assignment

In GridPlanner.py implement the following basic steps in of the grid-based navigation pipeline:

  1. Implement the obstacle growing (method grow_obstacles) to take into account the robot embodiment
  2. Implement a grid-based planning method (method plan) to find a path between the start and goal position; in GridPlanner.py there are prepared some A* helper functions, but the task can be solved by implementing another optimal grid-based planner
  3. Implement the trajectory smoothing method (method simplify_path) which purpose is to provide a smoother path for the path following controller that drives the real robot

Approach

  1. Obstacle growing - can be achieved by different means, typically methods of mathematical morphology are being used. In particular, binary dilation of the obstacles for a predefined distance which is set by a user to take into account the robot embodiment. Typically, half of the size of the being used as the distance. Further, methods based on distance transform can be also used to grow the obstacles or as a heuristic function for the planner to stay away from walls.
  2. Planning - grid-based path planning takes the grid map, starting position(start) and goal position(goal) on the input and provide a list of cell coordinates on the output. If the path is not found, the planner returns None
  3. Path simplification - path simplification is usually done by excluding navigation points from the path, that are not necessary in the sense that the robot does not have to visit them precisely. A typical approach to trajectory smoothing is to connect the neighboring segments one by one using straight-line segments (using Bresenham line algorithm) up to the point where the straight-line segment collide with an obstacle (grown obstacle) and then follow with another straight-line segment.

Evaluation

The code can be evaluated using the following script (also attached as t1d-plan-eval.py)

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
 
import sys
import math
import time
import numpy as np
import matplotlib.pyplot as plt
 
sys.path.append('robot')
sys.path.append('gridmap')
sys.path.append('gridplanner')
 
import Robot as rob
import GridMap as gmap
import GridPlanner as gplanner
 
PLOT_ENABLE = True
 
if __name__=="__main__":
 
    #define planning problems:
    #  map file 
    #  map scale [m] (how big is a one map cell in comparison to the real world)
    #  start position [m]
    #  goal position [m]
    #  execute flag
    scenarios = [("maps/maze01.csv", 0.1, (8.5, 8.5), (1, 1), False),
                 ("maps/maze02.csv", 0.1, (8.5, 8.5), (5.6, 3), False),
                 ("maps/maze02.csv", 0.1, (8.5, 8.5), (5.8, 2.5), False),
                 ("maps/maze03.csv", 0.1, (19.5, 23.0), (6.2, 26.4), False),
                 ("maps/maze04.csv", 0.1, (51.6, 53.3), (19.2, 12.4), False),
                 ("maps/maze05.csv", 0.1, (78.8, 79.2), (7.0, 14.6), False),
                 ("maps/maze02.csv", 0.1, (8.5, 8.5), (5.6, 3), True)]
 
    #fetch individual scenarios
    for scenario in scenarios:
        mapfile = scenario[0] #the name of the map
        scale = scenario[1]
        start = scenario[2]   #start point
        goal = scenario[3]    #goal point
        execution_flag = scenario[4] #execute the trajectory in vrep
 
        #instantiate the map
        gridmap = gmap.GridMap()
 
        #load map from file
        gridmap.load_map(mapfile, 0.1)
 
        #plot the map with the start and goal positions
        if PLOT_ENABLE:
            gmap.plot_map(gridmap)
            gmap.plot_path(gridmap.world_to_map([start, goal]))
            plt.show()
 
            #show the free/occupied space
            gmap.plot_map(gridmap, clf=True, data='free')
            plt.show()
 
        planner = gplanner.GridPlanner()
 
        #blow the obstacles to avoid collisions
        planner.grow_obstacles(gridmap, 0.4)
 
        #show the map after obstacle blowing
        if PLOT_ENABLE:
            gmap.plot_map(gridmap, clf=True, data='free')
            plt.show()
 
        #plan the route from start to goal
        path = planner.plan(gridmap, gridmap.world_to_map(start), gridmap.world_to_map(goal))
 
        if path == None:
            print("Destination unreachable")
            continue
 
        #show the planned path
        if PLOT_ENABLE:
            gmap.plot_map(gridmap)
            gmap.plot_path(path)
            plt.show()
 
        #simplify the path
        path_s = planner.simplify_path(gridmap, path)
 
        #show the simplified path
        if PLOT_ENABLE:
            gmap.plot_map(gridmap)
            gmap.plot_path(path)
            gmap.plot_path(path_s, color='blue')
            plt.show()
 
        if execution_flag:
            #instantiate the robot
            robot = rob.Robot()
 
            #execute the path
            for waypoint in path_s:
                #navigate the robot towards the target
                status1 = robot.goto(gridmap.map_to_world(waypoint))
 
                #check for the execution problems
                if not status1:
                    print("The robot has collided en route")
                    break


Appendix


FAQ and Known Issues
  1. Cell exactly distance of an obstacle are considered traversable. Use <, not .
  2. Mind the path direction. The path should always lead from start to goal.
  3. The plan function is parametrized with the neighborhood type. Both N4 and N8 should be supported.
courses/b4m36uir/hw/t1d-plan.txt · Last modified: 2019/10/22 19:32 by pragrmi1