PathFinder.jl - Sampling-Based Path Planning Algorithms in Julia

This Julia package provides implementations of two popular sampling-based path planning algorithms:

• Rapidly exploring random tree (RRT)

• Probabilistic roadmap (the simplified, k-nearest variant: k-Nearest sPRM)

Installation

This package is not registered, but you can install it using Julia's Pkg package manager:

(@v1.10) pkg> add https://github.com/B0B36JUL-FinalProjects-2024/Project_draholeo

Usage

1. Define your PlanningSpace

   using PathFinder
   
   dimension_of_my_configuration_space = 2
   limits_matrix_of_my_configuration_space = [0 1; 0 1]
   my_collision_checker = conf -> false  # environment with no obstacles
   
   ps = PlanningSpace(
       dim=dimension_of_my_configuration_space,
       limits=limits_matrix_of_my_configuration_space,
       collision_check=my_collision_checker,
       collision_resolution=0.01)

   where:

   • PlanningSpace.dim is the dimension of the configuration space
   • PlanningSpace.limits is (dim, 2) Matrix with [min, max] for each dimension
   • PlanningSpace.collision_check is a user-defined function that takes a configuration (a dim-dimensional vector) as input and returns true if the configuration is in collision with an obstacle, otherwise false
   • PlanningSpace.collision_resolution defines the resolution at which the planner checks for collisions along a path (smaller values result in finer collision checks but may slow down planning)

2. Select your favorite planner

   planner = RRTPlanner(step_size=0.05, max_iters=10000, tolerance=0.05)

   or

   planner = PRMPlanner(roadmap_size=200, k_neighbors=8)

3. Run the plan function

   start_conf = [0.1, 0.1]
   goal_conf = [0.9, 0.9]
   path, graph = plan(start_conf, goal_conf, ps, planner)

   which returns:

   • path from start_conf to goal_conf ((dim, num_of_waypoints) Matrix{Float64}) or nothing if the planning algorithm could not find a collision-free path
   • graph which is either a PathFinder.Graph for PRM algorithm or PathFinder.Tree for RRT algorithm

For a complete working example, see examples/example.jl.

Author

Leoš Drahotský (draholeo@fel.cvut.cz)

License

This project is licensed under the MIT License - see the LICENSE file for details.