non-holonomic-prm-planner

A PRM based non-holomic constraints based global planner implementation for ackermann vehicles. It is also capable of generating reverse paths.

Brief Overview

• Sample the C-space and collect collision free vertices
• Non-holonomic edge generation using steering curves
• Shortest Path Search over the roadmap

Running

Update constants.h file with the vehicle params go to the catkin workspace and build the package

    catkin build non-holonomic-prm-planner

Publish the obstacles polygons
(Make sure to pass one of the files provided in the obstacles folder)

  rosrun polygon_publisher polygon_provider_new_csv.py

Publish the global costmap

  rosrun global_costmap_publisher costmap_publisher_offline.py

Run the PRM node

  rosrun non-holonomic-prm-planner non-holonomic-prm-planner

Run Rviz and pass the config file in the Rviz folder

Headers Overview

• collisiondetectionpolygon.h => manages collision detection between convex polygons
• pathgenerator.h => responsible for generating a collision free path between the start and the goal points, given a roadmap
• sampler.h => implementation of intial sampling strategies
• roadmap.h => responsible for the roadmap generation
• kdtree.h => responsible for kdtree generation from the initial sampled points
• robotmodel.h => basic robot model representation
• steeringcurve.h => non-holonomic constraints modelling used to generate the roadmap
• utils.h => helper functions
• visualization.h => constains visualization helpers to see the generated roadmap, steering curves and the final plan

Screenshots

TODO

• Create launch file
• Parametrize ackermann vehicle params