Navfn Planner

The NavfnPlanner is a planning module that implements the nav2_behavior_tree::ComputePathToPose interface.

A planning module implementing the nav2_behavior_tree::ComputePathToPose interface is responsible for generating a feasible path given start and end robot poses. It provides the equivalent functionality to a GlobalPlanner in ROS1 MoveBase.

Status

Currently, NavfnPlanner’s core algorithm is a direct port from the ROS1 MoveBase Navfn planner. The Navfn planning algorithm is based on the Global Dynamic Window Approach.

Characteristics

In Dijkstra mode (use_astar = false) Dijkstra’s search algorithm is guaranteed to find the shortest path under any condition. In A* mode (use_astar = true) A*’s search algorithm is not guaranteed to find the shortest path, however it uses a heuristic to expand the potential field towards the goal.

The Navfn planner assumes a circular robot and operates on a costmap.

Task Interface

The Navigation System is composed of three tasks: NavigateToPose, ComputePathToPose and FollowPathToPose.

The DijkstraPlanner implements the Task Server interface for ComputePathToPose, specifically derives from nav2_behavior_tree::ComputePathToPoseTaskServer. The client to DijkstraPlanner is ‘NavigateToPoseTask’, which periodically sends requests to the path planner.

Next Steps

• Refactor Navfn. Currently difficult to modify/extend. Issue #244
• Implement additional planners based on optimal control, potential field or other graph search algorithms that require transformation of the world model to other representations (topological, tree map, etc.) to confirm sufficient generalization. Issue #225
• Implement planners for non-holonomic robots. Issue #225