recovery-behavior-tree

To configure Nav2 recovery behavior trees, use a two-child RecoveryNode that pairs a normal navigation PipelineSequence with a ReactiveFallback containing a RoundRobin of recovery actions like ClearCostmaps, Spin, Wait, and BackUp, triggered sequentially when the planner fails.

When ROS 2 navigation hits an obstacle, you should cycle through standard recovery actions using a RoundRobin strategy within a ReactiveFallback, including clearing costmaps, spinning in place, waiting, and backing up. A GoalUpdated guard ensures recovery is canceled if a new goal is received.

A RecoveryNode in a ROS 2 Nav2 behavior tree operates as a two-branch structure where the first child attempts normal navigation and the second child executes recovery actions upon failure. It retries the normal operation after a successful recovery action, up to a configurable number of retries.

Yes, you can limit recovery attempts in a Nav2 behavior tree using the configurable number_of_retries parameter. This parameter controls how many times the RecoveryNode cycles through its escalating recovery strategies before giving up and reporting a navigation failure.

A Nav2 planner may fail to recover if the behavior tree lacks a structured recovery sequence with a ReactiveFallback and RoundRobin to cycle through costmap clearing, spinning, waiting, or backing up. Without these escalating recovery strategies and a retry limit, the navigation pipeline cannot adapt to transient obstacles.