As mentioned in the move_base wiki:
~recovery_behavior_enabled (bool,
default: true) Whether or not to
enable the move_base recovery
behaviors to attempt to clear out
space.
If that's what you tried to turn false, there's not enough information to say why it's not working. If it's set in multiple places, one instance may be overriding the false setting. In any case, since you're planning to use your own recovery behavior anyway, you can override the default behaviors with your own nav_core::RecoveryBehavior plugin:
~recovery_behaviors (list, default: [{name: conservative_reset, type: clear_costmap_recovery/ClearCostmapRecovery}, {name: rotate_recovery, type: rotate_recovery/RotateRecovery}, {name: aggressive_reset, type: clear_costmap_recovery/ClearCostmapRecovery}] For 1.1+ series)
A list of recovery behavior plugins to use with move_base, see pluginlib documentation for more details on plugins. These behaviors will be run when move_base fails to find a valid plan in the order that they are specified. After each behavior completes, move_base will attempt to make a plan. If planning is successful, move_base will continue normal operation. Otherwise, the next recovery behavior in the list will be executed. These plugins must adhere to the nav_core::RecoveryBehavior interface specified in the nav_core package. (1.0 series default: [{name: conservative_reset, type: ClearCostmapRecovery}, {name: rotate_recovery, type: RotateRecovery}, {name: aggressive_reset, type: ClearCostmapRecovery}]). Note: For the default parameters, the aggressive_reset behavior will clear out to a distance of 4 * ~/local_costmap/circumscribed_radius.
If you would like more specific help, please provide the configuration and launch files you're using, so we can determine what the issue may be.
