Nav2 Optimization Guide

Targeted optimizations to improve Nav2 reliability and obstacle avoidance

---

This guide provides targeted optimizations for Beetlebot's nav2_params.yaml configuration to improve obstacle avoidance and navigation reliability, particularly when navigating tight spaces or corridors.

By applying these tuning suggestions, you can significantly enhance the smoothness of the robot's trajectories and prevent common edge-case navigation freezing.

1. Costmap Inflation Radius

Symptom: The robot attempts to drive too close to walls or pauses frequently when navigating narrow corridors.

Optimization: Increase the inflation_radius in both local_costmap and global_costmap.

• Current Typical Value: 0.25m to 0.3m

• Suggested Value: 0.7m to 1.0m (along with a slightly lower cost_scaling_factor)

• Reasoning: A larger inflation radius creates a stronger cost gradient away from walls, encouraging the global planner to naturally route the robot down the center of hallways rather than hugging edges.

---

2. Velocity-Scaled Lookahead

Symptom: The robot's trajectory becomes erratic or the robot stops near obstacles because it is looking too far ahead on the path.

Optimization: Enable dynamic scaling of the lookahead distance in the FollowPath controller (Regulated Pure Pursuit).

• Setting: Change use_velocity_scaled_lookahead_dist to true.

• Reasoning: When enabled, the controller will use the min_lookahead_dist (e.g., 0.2m) when driving slowly or maneuvering near obstacles, and the standard lookahead_dist (e.g., 0.4m) when driving at speed. This allows for much finer control and prevents looking "through" adjacent obstacles.

---

3. Rotate to Heading Angle Threshold

Symptom: The robot swings its corners into walls when trying to turn while moving forward.

Optimization: Reduce the threshold at which the robot decides to rotate in place rather than driving in a wide arc.

• Setting: Decrease rotate_to_heading_min_angle from 0.785 (45°) to 0.2 or 0.3 rad (11-17°).

• Reasoning: For a skid-steer base, rotating in place is often safer in tight quarters. A lower threshold ensures the robot rotates to face its path before driving forward, minimizing outward corner sweeping.

---

4. Transform Tolerance

Symptom: Jerky braking or delayed obstacle avoidance, especially when moving quickly.

Optimization: Reduce transform_tolerance across AMCL, costmaps, and the controller.

• Current Value: 1.0 seconds

• Suggested Value: 0.2 to 0.5 seconds

• Reasoning: A high tolerance can mask latency issues on slower CPUs, but it allows the navigation stack to use stale position data (up to 1 second old). Tuning this down ensures the robot reacts to obstacles based on its true, real-time position.

---

5. Reversing Behavior

Symptom: The robot gets stuck in corners and cannot back up fluidly to clear the turn.

Optimization: Allow the local controller to reverse.

• Setting: Change allow_reversing to true in the FollowPath configuration.

• Reasoning: Allowing the Regulated Pure Pursuit controller to plan reverse maneuvers provides an immediate, smooth escape from tight spots without triggering the slower, generic backup recovery behavior.