Local planner (DWA, TEB) in ROS - Performance & Optimization

Performance: Local planner (DWA, TEB)

HIGH IMPACT

This concept affects the real-time responsiveness and smoothness of robot navigation by planning local paths quickly and efficiently.

Planning local robot paths in dynamic environments

ROS

Use DWA (Dynamic Window Approach) or TEB (Timed Elastic Band) local planners that sample velocities and optimize trajectories efficiently.

These planners quickly generate feasible velocity commands with low CPU overhead, enabling smooth and responsive navigation.

📈 Performance GainReduces CPU blocking to under 10 ms per cycle, enabling real-time updates.

Planning local robot paths in dynamic environments

ROS

Use a global planner only or a local planner with high computational cost and no optimization, e.g., a naive full search without velocity sampling.

This causes slow path updates and high CPU usage, leading to delayed robot reactions and jerky movements.

📉 Performance CostBlocks CPU for 100+ ms per cycle, causing lag in navigation updates.

Performance Comparison

Pattern	CPU Usage	Planning Latency	Responsiveness	Verdict
Naive full search local planner	High (80-100%)	100+ ms	Low (laggy)	[X] Bad
DWA local planner	Low (10-20%)	5-10 ms	High (smooth)	[OK] Good
TEB local planner	Moderate (15-30%)	8-15 ms	High (smooth, optimized)	[OK] Good

Rendering Pipeline

Local planners process sensor data and robot state to compute velocity commands in a loop, affecting the control pipeline rather than visual rendering.

→Computation

→Control Loop

→Sensor Processing

⚠️ BottleneckComputation stage for trajectory optimization and collision checking

Optimization Tips

1Use velocity sampling and pruning to reduce CPU load in local planners.

2Optimize trajectory calculations to keep planning latency under 20 ms.

3Avoid naive exhaustive searches that block the control loop.

Performance Quiz - 3 Questions

Which local planner approach is best for minimizing CPU load during real-time robot navigation?

A Dynamic Window Approach (DWA) with velocity sampling

B Naive full search of all possible paths

C Random path generation without optimization

D Global planner only without local updates

DevTools: ros2_tracing or rqt_graph

How to check: Use ros2_tracing to record planning node CPU usage and latency; visualize with tracing tools. Use rqt_graph to verify planning node frequency.

What to look for: Look for low CPU usage spikes and consistent planning cycle times under 20 ms for smooth navigation.

Practice

1. What is the main purpose of a local planner like DWA or TEB in ROS?

easy

A. To control the robot's battery usage

B. To map the entire environment for the robot

C. To help the robot navigate safely around obstacles nearby

D. To communicate with other robots

Solution

1. Step 1: Understand the role of local planners

   Local planners focus on immediate surroundings to avoid obstacles and plan short paths.

2. Step 2: Compare options

   Mapping the environment is done by global planners or mapping nodes, not local planners.

3. Final Answer:

   To help the robot navigate safely around obstacles nearby -> Option C

4. Quick Check:

   Local planner = safe nearby navigation [OK]

Hint: Local planners handle nearby obstacles, not mapping or communication [OK]

Common Mistakes:

• Confusing local planner with global planner
• Thinking local planner maps environment
• Assuming local planner manages robot hardware

2. Which of the following is the correct way to enable the DWA local planner in a ROS navigation stack configuration file?

easy

A. base_local_planner: "dwa_local_planner/DWAPlannerROS"

B. base_local_planner: "teb_local_planner/TEBPlanner"

C. planner: "global_planner/GlobalPlanner"

D. planner: "dwa_local_planner/DWAPlannerROS"

Solution

1. Step 1: Identify the correct parameter name for local planner

   The parameter to specify the local planner plugin is usually 'base_local_planner'.

2. Step 2: Match the correct plugin name for DWA

   DWA local planner plugin is 'dwa_local_planner/DWAPlannerROS'. base_local_planner: "dwa_local_planner/DWAPlannerROS" matches this exactly.

3. Final Answer:

   base_local_planner: "dwa_local_planner/DWAPlannerROS" -> Option A

4. Quick Check:

   Correct plugin name and parameter = base_local_planner: "dwa_local_planner/DWAPlannerROS" [OK]

Hint: Look for 'base_local_planner' key with DWA plugin name [OK]

Common Mistakes:

• Using 'planner' instead of 'base_local_planner'
• Confusing DWA and TEB plugin names
• Using global planner plugin name for local planner

3. Given the following ROS parameter snippet for TEB local planner:

teb_local_planner:
  max_vel_x: 0.5
  max_vel_theta: 1.0
  min_turning_radius: 0.2

What will happen if the robot tries to turn with an angular velocity of 1.5 rad/s?

medium

A. The planner will allow the turn at 1.5 rad/s

B. The planner will limit the angular velocity to 1.0 rad/s

C. The planner will stop the robot immediately

D. The planner will ignore the max_vel_theta parameter

Solution

1. Step 1: Understand max_vel_theta parameter

   This parameter sets the maximum allowed angular velocity for the robot.

2. Step 2: Compare requested angular velocity with max_vel_theta

   The requested 1.5 rad/s exceeds the max of 1.0 rad/s, so the planner limits it to 1.0 rad/s.

3. Final Answer:

   The planner will limit the angular velocity to 1.0 rad/s -> Option B

4. Quick Check:

   Angular velocity capped by max_vel_theta = The planner will limit the angular velocity to 1.0 rad/s [OK]

Hint: Planner caps velocities at max limits set in parameters [OK]

Common Mistakes:

• Assuming planner allows velocities above max
• Thinking planner stops robot on limit breach
• Ignoring max_vel_theta parameter

4. You configured the DWA local planner but the robot keeps colliding with obstacles. Which of the following is the most likely cause?

medium

A. The obstacle distance parameter is set too low, causing late avoidance

B. The global planner is disabled

C. The robot's battery is low

D. The robot's wheels are not calibrated

Solution

1. Step 1: Identify parameters affecting obstacle avoidance

   DWA uses parameters like obstacle distance to decide when to start avoiding obstacles.

2. Step 2: Analyze effect of low obstacle distance

   If set too low, the planner reacts too late, causing collisions.

3. Final Answer:

   The obstacle distance parameter is set too low, causing late avoidance -> Option A

4. Quick Check:

   Late avoidance due to low obstacle distance = The obstacle distance parameter is set too low, causing late avoidance [OK]

Hint: Check obstacle distance parameter if collisions occur [OK]

Common Mistakes:

• Blaming global planner for local planner issues
• Ignoring parameter tuning for obstacle avoidance
• Assuming hardware issues cause collisions without evidence

5. You want your robot to navigate a narrow corridor smoothly using the TEB local planner. Which combination of parameters should you adjust to improve smooth turning and obstacle clearance?

hard

A. Increase min_turning_radius and decrease obstacle_avoidance_force

B. Increase max_vel_x and decrease max_vel_theta

C. Decrease max_vel_x and disable obstacle avoidance

D. Decrease min_turning_radius and increase obstacle_avoidance_force

Solution

1. Step 1: Understand min_turning_radius effect

   Decreasing min_turning_radius allows tighter turns, useful in narrow corridors.

2. Step 2: Understand obstacle_avoidance_force effect

   Increasing obstacle_avoidance_force makes the planner keep safer distance from obstacles, improving clearance.

3. Step 3: Combine effects for smooth navigation

   Tight turns plus stronger obstacle avoidance helps smooth, safe navigation in tight spaces.

4. Final Answer:

   Decrease min_turning_radius and increase obstacle_avoidance_force -> Option D

5. Quick Check:

   Tight turns + strong avoidance = Decrease min_turning_radius and increase obstacle_avoidance_force [OK]

Hint: Tighten turning radius and boost avoidance force for narrow spaces [OK]

Common Mistakes:

• Increasing turning radius makes turns wider, not tighter
• Disabling obstacle avoidance causes collisions
• Confusing velocity limits with turning parameters