Overview
The Dynamic Window Approach samples translational and angular velocities that the robot can reach within acceleration limits, simulates short trajectories, and chooses a collision-free command with a good score.
CoreSource checked
Dynamic Window Approach
Chooses robot velocity commands by simulating short feasible motions and scoring them.
Quick Facts
Compact metadata for implementation and discovery.
- Category
- Pathfinding
- Difficulty
- Intermediate
- Family
- Robot Navigation
- Tier
- Core
- Status
- Verified
- Review
- Source Checked
- Sources
- 1
- Formulas
- 0
- Relationships
- 3
- Code example
- Ready
- Visualization
- Ready
Core Concepts
local robot navigationrobot navigationrandom samplingpathfinding
Quick facts
Complexity
Runtime and memory behavior at a glance.
Best
O(n)
Average
O(n log n)
Worst
O(n^2)
Space
O(n)
When to Use
- Use when a robot needs local collision avoidance from current state and sensor data.
- Use when a global planner can provide a goal direction but local dynamics still matter.
When Not to Use
- Avoid when long-horizon global optimality is required by itself.
- Avoid when dynamics, obstacle prediction, or footprint checks are too inaccurate.
What It Does
- Samples reachable velocity commands
- Predicts short trajectories
- Scores commands by safety, goal progress, and speed
Use Cases
- Local mobile-robot navigation
- Reactive obstacle avoidance
- Velocity-space planning with acceleration limits
Inputs
- Robot Navigation input instance for robot navigation
- Configuration parameters controlling local robot navigation
Outputs
- Computed robot navigation solution produced by Dynamic Window Approach
- Diagnostic state over local robot navigation and robot navigation
Pseudocode
- 1.Compute the dynamic window from current velocity and acceleration limits
- 2.Sample candidate linear and angular velocities
- 3.Roll out each candidate for a short horizon
- 4.Reject trajectories that collide with obstacles
- 5.Score remaining trajectories and return the best command
Canonical Example
- Input
- robot at (0,0), obstacle ahead, candidate velocities turn-left and go-straight
- Output
- turn-left command if go-straight collides in the rollout
- Explanation
- The chosen velocity is feasible under acceleration limits and scores better after collision checking.
Implementation Notes
- Separate velocity sampling, trajectory rollout, collision checks, and scoring.
- Tune heading, clearance, and velocity weights on representative maps.
- Test narrow passages and moving-obstacle edge cases.
Common Pitfalls
- Ignoring acceleration limits makes sampled commands infeasible.
- Short horizons can get trapped in local minima.
- Footprint and clearance calculations must match the robot geometry.
Interactive Mechanism Walkthrough
2D Grid Wavefront & TrajectoryShow sampling feasible velocity commands and scoring short simulated trajectories on a small local robot navigation example.
S
#
#
#
#
#
#
#
G
Step 1 of 5: Compute the dynamic window from current velocity and acceleration limits
pathfinding-mechanismNotes
DWA is a local planner; it is commonly paired with a global planner or navigation objective.
Sources (1)
What each reference supports for this entry.
Dynamic Window Approach
PaperOriginal dynamic window approach paper for velocity-space local robot collision avoidance.
Open sourceExplore Next
Good next jumps from the relationship graph, ranked for useful discovery.
Conceptual neighborA* Search
Pathfinding · Core
Dynamic Window Approach and A* Search both relate to local robot navigation, but use different structures or assumptions.
Compare with thisRelationships
View full relationship pageConceptual neighborA* Search
Dynamic Window Approach and A* Search both relate to local robot navigation, but use different structures or assumptions.
Conceptual neighborD* Search
Dynamic Window Approach and D* Search both relate to local robot navigation, but use different structures or assumptions.
Conceptual neighborLifelong Planning A*
Dynamic Window Approach and Lifelong Planning A* both relate to local robot navigation, but use different structures or assumptions.
Related Algorithms
Tags
PathfindingGraphRandomness