gdc2006 miles david pathplanning

8/12/2019 Gdc2006 Miles David Pathplanning

1/51

Crowds In A Polygon Soup

Next-Gen Path Planning

David Miles

3/23/2006

Copyright 2006 BabelFlux LLC


2/51

Large-Scale Fluid AI Navigation in

Complex, Changing Worlds

1. Build a usable free space nav graph

2. Update graph for dynamic obstacles

3. Use graph for fluid navigation


3/51

1. Building the Nav-Graph

Automated Build Process

1.Large, Complex Worlds2.Polygon Soup Mesh

3.Multiple Unit Sizes


4/51

2. Dynamic Obstacle Updates

1.Obstacles reconfigure the world

2.Dynamic areas replace static areas

3.Underlying algorithms work exactly as before


5/51

1. Fast ray casts

2. Edge detection

3. Potential Fields4. Path Following

3. Fluid AI Navigation


6/51

Automated Build Process


7/51

1. Voxelize Polygon Soup


8/51

2. Extract Edge


9/51

3. Simplify Polygon


10/51

4. Partition Into Convex Areas


11/51

Convex Partitioning


12/51

Recursively Partition


13/51

Convex Area Graph Complete


14/51

Pathfinding


15/51

Holes in the Free-Space


16/51

Splice Location


17/51

Add Splice Segments


18/51


19/51

Convex Area Graph Complete


20/51

Overhead Obstacles


21/51

Walkable Surface Removal


22/51


23/51

3d Partitioning


24/51


25/51

Different Creature Shapes


26/51


27/51

Non-Uniform Voxelization


28/51

Part2: Dynamic Obstacles

Dynamic obstacles reconfigure world

Underlying algorithms unchanged


29/51

Boolean Subtract

-

=


30/51

Partition Non-Convex Areas


31/51

Multiple Obstacles


32/51

Memory Layout


33/51

Part3: Fluid AI Navigation

What can I do with my convex area graph?

Very fast navigational ray-casts

Distance to edge queries Repulsion fields on edges

Smart path following


34/51

Ray Casts


35/51

Smart Path Following

Desired flocking behavior

Common approach attracts

each flock member to the

minimum distance path. Does

not adapt to available

freespace.

Repulsion from edges can

easily trap individuals in local

minima.


36/51

For The Pathing

Connoisseur

Unneeded return to

the original path

annoys the pathingconnoisseur

Ahhh, much better


37/51

Next Corner

Each member of the flock steers left

or right to head towards the next corner.


38/51

Finding the Next Corner


39/51


40/51

Initializing the Loop


41/51

First Iteration, Left Side


42/51

First Iteration, Right Side


43/51

First Iteration Complete


44/51

Second Iteration, Left Side


45/51

Second Iteration, Right Side


46/51

Second Iteration Complete


47/51

Third Iteration, Left Side


48/51

Final Output


49/51

Next Corner Summary

Easy to compute from convex graph

< 2 cross-products per iteration

Can limit iterations Never explicitly create min-dist path

Fluid path following even with large

disturbances


50/51

Conclusions

Automated build of convex area graph efficiently represents the usable free space

operates on polygon soup mesh

settable enemy size and shape Dynamic obstacles update graph

no speed overhead once updated

Fluid AI navigation using the graph Many useful queries performed rapidly


51/51

Special Thanks

Meilin Wong

Hong Park and David Modiano

Crystal Dynamics

[email protected]

www.navpower.com
mailto:[email protected]:[email protected]

gdc2006 miles david pathplanning

Documents