[icon ] blenderdumbass . org [icon scene] Articles

Path Finding in UPBGE for Dani's Race

[avatar]  Blender Dumbass

February 04, 2025

👁 75

https://mastodon.social/ : 👁 4
https://lemmy.dbzer0.com/ : 👁 1
https://blenderdumbass.org/ : 👁 2
https://blenderdumbass.org/search?text=%23UPBGE&tags=on : 👁 1
https://blenderdumbass.org/articles : 👁 1

#DanisRace #MoriasRace #Gamedev #Game #UPBGE #blender3d #animation #GTAClone #programming #project #Navigation #NPC #AI

License:
Creative Commons Attribution Share-Alike
Not AI Generated! [icon internet] See Proof
Audio Version





Here is a scene for you. You messed up a race. Your Paps is very very mad at you right now. He doesn't trust you to drive. Yet he wants to teach you to drive. So he takes you to a karting club. This is a scene I'm working on in Dani's Race right now. A scene in a video game I'm making which should not have cut scenes. Which means, for this scene to work, Paps needs to walk from his room to a car, wait for you, the player to sit with him into said car, and then drive you across a town to a completely different location. Seamlessly.


[embedded image]




Uno momento...

In the Dani's Race Matrix Room Franzo said that when Petition for Version 2025-01-19 will reach 235 signatures, he will stream Dani's Race once again. It's already at 230. Make it happen!

Anyway...


It should be simple, you say. Just use the Steering actuator inside of UPBGE, pair it with a basic Navigation Mesh and you are done. Wrong! As you maybe know Dani's Race Cars are not simple. And to control them properly ( as in: in a way that will not break the flow of the game ) is more tricky than that.

I have to give the car specific control commands to accelerate, decelerate, how much to turn the front wheels and stuff. And the steering actuator does non of that. I though I was screwed.


[embedded image]


There was already one scene where you could see a character taking a car and doing something with it autonomously. Which wasn't a big deal anyway since even if it would fail, it would not actually impact the gameplay much. Still, the fact that I managed to make it work, made me feel like I could do that.

This stunt in particular was accomplished in two stages. First a very simple path finding operation calculated the path of the car from outside of the racetrack to the inside of the racetrack. From which a different algorithm took over. Which is the racing algorithm I use for racing. But which is implemented smartly enough that I could use it manually for a specific car in a specific scene.


[embedded image]


Both of those methods used navigation points placed in different places, from which the cars could figure out where to go. For racing it was a rather straight forward thing to do. Just place enough of those checkpoints, make sure that they are all in order. And instruct the car to race as fast as possible towards one of them. And as soon as it reaches the checkpoint, make it go to the next one.

vect = car.getVectTo(checkpoint)
turn = vect[2][0]
Accelerate(car)


This is the most basic algorithm that you can write. Well... Okay. I ended up spending a lot of time actually making it good. Making the cars know when to speed up and when to slow down. Making them see the road and obstacles. Teaching them the optimal racing lines. All of that had to be done at some point. You can look at the RacingAI function to see the extend of the insanity. But you get the point. You rotate the front wheels toward the target and accelerate.

With a race it's not hard. It's literally a line. There is no path to find. But what about navigation through the town? Originally I was stupid enough to just bombard the town with bazillion points on each road, hoping that I can write something clever enough for cars to self-navigate themselves using those. But unfortunately it's not that simple.

Path finding algorithms require you to have connections between points. And with my "just points" idea, everything was technically connected to everything else. There were no walls ever. The cars would attempt to cut through everything. Which did not resemble sanity what so ever.

So I decided, since I have to figure it out for this scene to make sense, I will deal with it somehow. And so I made this:


[embedded image]


A navigation mesh made out of points, but with those points being connected together with lines, called edges. Giving me something with which I could maybe potentially implement a path finder algorithm.

But then it struck me.

Wait a second... Blender kind of has a path finder built into it. You can select a vertex and then with a control select another vertex. And guess what... Blender will select all of the vertices in between, finding the shortest path. And because UPBGE is technically blender under the hood. And because they finally figured out how to make bpy module run inside of the game itself. This meant I could do this:

NavigationMesh = bpy.data.objects['NavigationMesh']
bpy.ops.object.select_all(action='DESELECT')
NavigationMesh.select_set(True)
bpy.context.view_layer.objects.active = NavigationMesh

bpy.ops.object.mode_set(mode = 'EDIT')
bpy.ops.mesh.select_all(action='DESELECT')
bpy.ops.object.mode_set(mode = 'OBJECT')
NavigationMesh.data.vertices[index1].select = True
NavigationMesh.data.vertices[index2].select = True
bpy.ops.object.mode_set(mode = 'EDIT')
bpy.ops.mesh.shortest_path_select()
bpy.ops.object.mode_set(mode = 'OBJECT')


Let's go over it, line by line, what it does: We find our Navigation Mesh object. We deselect everything, so nothing will be technically selected in the blender side of things. We select out Navigation Mesh. And we make it active too. Blender has a weird way to make selection complicated.

We go to the edit mode with that mesh. We deselect everything inside of that mesh. Then it's a little bit strange. We want to select our start and end points on the mesh. But for some reason you need to switch to Object mode when doing it. Anyway. We do that. We hop to the object mode, select our start and end points. Then hop back in. And do the magic part. We tell blender to select all vertices between the two selected ones. This gives us our path.

Well... Actually... there is more code to sort those points. And more code to execute the path that was found. Both of which will be released at some point. But first 2025-01-19 should get released. Anyway... This way I can find a path made out of little checkpoints that now I can use, like in the race, to make Paps drive Dani to the karting club. Or anybody drive anywhere for any future scenes that I might need.

Happy Hacking!!!


[icon terminal] JSON [icon markdown] Markdown

[icon question] Help

Subscribe RSS
[icon link] Author
[icon link] Website
Share on Mastodon


[icon question] Help


You can comment from Mastodon.










[icon reviews]Close Encounters Of The Third Kind

[thumbnail]

[avatar]  Blender Dumbass

👁 42 💬 0



During the making of Close Encounters Of The Third Kind Steven Spielberg was already kind of a big man in Hollywood. But from the other side he was still young. Technically speaking this was his third theatrical film. He did work for television from the late 60s. At that time he already directed a number of feature length TV movies. One of them was the famous Duel. And then he did only 2 theatrical pictures: the 1972's The Sugarland Express and the 1974's Jaws. And now there is this movie.


[icon petitions]Release: Dani's Race v2025-03-17

[thumbnail]


23 / 50 Signatures

[avatar]  Blender Dumbass

👁 280 💬 2



Dani's Race version 2025-03-17


#DanisRace #MoriasRace #Game #UPBGE #blender3d #project #petition #release


[icon reviews]The Substance Will Make You Puke While Cumming

[thumbnail]

[avatar]  Blender Dumbass

👁 61 💬 0



I am still unsure whether Coralie Fargeat meant for The Substance to be taken seriously or not. There was a similar misunderstanding when it comes to Lars Von Trier's The House That Jack Built, where the audience were laughing, unable to comprehend in the intensity of the film, while the director was dead serious. The experience I had watching The Substance reminded me of this confusion. The film is so over the top, it beats the absurdity of Sam Raimi's horror-comedies.



[icon reviews]Miami Vice

[thumbnail]

[avatar]  Blender Dumbass

👁 56 💬 0



Good film directors tend to take upon themselves projects that risk being misunderstood sometimes. An artist can't just draw the same drawing over and over. He wants to experiment. Director Michael Mann is famous with his films about crime. And the realistic depiction of what crime and police work actually looks like. But if you are doing this over and over, you tend to become interested in something else. Which is terrible for a director who has fans with certain expectations.


[icon codeberg] Powered with BDServer [icon python] Plugins [icon analytics] Analytics [icon mastodon] Mastodon [icon peertube] PeerTube [icon element] Matrix
[icon user] Login