Robot Rumble 2.0 > Robot Rumble 2.0

Robot Rumble 2.0 - Robot Combat Simulator - Under Development

<< < (2/472) > >>

Tashic:
Oooo! Nuts and bolts too!

Nice ampflow model as well.

Loving the details :)

cjbruce:

--- Quote from: Tashic on July 14, 2017, 10:16:29 AM ---Nice work there! Very impressed.

It's cool that you are trying to make it a simulator, rather than just a game.

I see that the wheels are very smooth, how did you achieved that? Did you use one of the standard colliders or something else?

--- End quote ---

Thank you!

For the original version of the game (Robot Rumble 1.0), I used Oimo.js physics, which uses cylinder colliders and is very smooth.  I just started with Unity, and it looks like PhysX is the only option, so I am experimenting with different drive mechanisms.  So far I have tried the following:

1. Forces applied directly to the model when keys are pressed (not at all realistic).
2. The standard Unity wheel colliders + white frictionless "casters" in the corners.  This is the model that you see in the link.
3. Cylindrical mesh colliders with hinge joints.  I made 16-sided, 32-sided, and 64-sided meshes in Blender and pulled them into Unity to see how they behave.  32- and 64-sided meshes work fine, but aren't quite as smooth as the wheel collider.

I am also considering an improved wheel collider that ray-casts in multiple directions for each wheel, but for now I am pretty happy with solution #2.  It has really low overhead and works wheel for protected wheels.  For flippable robots I will need to take a different approach, but right now I am just worried about getting everything running.

My target platform will be desktop with game controllers (Xbox and PS).  I would like to get it running with the Spectrum DX6i controllers we use for our real robots.  Assuming I can get them working, I will post a "how to" for this so people can use the real controllers.  For now, the simulator is intended to be 2-player multiplayer and 1-player vs AI.  Since I'm working in Unity this time, I would like to export to as many platforms as possible, potentially even to mobile and AirConsole, though I would need to think a lot about control scheme for those.

cjbruce:

--- Quote from: Tashic on July 14, 2017, 10:22:16 AM ---Oooo! Nuts and bolts too!

Nice ampflow model as well.

Loving the details :)

--- End quote ---

Re: Nuts and bolts - Of course! :)  It might be a little crazy to try to include every little screw that we use in a real robot, but right now I am exploring the limits.

I just received permission from AmpFlow to use models and the AmpFlow logo in-game.  I'm hoping to get permission from the other component manufacturers as well.  Hopefully everything ends up looking and feeling like the real thing.

The big problem we have is that we don't have our own arena, so students have to learn how to drive on the day of the competition.  Inevitably, this doesn't turn out well.  The goal is to give students driving experience beforehand so that they aren't completely befuddled by the controls when their robot is turned around in an odd orientation.

Tashic:
Thanks for the info!

I have tried as well to make a robot combat game, didn't go very far, but I did encounter the problem of the wheels, I used a method where I have the collision model of the wheel not completely fixed to the wheel itself, but I used a configurable joint to make it like a shock absorber radially, so it doesn't jump because of the not completelly smooth collision mesh.

But I'm sure you can find a much better and less messy solution than that!

I love the Idea of using the real controllers!

cjbruce:

--- Quote from: Tashic on July 14, 2017, 10:47:11 AM ---Thanks for the info!

I have tried as well to make a robot combat game, didn't go very far, but I did encounter the problem of the wheels, I used a method where I have the collision model of the wheel not completely fixed to the wheel itself, but I used a configurable joint to make it like a shock absorber radially, so it doesn't jump because of the not completelly smooth collision mesh.

But I'm sure you can find a much better and less messy solution than that!

I love the Idea of using the real controllers!

--- End quote ---

I think you might be overestimating my abilities!  :smile:

How much of an issue did you find the jumpiness to be?  Did you try increasing the number of sides?

I also found that adding sphere colliders as skids seems to reduce the jumpiness.  Apparently PhysX computes friction for every contact point, so as the number of contact points changes, the friction changes. Maybe this is the cause of some of the jumpiness?  Spheres only have one contact point, which I am thinking would smooth things out.

Also

Navigation

[0] Message Index

[#] Next page

[*] Previous page

Go to full version