Kinematic Joint Error Message

[SpiderPig]

The code below gives 3 different error messages (if you press ignore each time).  Here's the first one;

Only get the errors in debug mode.  In release mode the object is shot away at speed.

#include "UltraEngine.h"

using namespace UltraEngine;

int main(int argc, const char* argv[])
{
    auto displays = GetDisplays();
    auto window = CreateWindow("Ultra Engine", 0, 0, 1280, 720, displays[0], WINDOW_CENTER | WINDOW_TITLEBAR);
    auto framebuffer = CreateFramebuffer(window);
    auto world = CreateWorld();

    auto camera = CreateCamera(world);
    camera->SetClearColor(0.125);
    camera->SetPosition(0, 0, -8);
    camera->SetDebugPhysicsMode(true);

    auto light = CreateDirectionalLight(world);
    light->SetRotation(35, 35, 0);

    auto box = CreateCylinder(world, 0.5f, 1.8f);
    box->SetMass(1.0f);
    box->SetColor(0, 1, 0);

    auto floor = CreateBox(world, 10, .1, 10);
    floor->SetPosition(0, -2, 0);

    auto joint = CreateKinematicJoint(box->position, box);
    joint->SetMaxForce(100);
    joint->SetMaxTorque(100);

    auto pivot = CreatePivot(nullptr);
    
    float a = 0, y = 0;
    while (window->Closed() == false and window->KeyDown(KEY_ESCAPE) == false)
    {
        pivot->AlignToVector(Vec3(0, 0, -1));
        auto quat = pivot->GetQuaternion();

        joint->SetPose(Vec3(0, y, 0),  quat);

        world->Update();
        world->Render(framebuffer);
    }
    return 0;
}

 

[Josh]

It seems to be because the target rotation is 180 degrees away from the current orientation. Maybe the solver can't decide which way to turn?

Line 337 of dgCustomKinematicController.cpp:

dAssert(lateralDir.DotProduct3(lateralDir) > 1.0e-6f);

Simplified example:

#include "UltraEngine.h"

using namespace UltraEngine;

int main(int argc, const char* argv[])
{
    auto displays = GetDisplays();
    auto window = CreateWindow("Ultra Engine", 0, 0, 1280, 720, displays[0], WINDOW_CENTER | WINDOW_TITLEBAR);
    auto framebuffer = CreateFramebuffer(window);
    auto world = CreateWorld();

    auto camera = CreateCamera(world);
    camera->SetClearColor(0.125);
    camera->SetPosition(0, 0, -8);
    camera->SetDebugPhysicsMode(true);

    auto light = CreateDirectionalLight(world);
    light->SetRotation(35, 35, 0);

    auto box = CreateCylinder(world, 0.5f, 1.8f);
    box->SetMass(1.0f);
    
    auto joint = CreateKinematicJoint(box->position, box);
    
    while (window->Closed() == false and window->KeyDown(KEY_ESCAPE) == false)
    {
        joint->SetPose(Vec3(0, 0, 0), Vec3(0,180,0));
        world->Update();
        world->Render(framebuffer);
    }
    return 0;
}

 

[Josh]

Reported here:
https://github.com/MADEAPPS/newton-dynamics/issues/321

[Josh]

I tried commenting out the assert statement. In my code below, any rotation closer to 180 causes the joint to just not react. I suppose this is better than crashing, but still not ideal.

#include "UltraEngine.h"

using namespace UltraEngine;

int main(int argc, const char* argv[])
{
    auto displays = GetDisplays();
    auto window = CreateWindow("Ultra Engine", 0, 0, 1280, 720, displays[0], WINDOW_CENTER | WINDOW_TITLEBAR);
    auto framebuffer = CreateFramebuffer(window);
    auto world = CreateWorld();

    auto camera = CreateCamera(world);
    camera->SetClearColor(0.125);
    camera->SetPosition(0, 0, -4);
    camera->SetDebugPhysicsMode(true);

    auto light = CreateDirectionalLight(world);
    light->SetRotation(35, 35, 0);

    auto box = CreateCylinder(world, 0.5f, 1.8f);
    box->SetMass(1.0f);

    auto joint = CreateKinematicJoint(box->position, box);

    while (window->Closed() == false and window->KeyDown(KEY_ESCAPE) == false)
    {
        if (window->KeyHit(KEY_SPACE))
        {
            joint->SetPose(Vec3(0, 0, 0), Vec3(0, 178.6, 0));
        }
        world->Update();
        world->Render(framebuffer);
    }
    return 0;
}

 

[Josh]

Newton forum is back up:
http://newtondynamics.com/forum/viewtopic.php?f=12&t=9825

[Josh]

As I suspected there is no perfect solution when the target rotation is 180 degrees opposite from the current rotation, because one degree of freedom is lost. I'm not sure at this point what should be done or if this is just the way things work.