Ignoring mass from welded joints in box2d

Question

What I have is a circle that rolls over other objects and 'picks' them up. My current setup uses WeldJoint to join and place the picked up object at the circle's edge. I then edit these object's fixtures to become a sensor to avoid collision response. The circle is pushed around using applyForceToCenter() and a slight impulse to get it going. This all works.

The problem is that the the accumulating mass from the new bodies is limiting the main circle's movement. Too realistic for my game. The bodies keep tugging the circle around. I want to zero out any physics-based attributes affecting the circle.

I've tried a variety of methods using resetMassData(), MassData.mass = 0, setAwake(false), setActive(false) ...

Any help is appreciated, thanks.

Edit: I've thought about using a math function to spiral the object's around the circle, e.g. Make a sprite go in a circle. I'm not so sure if that's the route to go with box2d in hand.

Answer 1

When you are using weld joints, you aren't deleting any of the bodies (which have their own mass) as you attach them to your circle. Since you are content to simply assume that once something becomes stuck to the circle that it no longer needs to collide with anything, a good solution would be to add them as sensors to your circle body. This will require you to delete the body and add the fixture to your circle.

I think it would be cool to have more objects result in more mass for your circle. Why not make a routine that determines the total additional mass currently stuck on your ball, and scale your ball's push force accordingly? You'll end up with the same acceleration but you will get a realistic increase in collision forces generated for free. Not only that, but your welded sensors will also effectively modify the inertia so your ball's center-of-gravity will now reflect how things are stuck on it! It's free realism, so I'd try to take advantage of it. Plus it will be easier than writing code to correctly move the fixtures.

You could also in addition reduce the density of a newly "stuck-on" object by some factor to reduce the effect as necessary.