I can’t point you to a checkbox that will solve everything, but I think my solution will be faster than what you’re doing now.
Start with a collision shape that is slightly larger than your ship. Then, on tick, add (velocity * elapsedSeconds) to the dimension(s) of the collision shape. That way, the faster you’re going, the larger the collision shape around the ship, and thus the earlier the collision will be detected. I take the (velocity*seconds) value and multiply it by some small amount (like 4) to add a little buffer room.
The easiest thing is to grow all three dimensions by the length of the velocity vector. That would be good for an initial test, but with some math you can grow the shape along individual axes based on the actor’s rotation and velocity along that axis.
On collision shapes, the only way to not use a standard collision shape is to use ‘complex’ collision. I’ve never seen that work right though and it sucks up CPU like an infinite loop.