[hddqsb]

It's robust. I hamstrung the model solution by forcing pistonAcceleration to be 0, and forcing hingeAcceleration to be 0 during the first second, but it still "caught" and balanced the ball.