The cone collocated on two inclined rails doesn't want to roll down. Instead if we leave it on the basis of the rails, it will roll to the top by itself.

In reality, during the motion to the top, the cone advances in the direction where the rails diverge, leaning on its always thinner part. By an appropriate choice of the rails' aperture angle and of the cone's angle, we obtain the motion of the centre of mass and the apparent paradox disappear.

A similar case are those "miraculous" roads where cars ascend without engines. If we look carefully at the slope, we realize that it's just an illusion.

Let us view the cone in the cross-section and the rails from above. The aperture angles of the cone and the rails are chosen in a way that allows for the gravity centre of the cone to be positioned higher at the beginning of the rolling up movement than at the end of it. Thus, it is essential that the difference in height between the start and finish points of the rails should be greater than the radius of the base of the cone.

Note that, when it is approaching the end of the v-track, the cone rolls faster and faster as the radius gets shorter towards the tips. Although viewed from the side the motion of the cone is accelerated, it is slower compared to the movement of sphere going down an inclined plane - similarly as it is in the case of Maxwell's pendulum a considerable amount of potential energy is transformed into the spinning motion energy.