Consider the case of an accelerating motorcycle, the engine converts potential energy into mechanical energy which is used to generate a Newton third law pair of forces at the rear tire's contact patch, a backwards force exerted from the contact patch onto the roar, coexisting with a forwards force exerted by the road onto the contact patch. To simplify things, assume there are no losses in the process, no drag, no rolling resistance, etc., so that any decrease in PE (chemical potential energy of the fuel/battery) ends up as an increase in KE (kinetic energy).
The "abstract" object in this case is the contact patch of the rear tire. Although there is no relative motion between the surface of the tire and the road at the contact patch (static friction), the contact patch itself is moving at the same velocity as the motorcycle (ignoring load related deformations). This is easier to visualize if you use the center of area of the contact patch as the instantaneous position of the contact patch.
The road can't generate power, but the point of application of the force the road exerts onto the contact patch moves with the same velocity as the motorcycle. So power could be stated as the force exerted by the road times the velocity of the point of application of that force, the contact patch, which is the same as the velocity of the motorcycle (assuming a flat road).
The road also can't perform work, but the integral sum of force(versus position of contact patch) times distance the contact patch moves, could be used to calculate the "work" done that originated from the engine.
I did a rethink on this. Power = force exerted on motorcycle · velocity of motorcycle. The fact that the road is not moving doesn't affect the road's ability to exert a force on the moving motorcycle, because it applies the force to the contact patch of the tire, where the tread is not moving with respect to the road, but is moving with the negative of the motorcycles velocity with respect to the motorcycle. Due to the rolling motion and the torque from the engine, the motorcycle's tire and wheel transmit the force from the road onto the rear wheel axis with the same force from the road and at the velocity of the motorcycle. The road could be considered as part of the power transmission sequence that uses the engine's power to accelerate the motorcycle.
In this case the velocity of the contact patch is the same as the velocity of the motorcycle, but consider a drum being angularly accelerated by a spinning tire, in this case the contact patch is not moving, but the surface of the drum is. The drum could be replaced with a cable that loops between two spools, so that the acceleration of the cable at the contact point is linear. In this case, the contact patch is not moving, and power = force exerted on cable · velocity of cable.
The fact that the tire surface is not moving with respect to the road at the contact patch is the reason that a non-moving road can apply a force to a moving motorcycle.
So what I refer to as an abstract object is just a way of referring to something that moves at the same speed as the object the force is being applied to, and was my attempt to deal with rolling motion of the rear tire in the case of the motorcycle.
The point of application of force being at the contact patch does have consequences, such as a wheelie if the acceleration is sufficient.
From a strict physics viewpoint, the interface between tire and road convert angular power (torque times angular velocity) into linear power (force x linear velocity), so no net work is done. However, it is common practice to state what the rear wheel horsepower is for a motorcycle, and this can be calculated as force times speed. This can be done using a chassis dynamometer, but it is also possible to determine force through torque sensors (tranducers), allowing rear wheel horsepower to be determined in real time while riding, and some riders buy the equipment that includes torque sensors and data recording for their (racing) track bikes.