The main reason that retroreflective paint is successful is that the reflected light is much brighter than normally dispersed light. In the case of the driver in his own car, the angle formed between the headlamps, paint, and driver is of a limited range. It is much easier to predict how the beam should reflect in order to be seen. In the case of the driver opposite him, there is a much wider range of angles that the beam can take. Also, any light redirected in this way would subtract from that directed to the original driver.

Perhaps phosphorescent paint is in order here? Maybe some way of designing the paint so that it can store energy, and release it as brilliant light when exposed to some form of automotive stimulus?

After having studied driving itself, and vehicle dynamics, I think we should be focusing more on improved driver training and removing poor habits than modifying the roads to suit horrid drivers. Everyone feels they are a good driver - and very, very few people are. Everyone who drives should be able to explain basic driving concepts, such as shifting a car's balance explain the traction circle. People complain about speed limits without understanding the technical features of a speed limit, or how traffic flows down a road.

I think the problem you are trying to address exemplifies this perfectly.

With proper technique and practice, you can see lanes far down the road in both darkness and heavy rain simultainiously. This is with no reflectors, and actually extends farther out than the retroreflective paint's useful range.

See RetroReflectors. Something might work as you intend. But I don't think sufficient incident light is available to reflect towards oncoming traffic.