So people who can afford a self driving car will be subsidized by people who can't.

Who will put those magnets in the road first place? Here where i live there are roads that do not even have pavement yet

Slowburn, 'tis always the way with infrastructure projects.
One problem with all these upgraded super road ideas, is that they are exceedingly easy to sabotage.
Either magnets or metallic objects, embedded in the road surface (or sub surface) for detection by the vehicle's on-board sensor suite, similar objects can be implanted in the road by nefarious individuals.
Note that using a detectable metallic object on the road surface, will be much cheaper to deploy, than placing billions of ferritic magnets across the national grid. In the real world, any blind sensing system must be augmented with machine vision (visible light and-or additional bandwidths).
The technology of a line following robot has been around for decades, the reason why this hasn't been accepted anywhere for vehicles on public roads is the unreliability of a non-intelligent system.
VIsion, (TLDR if you have ADHD etc) We are getting there. For countries with cats-eye markers on the lane dividers, placing an RFID tag in the cats-eye provides for a semi-secure means to achieve the affect of a reliable lane holding system for widespread use (reliable in vision denied conditions- using sensor fusion), combined with cameras, GPS and IMU, if a few of the markers are missing the system will carry on along the planned route until the next reliable marker is detected... Leave the implants on the lane lines, hey if a line + RFID tag (cents per tag, most likely cheaper than a magnet) on either side of the lane isn't good enough where have we come (not) with technology over the last 20 years.
Of course the use of lanes at all, is also a bit on the old-tech side, why not have a more organic approach to traffic, with collision detection and avoidance algorithms , local area state and intent broadcasting (similar to ADS-B) and global automated traffic management broadcasting (like automated ATC), the ability to manage traffic flows in an on-demand manner is becoming more feasible..
One hold-up, we all need to be driving semi-automated or fully autonomous cars / personal-public transport. In aviation, if an aircraft does not have adequate radar transponders on-board they are not permitted in certain classes of airspace. This will eventually happen on certain roads, where hopefully automated traffic control and operation will mean free'r flow and safer commutes and long distance travel... Pipe dream, probably, maybe we all should just fly, it is simpler in the air.

Yep I can see why Volvo drivers do not want to drive, but I actually enjoy driving, so I always want to control the car myself. Plus I trust my own judgement and motor coordination (pun intended), much more than a complex system of programming and infrastructure.

In 1980, I worked in Houston, Texas at Texas Instruments. We had this robotic mail cart that used a magnetic painted strip on the carpet to know where it was. We use to take great pleasure in taking our shoes, scuffing up the carpet, rolling a screw driver magnetizer around to screw up the mail cart, which would STOP if it didn't sense the magnetic strip in the carpet. Security & the mail room people didn't think it was funny, but we did.

@MD Very good idea with the rfid tags. They could come in specific types that responded to requests from an on-board receiver with the number of the lane they were marking. Instead of putting them in cats-eyes, they could be put in the centre of each particular lane. That way they could be applied anywhere, even in car parks, which might help a few people who struggle to park their cars correctly.
Taking the rfid idea further, I assume that it is not beyond the wit of humankind to have special road maintenance plant that formatted the rfid tags with not only their lane number, but also their location via GPS. these could be driven along specific lanes, guided by GPS and applying the tags with a dab of asphalt or the like every meter or so. This could be done on roads at any time when weather conditions allowed, not just when the surface is being relaid. Given the location content, calculating speed would be easy and thus provide control information regarding speed limits and when approaching light controlled junctions that were on red or about to do so and the vehicle needed to slow down. (Result: no speed cameras, no red-light cameras, thus stealth taxes in the form of fines for speeding and crossing on red.)
Can't say I like the idea of self-driving cars, but we do have an aging population and we have a society that has put the car at its centre. Those two facts have to match a lot better than they do at present. Any technology that makes driving safer has to be a good thing for all of us surely and would enable the elderly to be mobile to an older age. With any luck, it might also be possible to relax the drink driving laws if cars could be made to deny people the opportunity to indulge in dangerous behaviour, no matter how brave their intoxicated state of bravado made them feel.
People could then return to going out for the evening and enjoy a couple of beers or half a bottle of wine without worrying that they might lose their licence. Or have I got it wrong and enjoying life is not what it is all about. There would, of course, be one rather large dividend for the owners of such vehicles: their car insurance would reduce by leaps and bounds.
If people insist on trying to prove that they could be F1 drivers if they could only get the chance to prove it, let's open up the racetracks for them at weekends, say. We could put on extra staff at the local A & E and warn the undertakers in the area that it might be wise to have someone on hand should their services be required. (Expenses could be met by selling tickets to watch the spectacle.) Such an arrangement could prove to be one of nature's better self-regulating feedback loops.

In other words, it is not self-driving, it's a rigged demo on a customized track.

kids of tomorrow will get a thrill tossing magnets onto the roadway, they can cause as much damage as tossing pennies onto rail road tracks or putting rocks on them

This magnet-based vehicle automation technology has been around for almost 20 years. I worked on it in grad school starting in 1996. UC Berkeley's PATH program automated the cars and UC Davis' AHMCT program built a diagnostic system to test the magnets and their polarity for any damage (on an autonomous car driven by a vision-based system developed by Lockheed Martin). Seems funny that Volvo is looking at it now. Costs for magnet installation were very high; perhaps they've gone down. See here for video's of the cars in action in San Diego in 1997. http://www.path.berkeley.edu/PATH/Research/Demos/
After that they put the magnetomers on a Caltrans snowplow and my partner and I developed a radar-based collision warning system and an in-vehicle display showing your position on the road by reading the magnets to enable driving in white-out conditions. The snowplows were never automated, though. Man, this takes me back! http://www.path.berkeley.edu/PATH/Research/snowplow/

Hahaha! No! This is dead wrong on so many levels!