Time spent in the car has an opportunity cost. Those 20 minutes spent driving to work could be 20 minutes more sleep, or breakfast, or any number of useful things. The value of commuting might be a cheaper home, or a better workplace, or the ability to live (or work) near your friends and family.
Self driving cars don't change the opportunity cost, but they do add to the value of commuting. Suddenly, you get your entire commute to do something else in because the car drives itself. To make things even better, it reduces the cost of commuting by removing the need for car ownership (self driving cars will work like uber).
For people who know a little about microeconomics, this has a similar effect to shifting the demand curve up. And as the system equalizes, that 20 minute commute might become 30 or 40 minutes.
I expect this increased congestion to accelerate the transition to self driving cars once we see the first cars on the road. After all, the person in the self driving car is both spending less and getting more done than the driver.