4-5 years seems optimistic to me, I think we're about a decade away but I expect it will definitely happen within our lifetime.

Humans are the cause of more errors and crashes than machines so I expect it will happen ultimately but both the liability and confidence issues are big ones. There's also an issue that the drivers sometimes aren't just drivers but also eg pack and unpack their vehicle - something a driverless vehicle may struggle with. A business I used to work with got overnight deliveries, the drivers has keys to the building and the alarm code and would park up overnight, unpack the delivery into the fridge and then lock up again and continue their route - no vehicle would ever be expected to do that.

Unions will also be an issue for many places - the technology for automated trains has existed for many decades (and trains are a lot simpler running as they do on tracks) yet despite that we still have Tube drivers on massive salaries. The Dockland Light Rail was built as a new thing driverless from the start but the Tube has never been switched over to being driverless as the drivers would go on strike to prevent it and it isn't possible to just switch overnight. New businesses and routes may find it easier to be driverless from the beginning than existing ones do - which in the private rather than public sector may result in some existing businesses going out of the business and being replaced by more economical competitors.