I have a 2018 Honda with Lane Assist and Adaptive Cruise Control. I commute about 24 miles each day between the city and my home on a rural highway. I've also taken a few 2 hour late-night trips up and down the Interstate.
First off, it is awesome. It largely works remarkably well. As I mentioned, I've made a few of those 2 hour I-5 trips, albeit late at night when the traffic is virtually nil. But pretty much the only steering I did was a quick jiggle of the steering wheel to shut off the driver attention alerts.
It was utterly remarkable. The car autonomously drove itself for almost the entire two hours.
But it has its shortcomings. It is not a relatively simply engineering challenge. Roads have to be relatively well-marked for the Lane Assist to work well. LA gets confused when the Sun is low on the horizon and shining just right into the sensor. Heaby rain and ice often interfere with the sensors. Sometimes there's a weird marking or an on-ramp and LA wants to drive me into a barrier. I keep a mental catalog of the sections of the highways where this occurs. Also, the ACC sometimes like to brake way, way too hard or jerky. Most of the time it's good, but sometimes it really lays on the brakes.
Now, when I drive other cars, I feel minor annoyance at having to steer and maintain a safe driving distance. It's something that's super useful and an enhancement to driving. But I wouldn't dare trust it or take my eyes off the road.
A truly successful system would need a variety of sensor inputs (visual/mwave/ultrasonic/etc). You would need software that takes the data collected by all those sensors and correlates it into a bigger picture and then makes life-or-death decisions based on many, many highly-variable factors.