The outspoken CEO did caution that regulators may then take between one to five years to approve the use of such self-driving technology. During that time, the company would operate its autonomous driving software in "shadow mode", collecting data on what the car's computers would have done compared to what human drivers do.
Musk believes that once it becomes "statistically clear that an autonomous car is safer, I think, regulators will be comfortable with allowing it".
Audi's Keogh believes that autonomy will come in stages, with "completely gridded, completely mapped" urban areas where "all the data and intelligence will be". There will also adaptive cruise control systems, like the ones seen on many high-end luxury cars today, that can operate in stop/start traffic and keep to the current lane.
He noted that interpreting human behaviour during low speed driving may be one of the toughest technical problems for those engineering self-driving software. Keogh cited as an example the four-way stop sign that's common sight in the US:
"We all know there’s a rhythm to how you get across that thing, you stick your nose out a little more, you make eye contact, there’s a common understanding, and then off you go. If the system literally tries to do it by the letter of the law ... you might be there a couple of hours while people are all edging in."