I'm surprised to see Toyota have someone with such a defeatist attitude in charge of their self driving endeavors.
As they say, perfect is the enemy of good. Toyota is about taking something and making it cheaper, faster etc. They are lot more cautious in coming out with new innovative products as they want to be perfect. They cannot.
With driver-less cars, the question is usage conditions. Having autonomous driving in a geo fenced area of Phoenix, where streets are broad, weather conditions are usually good and traffic is manageable, is relatively easier, compared to say autonomous driving in Sao Paulo or Shangai or New Delhi or Tel Aviv. Here, traffic may be more chaotic, streets narrower etc. To this we can add weather and conditions such as blizzard and weather conditions, unmarked roads, construction delays, special events etc.
So we have standard conditions and non standard or location specific conditions. Now Waymo and other are going for standard conditions, not the boundary conditions. Toyota is going for the entire enchilada, they are looking at all and every condition. And I agree that is correct, that we are probably far away from there. So I would be hesitant to buy a car where manual intervention is not possible. However we cannot wait till every possible condition is covered. You have to begin somewhere, even if it covers say 80% of the conditions, rather than 100%. Waymo may be at say 95% in the Phoenix area, but may be only 50% elsewhere. That has not stopped them, they are starting where they can. (Even though Waymo is offering completely driver-less service in my area of the Phoenix suburbs, they can pull their service at any time i.e. if there is a dust storm like we sometimes have here, they can stop their service or only have a manned service where the driver can take over.) My take is that Toyota wants to be at 99% or over (a sure thing). Unfortunately, you have to get to that through machine learning, which may start at 0% and work yourself up to wherever.
Now I do not disagree with Toyota that the cost of the additional hardware in the car and software (additional complexity and compute power) to get to say 99% or higher is going to be expensive. However that does not mean that you cannot provide hardware and software for say 80% of the conditions fairly inexpensively and provide more capabilities in a sub-set of cars that are constantly used.
Both can co-exist (some cars at 80% and other at 99%) and but Toyota seems to wants to stick to what they are comfortable with. Let us remember, software development is not Toyota's skill or for that matter, for most automotive manufacturers. Developing some embedded software code is far cry from developing a complex system. Tesla has software roots to begin with and it was easy for them. Google is a software giant. GM bought Cruise and allowed others like Honda to invest in it, as Cruise is a Silicon Valley start up. Others like Nissan turned to Mobileye (which is now owned by Intel a hardware giant) or to Nvidia. Toyota on the other hand is not comfortable with working with vendors they do not control. That is why they are struggling as this is outside their comfort zone. That said I agree with the larger statement that we are still far a ways from all conditions autonomous driving, but most manufacturers need to get to L3 at least. Toyota is behind the curve.
