Governing Autonomous Vehicle: Heaven, Hell and Who Decides

The first thing I will say is how to merit the right thing.

This is a graph in San Francisco, top of the growth of Waymo.

It increased from below 100,000 trips to 700 trips within a year, like a tenfold increase.

But more importantly, to look at this is the passenger onboard miles.

and the red part is empty miles, and this is in route pickup miles.

This is the critical aspect about the vehicle occupancy.

That's the foundation for the discussion of the efficiency.

I like to suggest this metric, that is the passenger mile to vehicle mile ratio.

If you consider total passenger mile, If you have a share of two, if your bus is 20, it's a contribution to society.

And the total vehicle miles, whether it's deadheading with people or not, this is the total cost to society.

If you took this P2V ratio, that's a metric to measure the overall efficiency.

So we hope that all EV service providers, Uber, buses, planes should all provide this ratio and have a fair comparison.

In terms of PVR, can AV be better than each of those current services? Be better than private ownership, Uber, dial-right, suburban buses, urban buses, or subways? And for specific location and time, that would be a good metric to check.

Does it make sense for AV to play a positive role here? We can carry this search ratio today, but also if Waymo were to 10 times the operation, let's say, they're moving mostly from the downtown core to a more suburban, lower density operation, would the search ratio decrease or increase? Can it maintain a good, efficient operation? Point number two, what's the difference between Waymo and Uber? Waymo, where you can only incentivize behavior through pricing, through information.

But in Waymo, you can dictate behavior.

If we have 100% of control, what can we do? We've been discussing this dynamic bus lane for a while, but we just couldn't enforce it.

But with AV, potentially we can have a meaningful dynamic bus lane.

The lane used by bus, when bus is a big gap, you can insert the AV operation in it.

Second one is the precise pickup and drop-off location management.

Next one, dynamic parking.

It's been discussed for a long time, but we can only implement to a very limited degree.

With AV, can we introduce a series of smart dynamic pricing? We introduced many of the speed limits, particularly in this highway system, but what caused accident at one speed, but the other important one is the speed variation.

So with AV, it's possible for us to regulate Speed variation, in addition to speed limit, that's another possibility that is never enforceable, implementable today, but with new technology, can we imagine that? Last one is AV as a traffic controller.

by treating AV either as a moving traffic light or AV as a policeman to regulate the overall traffic if we have a fleet-based AV.

With AV potential, we don't need congestion pricing.

The first thing I will say is how to merit the right thing.

This is a graph in San Francisco, top of the growth of Waymo.

It increased from below 100,000 trips to 700 trips within a year, like a tenfold increase.

But more importantly, to look at this is the passenger onboard miles.

and the red part is empty miles, and this is in route pickup miles.

This is the critical aspect about the vehicle occupancy.

That's the foundation for the discussion of the efficiency.

I like to suggest this metric, that is the passenger mile to vehicle mile ratio.

If you consider total passenger mile, If you have a share of two, if your bus is 20, it's a contribution to society.

And the total vehicle miles, whether it's deadheading with people or not, this is the total cost to society.

If you took this P2V ratio, that's a metric to measure the overall efficiency.

So we hope that all EV service providers, Uber, buses, planes should all provide this ratio and have a fair comparison.

In terms of PVR, can AV be better than each of those current services? Be better than private ownership, Uber, dial-right, suburban buses, urban buses, or subways? And for specific location and time, that would be a good metric to check.

Does it make sense for AV to play a positive role here? We can carry this search ratio today, but also if Waymo were to 10 times the operation, let's say, they're moving mostly from the downtown core to a more suburban, lower density operation, would the search ratio decrease or increase? Can it maintain a good, efficient operation? Point number two, what's the difference between Waymo and Uber? Waymo, where you can only incentivize behavior through pricing, through information.

But in Waymo, you can dictate behavior.

If we have 100% of control, what can we do? We've been discussing this dynamic bus lane for a while, but we just couldn't enforce it.

But with AV, potentially we can have a meaningful dynamic bus lane.

The lane used by bus, when bus is a big gap, you can insert the AV operation in it.

Second one is the precise pickup and drop-off location management.

Next one, dynamic parking.

It's been discussed for a long time, but we can only implement to a very limited degree.

With AV, can we introduce a series of smart dynamic pricing? We introduced many of the speed limits, particularly in this highway system, but what caused accident at one speed, but the other important one is the speed variation.

So with AV, it's possible for us to regulate Speed variation, in addition to speed limit, that's another possibility that is never enforceable, implementable today, but with new technology, can we imagine that? Last one is AV as a traffic controller.

by treating AV either as a moving traffic light or AV as a policeman to regulate the overall traffic if we have a fleet-based AV.

With AV potential, we don't need congestion pricing.