4 Comments

Thanks Phil. Few day back a colleague brought this stats for acceptance criteria and validation target discussion. Then, I try provide a few analogy of the dashboard stats.

- 84% fewer airbag deployed - These robotaxi operate major of the mileage without occupancy. In that case regardless the speed airbag doesn't deploy if no passenger detected.

- 73% fewer injury causing crashes - same as above along the explanation from F perkin.

- 48% fewer police reported crashes - many crashes remain unreported as no direct driver involvement or any severe damage due to low speed and defensive driving.

These are purly misleading numbers. It is advisable to use meaningful justifiable stats relevant to the use case.

Expand full comment

I appreciate the thoughts Abhash. As you observe a really meaningful apples-to-apples comparison is quite difficult.

Along the lines of your airbag point, it is also expected that robotaxis will have a 3x advantage on occupant injuries because there is no driver inside the vehicle to be injured -- even if they have the same crash profile as human driven vehicles. See: https://philkoopman.substack.com/p/why-robotaxis-have-an-inherent-3x

Expand full comment

Thanks.

Perhaps also worth noting that Waymo

operations to date have all been at low speeds which are only rarely associated with serious injuries or fatalities. Waymo has recently received permission to operate on freeways in San Francisco which will quadruple their (inherently hazardous) kinetic energy and reduce time available for adverse event reaction. Good luck to Waymo and the tens of thousands of unwitting vulnerable test participants in this poorly bounded dangerous experiment.

Expand full comment

Thanks for the comment Fred. As you know (but perhaps not everyone else knows), fatality probability per crash increase dramatically for speeds over about 20 mph. Indeed highway operation for robotaxis is going to be a whole different ballgame for safety.

Expand full comment