Blog Waymo Testing Vehicles Without Human Drivers

[vc_row][vc_column][vc_column_text]Waymo, the autonomous vehicle division of Alphabet, Google’s parent company, says it has been operating a fleet of autonomous vehicles in Arizona without a human behind the wheel since mid-October.

Soon, the company plans to invite regular people for rides in these fully self-driving vehicles.

The cars will be geofenced within a 100-square-mile area of the town of Chandler, a suburb of Phoenix. But, Waymo plans to expand the number of cars in the fleet and the range to an area larger than Greater London over time. The company is asking people to apply to be part of an “early rider program.”

Arizona has no restrictions against operating an autonomous vehicle without a person in the driver’s seat. California, however, requires self-driving cars to have a safety driver in the front seat.

Waymo shared the milestone in a Medium post and video, posted below.[/vc_column_text][/vc_column][/vc_row][vc_row][vc_column][vc_video link=”https://www.youtube.com/watch?v=aaOB-ErYq6Y” video_title=”1″][/vc_column][/vc_row]

 

[aWalkingShadow]

Their use of AI is somewhat limited, they concentrate on having good sensors and algorithms to decide what to do.

On what basis do you say that Waymo's AI is limited? Doesn't the fact that they are able to train it by running simulations indicate that their car's ability to learn is actually relatively sophisticated and well-developed?

 

[banned-66611]

Based on what they have said publicly about how it works, they use some AI for recognition of images but it's mainly for things like signs and traffic lights. They use lidar for most of the environment mapping and object detection.

As for decision making they use algorithms. I think Tesla will too, AI just isn't suitable. For example, Google do things like avoiding sitting in other car's blind spots and predicting the actions of pedestrians, and it's all programmed behaviour.

That gives you some idea of how far Tesla need to go... Or maybe not. Tesla may decide it's okay to mow down pedestrians who pick a bad spot to cross the road, because it's not their fault and at low speed they have emergency braking.

 

[banned-66611]

I wasn't even joking, Musk said that the threshold is "as safe as a human driver" which is way below what Google and probably regulators will accept.

 

[electracity]

It's unclear to me if the whether the primary hurdle will be regulation or insurance. What is the liability framework in place to allow Tesla to tell drivers that its cars are now level 3?

Waymo at level 4 is straightforward. Waymo is responsible. Level 3 is very messy. Tesla needs a camera pointed at the driver so they have the option of holding the driver responsible in an accident.

Tesla not only needs to avoid liability, but the burden of being a defendant in every serious accident involving a Tesla.

But the even bigger problem for Tesla has yet to emerge. Who wants their family members stepping in front of vehicles with autonomous vehicles with Musk's half-assed sensing hardware circa 2016? It won't happen, and it should not happen.