Live data from Hacker News

Just press go: designing a self-driving vehicle

googleblog.blogspot.com

121–130 of 322 posts

Re: Just press go: designing a self-driving vehicle

#121
post #81

Earlier quoted context omitted.

How about something simple like telling the car, "back up 10 feet" or "move to the spot across the street". This is dead simple short range routes...

How does one "tell" the car what to do without controls ? I hope you're not suggesting voice recognition because anything less than 100% accuracy will guarantee accidents and possibly deaths.

A smartphone app would probably suffice, given that I can control a drone aircraft with one already.

Re: Just press go: designing a self-driving vehicle

#122

Earlier quoted context omitted.

I think the eventual goal will be that most cars will be self-driving and that will negate most of those problems. I imagine that the vision of the future is a giant car-sharing program made up of self-driving cars all coordinated by Google. Think about it - you fire up an app to request a car, one appears at your door within a couple of minutes, takes you to work, then disappears. After work, a different car picks y…

I can't see that working. You'll require a ton of cars taking individuals or small groups to and from work in a big city. That's not very efficient, and the reason we have a metro system in the first place. 1. That's a lot of pollution. 2. You have twice as much traffic, since these cars need to come to your house every morning, and drive back to their charging station in the evening. They can take those trips off-pe…

I think you're underestimating the amount of pollution/resources that go into making a car. Reducing the overall number of cars produced would offset a huge number of those extra trips.

Re: Just press go: designing a self-driving vehicle

#123

This is going to change the world. No more car insurance. Vastly fewer driving fatalities. Shared cars. Less parking problems. I can't wait. Just take my money already.

I bet you'll still need insurance (or the provider will need it, in a service model).

It should be much, much cheaper though. Most vehicle accidents are caused by human error -- the insurance would just need to cover accidents caused by mechanical or software failure.

Re: Just press go: designing a self-driving vehicle

#124
I think Google could reach their goal faster if they started with more controlled environments - such as baggage vehicles at airports or mining trucks. There are plenty of people in my part of the world who are paid six figures to drive a truck up and down the same road of a mine all day.

Re: Just press go: designing a self-driving vehicle

#125
post #66

Earlier quoted context omitted.

Sure, Steer by wire is new. DBW generally refers to the connection between the accelerator pedal and the throttle. There, DBW has been the norm for many years.

Surely not the brake, though. Right? Right? Edit: http://en.wikipedia.org/wiki/Brake-by-wire

Mechanical failure was always possible with brakes.

Re: Just press go: designing a self-driving vehicle

#126
post #10

I think the biggest news was this sentence: If all goes well, we’d like to run a small pilot program here in California in the next couple of years. This is the next step in their very slow but, apparently, very steady march towards real self-driving cars that anyone can buy.

Why buy a self driving car when you could rent one as needed?

I never want to own another car. That is a huge amount of capital I could put towards Uber, or this.

Re: Just press go: designing a self-driving vehicle

#127

I feel as though they really havent embraced the driverless concept with their prototype. Why have seats that are fixed and facing forward? What about a couch? Why is there a dash? The people in the video are essentially watching the car drive itself, when they could be doing anything else. There are probably a multitude of reasons why the things above are the way they are. However, the prototype seems like just anot…

One of the cars my parents had when I was growing up was a VW Eurovan. The seat configuration in the car was a 'sofa' (fits 3) facing forwards and then 2 seats facing the sofa. On the left hand side of the car was a foldable table. As a kid it was fun to use it to play cards with the person in front of you or to eat, etc. etc.

0:50 over at https://www.youtube.com/watch?v=IBt-013d7Bw

I absolutely agree that cars can be much more fun when set up with different seating configurations.

Re: Just press go: designing a self-driving vehicle

#128
post #85

Earlier quoted context omitted.

I can't see that working. You'll require a ton of cars taking individuals or small groups to and from work in a big city. That's not very efficient, and the reason we have a metro system in the first place. 1. That's a lot of pollution. 2. You have twice as much traffic, since these cars need to come to your house every morning, and drive back to their charging station in the evening. They can take those trips off-pe…

1. Less pollution than what we have now given that many people spend time hunting for a park when they arrive, or drive five-seat vehicles even though they're alone or with 1-2 others. I read something saying that in a city centre, often 30% of traffic is simply people hunting for a park. I know when I arrive near my office, I can spend 5-10 minutes driving around trying to get an all-day park. Three times a week I'm…

1. Not sure I agree with the first point. People hunting for a parking space once a day likely causes less pollution then a car driving itself to a charging station or depot twice a day. Plus, if these are an affordable method of commuting, you're going to be pulling people away from public transportation, and getting more vehicles on the road. If they're more expensive, then you'll just have people buying their own, and then they're in your situation, where they buy an SUV model since it's more versatile.

2. Moving some depots outside the city will definitely help. Yes, some of the cars going into the city will be used for courier jobs, or tourist day trips, but that's an incredibly small percentage. You might have 500,000 automated cars taking people to work. 450,000 sit in a depot waiting for the evening commute, and 50,000 are used for people active in the day. Once again 500,000 people take their car home, and about 450,000 sit idle again waiting for the next day. How does this work as a business? If you use one of these cars for work, you need to pay for 90% of the cost, because that car rarely gets used outside of taking you to and from work. Sure, you could do a ride share with a couple of other people, but still, the few of you are renting that car full-time. It doesn't work.

3. If these can be solar charged, then surely a taxi can be solar charged, so there's no difference in fuel cost. You're paying a driver, and someone in an office answering calls, so there is an extra cost, and I think they have a use case here.

In short, these cars need to be on the road all day long, that's how they lower their prices and raise efficiency. That doesn't work for daily commuting. However, they would be an incredible taxi service. Imagine you have a fleet of 100 automated cars, people use an app to set their pickup/destination, and the cars find the most efficient way of organizing the routes and schedule. A car breaks down, and another one automatically reroutes itself. You could give people the option of carpooling with a checkbox. Let people register on the app, and upload a photo. So, you set your destination, and it says it'll be $10 and 15 minutes, or you can carpool with John (25/male), and it'll cost your $7.50, and 20 minutes. This could even happen while you're in transit. You get a pop-up on the dash. Want to lower your rate and share a ride with the person in this photo? Say yes, we'll credit $5 to your account, and take a 5 minute detour to pick them up. Worried about safety? You could review passengers, and only car pool with people that have a high rating and good feedback. I think that's exciting, it should offer lower rates to consumers, and you get to socialize with new people at the same time, so you might make some friends on your next taxi ride.

Re: Just press go: designing a self-driving vehicle

#129
post #48

Earlier quoted context omitted.

I really think that self-driving buses will be common long before most cars are. * They cover the same routes day after day * They are already centrally owned by large companies or governments * Driver cost is a big reason you can't have 2 small buses running twice as often as a single big bus. Or why buses can't run longer hours. * You still have limited space on roads and you need to use PT to get the person densit…

I think the driver on the bus has more functions than just being a driver. He's also an authority figure and can provide assistance.

There have been self-driving trains for some time now (Nuremberg, London Docklands). There are more people to be responsible for on a train and I do not see why a bus would be a special case.

Re: Just press go: designing a self-driving vehicle

#130
post #124

I think Google could reach their goal faster if they started with more controlled environments - such as baggage vehicles at airports or mining trucks. There are plenty of people in my part of the world who are paid six figures to drive a truck up and down the same road of a mine all day.

I know there are open-cast mines that already use remotely controlled/monitored and self-driving trucks. I think using it in an environment that they're used to, and which I presume would be most practical, makes sense. It is more helpful for them to research and build their technology where it will mostly be used, on the streets. There is more volume of traffic, which makes it possible to deploy 100 small cars cost-wise, and could lead to 'quicker' development as they won't always be slowed down by legal and other negotiations with airport companies and the like.

Just my thoughts anyways :)

Post reply on HN