Hacker News Reader

Comment by jitl | Hacker News Reader

I think these are two companies in very strong positions:

- Hyundai/Genesis has some of the best cars on the market right now between the GV line and Ionic lines, and they seem to be improving faster than other manufacturers

- Waymo is now the clear leader in self-driving in my mind after visiting SF recently and opting for Waymo driver over human driver 60% of the time.

I have a Mercedes on a short lease now but my hope/expectation is that in 3 years Hyundai will be making my ideal car since the GV80 is so close, but I’d like to see an electric-first design from them in the space.

I wish the Waymo deal meant I could buy a Waymo-equipped Hyundai myself though :-(

View on HN

Living in San Francisco, I feel much safer as both a pedestrian and a cyclist when I see a Waymo at crosswalks vs. human drivers. I was hit by a car once while cycling because the driver zoned out, realized too late they were about to miss a turn, and struck me while attempting a last-minute turn. Since then, I’ve become hyper aware whenever drivers are on their phones at stop signs or play pointless games of chicken, trying to beat red lights just to end up stuck blocking the middle of the street.

When I see a Waymo, it feels like it's aware of me and always stops at a safe distance, and I don't fear a collision. It doesn't give off the hostile or dangerously clueless vibe I often get from human drivers in the city.

I must say that I usually try to keep the screen view of the 3D world perceived by the Waymo Driver.

I prefer to sit in the front passenger seat. Once I've programmed my music and adjusted climate control, I sit back, relax, and watch the little icons of vehicles, pedestrians, roads and lanes, and the immediate route plan ahead.

The icons include pylons detected, as well as cyclists and emergency vehicles. There are some rather ghostly fades in/out at the peripheries, but it's clear that the Waymo Driver has perpetual 360° situational awareness.

This weekend, I'm celebrating 1 year with Waymo: 144 rides and 805 miles logged.

The Jaguar I-PACE is very comfortable, includes 2x USB-C charging ports, electric reclining front seat, plenty of cargo in trunk (and no cab driver storing his stuff) so I look forward to how the Hyundais measure up.

The intent is to eventually make this available to own

https://www.youtube.com/watch?v=sCY214KiwNs&t=1329s

They may say that, but OpenAI also said they're not for profit. They say the darndest things.

It seems to make much more sense to have a fleet owned and operated by a corporation. I actually doubt that any company that achieves proper self-driving will sell vehicles. Renting makes much more money, and corporations like money. Do you see Google selling their search engine as software that you can deploy on-premise? Do you see Amazon selling the internet store software? Microsoft was selling Windows and Office, and what do we see? They are trying to move as many customers as they can to the rent model. Logitech mentioned subscription for a mouse.

Nobody will sell self-driving cars, if they can make them.

>Do you see Google selling their search engine as software that you can deploy on-premise?

They did, until pretty recently.

https://en.wikipedia.org/wiki/Google_Search_Appliance

There are going to be a ton of really powerful network effects with large scale robotaxi networks. Rideshare network effects are already pretty strong and the startup costs are much lower than a company having to build and maintain hundreds of thousands or millions of vehicles. The first team with commercially viable wide scale self driving might not jump straight to consumer sale, but the second one will, and if not them the third. On top of that, anyone who wants to scale quickly will have to either be an automaker or play ball with automakers who will have little interest in putting all their autonomy eggs in someone else's basket. Even if an automaker does want to go all-in they will have to contend with dealers who are an extremely powerful political force and who would have a very diminished role in a rental only future.

I think its much more likely that we see premium cars for purchase with L2 and advanced safety features at no additional charge and a yearly subscription model for L3/L4 like we see with hands-free and L3 tech from Ford, GM, and Mercedes right now. Even so Ford now offers a one-time purchase option for Blue Cruise so maybe consumers are going to push back on that model as well.

What's scary so how networks externalize the costs while controlling the profit.

Ideally the internet could be host to actual marketplaces, versus the up-in-the-cloud data-keepd that dictate to the down-below dwellers.

Well they are not saying it’s a non profit, self driving cars will very much be for profit, a lot of it.

I'll be honest, I think they're just saying this so as not to spook anyone. But really I think they plan to build a massive, super convenient robot taxi service across the country and by the time their self driving tech is "ready" for sale that service will be completely ubiquitous.

There is still some value in having a dedicated/always available option like car owners are used to. I live in a car centric city and maybe drive somewhere on average of 8 times a day (it's probably 4 on weekdays and maybe 20 on weekends). Uber is always near but still usually takes a minimum of 8-10 minutes and probably an averages ~12-15 minutes in my experience before I am entering the vehicle after summoning it. You can do the math and see that I would spend hours a day waiting in current world. If every vehicle was a Waymo then sure they'd always have one on my block at the ready, in theory. But in reality, a business will eventually seek to maximize efficiency and this will lead to a sucky UX.

People also use vehicles as cargo carriages, not simply human transport devices. I know I don't always immediately unload my vehicle when I transport things, sometimes it gets stored there. Sometimes the load time and the unload time don't even make sense to do back to back. Sometimes I load something knowing I will unload it next week. Some things I just carry around with me on the off chance I might need them while out and about. Anyway, there's a lot of nuance to this type of change and why owning a car might still make sense for a lot of people.

I would expect self driving cars to be a lot quicker and more available than uber.

Plus lets be honest Americans love their cars, a lot of them are never going to give them up on the principle of the thing, but their kids are likely to reach an age where they would need to learn to drive and just not see the point, self driving robo taxi's will be good enough.

You get in and out of your car 20 times a day on Saturdays? What hell is that?

Probably. I’m guessing. But it equates to 10 destinations.

3-4 are kid activities sports/birthdays/play dates (we only have 1 kid but he’s a busy guy!)

2 weekly dog grooming appointment (doodle!) have to drop her off then pick her up later, so it’s counts as 2

2 we eat out twice on Saturday and once Sunday

Rest are other social, family visits, or hobby activities along with a plethora of random errands; groceries, home center, and a long tail of specialty stores that are individually irregular but always a few every weekend (pool supplies, makeup for my wife, post office/UPS, etc.)

We shop online so mostly avoid malls and department stores unless it’s more of a boutique thing or something we want to see before buying. We don’t do food delivery or groceries unless it’s just curbside thing which is another trip. I hate the online/app shopping experience for things like groceries.

Sometimes I just drive half a mile to get a Big Gulp fountain drink or an ice cream cone with the kid. I could walk or bike, but it’s car centric and 100F outside.

It’s also a huge range and varies a lot by season. We certainly have some weekends we barely leave the house.

That sounds exhausting. I've got 3 kids and a dog and don't go out in the car nearly as much as that. 5 or 6 times a WEEK! I don't know how you have time left for the actual hobbies. We do though, do most shopping online.

> a long tail of specialty stores that are individually irregular but always a few every weekend (pool supplies, makeup for my wife, post office/UPS, etc.)

I always save stuff like that up so I don't need to make an extra trip out

It can be. When I’m exhausted, I take a trip and it involves lots of car rides and flights too.

Hobby time has suffered the most. My philosophy is spend as much time as possible with my kid while he actually wants to be around me. So that’s my hobby now. But I’m also a nocturnal person so I do hobbies from 9pm-3am. Helps that I have a low stress low demand job. I work about 30 hours a week and can do that any time I want. Some weeks I’ll do a 2 on / 5 off schedule. Or work a lot for a month then take the summer off. I realized that’s a luxury most people don’t have.

I do the bulk purchase thing too on lots of stuff but still it seems like there’s always a few places we need to go on any given weekend.

Where do you live that you can take 20 car rides in a day? Dense suburb?

I have a hard time (for myself, not doubting you) imagining that many things to do.

I guess dense suburb.

I’m in the suburban core of Dallas Texas. The area that was originally developed in 1950s as suburbs. It’s very much considered “in town” now as the area has sprawled greatly since. But it’s still very neighborhood-like. Most of my trips are 10 minutes or less from my house of total drive time.

Before having a kid I never understood the busy parenting thing, but I’m actually finding myself want to be at all the things and it’s a huge social thing that nothing since college/high school compares too. Meeting new friends as an adult is actually very easy if you have kids and want to be involved in things.

Does it matter if you "own" your car if it is part of the Waymo fleet, getting regular updates and communicating with the Waymo network like any other robotaxi in the fleet? You could even opt into having it give people rides while you're working or sleeping. And then Waymo can have you pay for the maintenance, insurance, and cleaning. It's like having Uber drivers who don't have to drive!

I think the key issue is that "owning" a car comes with a lot of expectations on behalf of existing customers.

People are going to say that if they own it they should be able to mod it as they please, is that actually practical with a self driving car? What if they mod it in some way that makes it dangerous? Who's liable?

There's a whole host of issues like this and most of them go away completely if you can't buy a self driving car but you can rent it per ride. Plus, if they do run it as a taxi service I think they can iterate everything from hardware, software and even infrastructure that supports it much faster than if they're selling it.

I think google will be content to run a taxi service and let others sell self driving cars if and when they catch up because honestly it's a better business than outright selling cars.

I doubt it. The US is too spread out and large to have a centralized robo taxi system work for everyone. We will need individual vehicles.

Counterpoint: Hyundai/Kia has a terrible reputation in the car industry and was probably willing to give Waymo more favorable terms than any other manufacturer. Ford and GM are working on FSD by themselves, so they aren't interested.

Hyundai/Kia have had multiple scandals over the past 10 years due to widespread engine failures, fatal vehicle fires, and becoming a huge theft target due to failing to install even the most basic anti-theft technology (RFID key+immobilizer). Each of these has resulted in recalls and lawsuits.

And besides that, Consumer Reports ranks them #10 on the reliability chart.

https://www.consumerreports.org/cars/car-reliability-owner-s...

That link doesn't communicate what you seem to think it does. A rank of #10/11 puts their reliability in the upper third amongst major manufacturers, above every American auto manufacturer, as well as above every European car manufacturer except for BMW and Porsche.

Waymo works in specific areas only though. As soon as you get out of San Francisco, Phoenix, Los Angeles, or Austin, the car wouldn't have self-driving capabilities.

I suspect it'll be some kind of AI-powered assistant that will take self-driving vehicles to people, like [1]. You retrofit a device onto your car, which'll hook to the car's sensors (as well as bring it's own sensors, like camera) and get a better experience in self-driving. Still nowhere near reliable to just stop paying attention though.

Hopefully, they'll make reliable self-driving vehicles within the next 10 years or so...

[1] https://comma.ai/

Atlanta coming soon.

They scaled from 10k to 100k per week in a year.

They've been operating successfully on highways for months now.

Once they can do highways, a huge fraction of the US will open up. And generally speaking, driving outside of cities is quite a bit more easy.

They seem like they're moving very quicky to massive deployment. Comma is still just driver assist.

Custom vehicles that can strip out all the complexity of a steering wheel will reduce costs pretty aggressively.

Any taxi service will rely on several elements which limit their service area:

1. Availability. How many vehicles are in an area, that can respond to dispatch?

2. Demand. Can't have too many idle vehicles scooting around and taking up space.

3. Service. Must establish depots where the vehicles can return for charging, cleaning and maintenance.

4. Support. Waymo Support is a tap away, and if a vehicle gets particularly jammed up, they will dispatch a technician in their own vehicle. Support can also monitor telemetry and push commands to give nudges (but cannot remotely drive a car).

So with the current model, I do not envision Waymo service areas expanding beyond the reach of their depots, or offering long-haul trips between urban centers.

Even then... I happen to live in Phoenix, and locally wouldn't mind self-driving, but I like to drive on road trips as well. So electric first and limited self-driving as an option for my home area would be really nice, at least for me specifically.

Of course, I don't have enough trust in self-driving to even use the things. I know a lot of people love them, and have a couple acquaintances working on self-driving stuff. Just me being a bit hesitant with new tech in general.

I have similar feelings on a lot of the smart home stuff, well justified after companies dropping, deprecating hardware or outright dying. Maybe in another decade we might see better options all around.

Self-driving being affordable enough for people to own a self-driving car is going to rely on either massive price reductions on LiDAR sensors (an order of magnitude or 2) or a Tesla-style approach without LiDAR working out.

I worked in the space 5ish years ago, and the LiDAR sensors were monstrously expensive. They were the most expensive component we put in the car; if I'm remembering right, a single LiDAR sensor cost more than the car itself and there were half a dozen or so of them on the car. I want to say it was ballpark $500k to build out one of the cars, with the majority of that being LiDAR sensors followed by the servers in the car.

Comma doesn't appear terribly close to cracking that based on their features. It's certainly a neat project, but I wouldn't put them much closer to cracking self-driving than someone just starting out today.

Your numbers are way off. Waymo reported 5 years ago their Jaguars cost between $120k-$150k. That was after reducing their LiDAR cost by 90% from $75,000 to ~$7500. Their newly announced 6th gen sensors are even cheaper.

LiDAR sensors are no longer expensive. We are seeing automakers including sub $1000 LiDARs. Granted they only offer L2 ADAS, but the rapid price cratering means it's only a matter of time for consumer-owned self driving cars.

That's likely still too expensive; sub-$1,000 would probably work. The currently deployed Waymos have 5 $7,500 (at cost) LiDAR sensors in them. That's $37.5k, almost exactly $10k short of the average new car sale price.

To be clear, I'm talking about the point where normal people get new cars that have self-driving. It's affordable for wealthy people now, and was available to the super wealthy before that. I think the whole package is going to have to get somewhere in the $5k-$10k range before it's "normal" to have a self-driving car and be able to afford the maintenance.

Even those super cheap LiDARs may be over $10k installed by the time you add the other sensors, compute, labor, etc. I have no idea what the lifetime of a LiDAR sensor is, so maintenance is a concern too.

I think we'll get there, but it'll be a while after robotaxis become a thing. The news seems to think LiDAR sensors will settle around $200, which would definitely be affordable.

> The currently deployed Waymos have 5 $7,500 (at cost) LiDAR sensors in them.

Only the "dome" LiDAR at the top costs $7500. The smaller perimeter LiDARs are much cheaper.

> The currently deployed Waymos have 5 $7,500[...]

This price from 5 years ago. How confident are you the unit price is still $7,500?

> They were the most expensive component we put in the car; if I'm remembering right, a single LiDAR sensor cost more than the car itself and there were half a dozen or so of them on the car.

Some of that was Tesla/Elon propaganda when he was talking about being camera-only.

Comma? Geohit's scarce attention already flitted away, quite some time ago.

> Hyundai/Genesis has some of the best cars on the market right now between the GV line and Ionic lines

This also includes Kia, and the sibling to the Ioniq 5, the EV6

>Waymo is now the clear leader in self-driving

Always has been.

Why was this downvoted? Waymo has always been the clear leader in autonomous/self driving capabilities, at least for the true driverless goal. Briefly there were one or two milestones that Cruise got neck and neck on, but still think the parent’s sentiment holds true.

You could argue Tesla has a lead since it is more general/widely deployed, but IMO Waymo is definitely the better long-term approach.

Tesla is never going to ship fully automated driving IMo

The keys to Waymo's success and reliability are infrastructure and experience.

(Waymo is a subsidiary of Alphabet, Inc. That makes Google a sister company. Google doesn't operate Waymo.)

Google has built infrastructure and experience from the get-go, from driving Street View cars and building Maps, operating their own data centers, plenty of GPU application experience, teams of human support agents (yes they exist and respond), and just being really good at collecting, organizing, analyzing telemetry.

That's why I have confidence in the vehicles and Waymo Driver. This service is an iteration and evolutionary step. The Hive Mind behind the car has everything well in hand. Everyone else is just makin' cars and drivin' around alone. NO COMPARISON!

I can get into my Tesla, that I own and don't have to rent through an app, right now, and have FSD take me to the airport, or literally anywhere I want to go in the country. I don't have to be in one of San Francisco, Austin, Los Angeles, or whatever cities Waymo operates in. Tesla is leagues ahead, and to say Waymo is the "clear leader in autonomous driving" is a total joke. And this is coming from someone who regularly uses Waymo in SF. It's a cool tech demo but that's about it.

I'm not comparing Waymo to FSD because FSD has not shown the ability to be "truly diverless", aka Level 4+, which was an important distinction I made (perhaps not too clearly) in my comment.

FSD is probably the leading Level 2 autonomy system. As you point out it covers a massively broader set of geographies and conditions than Waymo can. On the flip side, it won't even take you around the block if you just get in the back seat and expect it to handle everything. Perhaps next week we will see major update from Tesla towards Level 4, but until FSD breaks past L2 supervised capabilities they are different products solving for massively different trade-offs.

> Hyundai/Genesis has some of the best cars on the market right now between the GV line and Ionic lines, and they seem to be improving faster than other manufacturers

Best in what ways? I don't follow the car market so I'm curious for more specifics here.

Warranty, charging rate (they pioneered the 800V architecture IIRC), bang-for-buck; Kia features are generous for the trim-levels compared to the competition.

Anecdotally I’ve enjoyed the two Hyundais I’ve had. They are ICE but had great safety/usability features for the price.

I’m not super into cars though so I don’t have much to compare to.

Their charge curve is second only to Porsche at the moment, and at a much lower price point. The Ioniq 5 is arguably the best EV crossover on the market.

Reply on news.ycombinator.com