Self driving cars are still decades away

IAN bob_2 and cannot speak for him, but the impression I got from his post was that he means it’s fun to show off the “self-parking” feature to a passenger when the parking space has very little margin of error.

This is about as science fiction as Star Dreck; meaning they’ll both be real in the same date, hundreds of years from now. The end game of this is that every moving vehicle (& pedestrian) is communicating to all of the others on the road. There will be no more traffic lights or stop signs at intersections as your vehicle will slow down or speed up to completely miss but pass within inches/feet of a crossing vehicle much like figure 8 car racing. The tech isn’t there yet, & when it is, it’ll eventually be required on new cars, & then at some point after that be required to be retrofitted on older cars & then some type of battery pack for cyclists & pedestrians.
You know some cyclist/pedestrian will either not charge it or totally forget to take it with them when they drive somewhere to start their ride/run. I’m sure the wire will come loose in some car at some point. & this says nothing of the wildlife that will never have one or a vehicle that breaks down & it turned off (there’s videos on YouTube of a vehicle losing a wheel & they’re stuck where they are), or someone losing something (mattress) from their roof or pickup bed, or of someone running out of fuel on a road w/o a shoulder/parking lane. IOW, even when we eventually have this great vehicle to vehicle communication, we’ll still need object avoidance technology.
We’ll also need some idiot-proof way to accommodate for the extra length when someone is towing a trailer; there’s a lot of length difference between a boat being towed & a bike on a hitch rack.

How will these cars handle parking at something like a festival or a tourism farm or a little league/soccer field where parking is in a big grass/dirt field & no lines to follow?

Well, no. A fully automated link was implemented in San Diego in 1997.

I was thinking of something simpler, like a transponder system. The car could ping objects in view to accurately determine their identity and status. The fleet would acquire such systems over time. Emergency vehicles, highway equipment, highway construction sites initially then phase in passenger cars, roadside facilities, bicycles, cats, dogs, etc…

Good luck getting the transponders on the squirrels; I tried and they are difficult to catch and then hold down long enough to install them. And then getting them to understand the walk/don’t walk lights; we’re still working on that.

Excellent point, but we can start with fire engines and save the squirrels for later.


A proof-of-concept test on one road is not an implemented system.

Are there any vehicles being sold today with this technology? How/when is it engaged? How when is it disengaged (if I want to exit the expressway at an earlier exit than the car ahead of me)?

How does it play with vehicles that don’t have this tech? IOW, only in the left lane of a limited access highway with only right side exits; otherwise one is going to start to drive past this train & then suddenly realize that they can’t break thru 30 cars 3’ apart to get to their exit, which would necessitate hard braking in the middle lane, which will now cause flow disruption issues for everyone else.

Police vehicles would need some way to temporarily disable their transponder in certain circumstances; otherwise, there goes all speedtrap revenue.

I think the point is that the technology for intervehicle communication exists, but there are many other huge roadblocks to implementing it. Other road occupants who are not using it is the biggest one.

From the Wiki link on “platooning” that the article points to:

In a fully smart car (whatever that means), I would program my destination, it would join a platoon and then exit both the platoon and the highway at the appropriate time.

They are not. As I pointed out above, that is the problem. It was assumed that V2V would be implemented long before self driving cars were even considered. The need will become obvious with time.

The 1997 platooning experiment investigated a possible remedy to the problem of traffic congestion. It worked very well.

In another 20 years traffic congestion will double compared to 1997. We are only half way there and places like the bay area are close to a standstill… Some things will have to change.

“We’re trying to have a single stack for both highway & city streets, but it requires massive NN (neural network) retraining.”

First, I didn’t know that they had separate stacks for street and highway, but I think that makes sense, or I guess it made sense.

Second, retraining both of those into one stack does indeed seem huge. Especially when the street stack has problems.

Waymo has begun a pilot program for customers to hail rides from their autonomous vehicles in San Francisco, the next city to be served after the Phoenix metro area.

However, there will still be a human safety driver behind the wheel, and not surprisingly they have initially geofenced the service region to exclude the downtown area, which is more challenging and has the most potential for ****ups to occur, even though that area is where most of the Uber/Lyft pickups take place.

Waymo claims they will expand this service region over time, although it remains to be seen how effectively they can deliver on their promise.

I suppose it was extremely quiet so he couldn’t hear it coming.

New information from the Tesla self driving crash outside Houston, in which the driver and passenger were reportedly in the back seat at the time of their death.

As reported by Bloomberg, the autopsy report shows that both people were intoxicated with a BAC well over the legal limit. Previous reports had been “a drink before dinner,” but this shows it was much more than that.

Proving that Tesla autopilot is not capable of being the designated driver, especially when the drunk person is the one issuing commands to the autopilot.

Could someone review the current information? As I remember it Tesla was saying there was no possible way Tesla was doing the driving [and the driver either moved into the rear seat or was thrown into the rear seat at the time of the crash].

Walmart has been rolling out “pilots“ of this since late 2019, but I haven’t seen any public results from them. But it must be working to some extent if they are extending it?

I mean, I’ve already shared my doubts (many times) about fully autonomous vehicles, but a geo-fenced usage makes perfect sense.