The robots are coming, the robots are coming
-
Despite my previous post about the reality checks slowing down self-driving initiatives, a company with far less name recognition than Tesla or Waymo is making noise with autonomous trucking.
Working with truck manufacturer Navistar as well as shipping giant UPS, TuSimple is already conducting test operations in Arizona and Texas, including depot-to-depot autonomous runs. These are being run under what’s known as “supervised autonomy,” in which somebody rides in the cab and is ready to take the wheel if needed. Sometime in 2021, the startup plans to begin doing away with human supervision, letting the trucks drive themselves from pickup to delivery without anybody on board.
By 2024,TuSimple plans to achieve Level 4 autonomy, meaning that its trucks will be able to operate without a human driver under limited conditions that may include time of day, weather, or premapped routes. At that point, TuSimple would start selling the trucks to fleet operators.
-
@just-jeepin who's accountable if it kills someone?
-
@sony1492 that’s the trillion dollar question
-
Example of the challenges automation faces: cattle grids.
-
@sony1492 said in The robots are coming, the robots are coming:
@just-jeepin who's accountable if it kills someone?
Contact your government representative and ask them. They're the ones that allow them to operate on public roads.
-
@just-jeepin its annoying that they rebuilt the road because our technology never took into account Non-urban infrastructure
-
Somewhere towards the end of that thread I had mentioned them and Plus(dot)ai. Both companies seem to be doing a slow, thorough, and quiet ramp up. To me this works as a better business case, but is not as helpful for funding unless you can get tied to a bigger company. In my experience it brings better results, but that sample size is small...
Navistar being forward thinking and proving out with development is an odd thing for me. When I was there, the R&D side was working on some really trick stuff as the Research side was really good, but C-suite just killed any Development constantly for quick money making BS that ended up in litigation or put trucks in endless warranty hell. Ustian and Campbell absolutely wrecked NAV and I guess they're finally crawling out of their hole.
Litigation-wise, I bet it gets spread between the kit provider, Nav, and the carrier. These systems pull enough data that fault isn't hard to suss out pretty quick. The bigger fish is going to be Nav or the carrier in this case. The trials for the Uber pedestrian death aren't all over I don't believe (maybe the safety driver is left?), but Uber settled with the family already I believe.
-
@looseonexit wait, what did Waymo have to do with the pedestrian death? That was Uber (and IIRC Volvo, indirectly, since their auto-braking software had been disabled).
-
Big mistake on my part. Not sure why I threw Waymo in there. Going to edit and remove.
-
@just-jeepin Trucks are a better target for automation than passenger cars.
-
The fleet managers may have fixed routes or a limited number of routes. These can be well mapped with additional information recorded in the navigation system.
-
In theory, fleet managers can do a better job of regular maintenance on the sensors.
-
It makes more sense to add $50,000 in sensors to a $300,000 truck if it saves you the cost of paying a driver. That's an easier financial calculation than adding $15,000 in sensors to a $40,000 car for a personal vehicle. This could allow more sophisticated sensors with more redundancy.
-
Large trucking companies already have liability policies in place for accidents by their drivers. They may be more willing to assume the risk of an incident by one of their self-driving trucks.
-
-
@roadkilled Yep. The model that I expect to win out is a pool of drivers at endpoints waiting for trucks to arrive, in order to finish the route/deal with warehouses & such.
-
@just-jeepin the robots are already here.
-
@frinesi2 I know, but in this week of revolutionary fervor it seemed apropos
-
@just-jeepin I always thought that depot-to-depot would be the best real-world scenario for unmanned vehicles. Also, I am surprised that someone hasn't come up with unmanned depot-to-depot with remote control abilities, where a centrally located "drone operator" can handle multiple vehicles in case there are issues which require manual intervention.
-
@musashi66 I think systems like that are either deployed, or nearly so, but I’m trying to remember where I read about it.
-
@just-jeepin I've heard nothing of remote operator capabilities, but I guess I would be surprised if the machines being tested don't have those capabilities.
-
@musashi66 said in The robots are coming, the robots are coming:
@just-jeepin I always thought that depot-to-depot would be the best real-world scenario for unmanned vehicles. Also, I am surprised that someone hasn't come up with unmanned depot-to-depot with remote control abilities, where a centrally located "drone operator" can handle multiple vehicles in case there are issues which require manual intervention.
@musashi66 said in The robots are coming, the robots are coming:
@just-jeepin I've heard nothing of remote operator capabilities, but I guess I would be surprised if the machines being tested don't have those capabilities.
I had mentioned that in the other thread.
@looseonexit said in Driverlessless:Maybe? I don’t think we’re too far off from remote or driverless tests for trucks. Currently the drivers are much like stewards (“safety operators”), sitting at the controls just in case. TuSimple w/UPS was running trucks for awhile along I10 with little fanfare in 2018. Plus.ai did a a Level 4 autonomous run across America delivery in late ‘19 with minimal stops - 41hr trip. Australia was looking at doing test with road trains at one point. Programs usually start with someone there in the seat for safety (1), remove the driver/operator and move to someone remote watching one (2), and then one person supervising multiple vehicles (3). Now we can remove cabs and any human elements for cost savings and efficiency. Next would be one vehicle supervisor over everything in a mine or one per in a state/locale in public domain for trucking) (4) with ideally vehicles not needing any supervision (5). Mines are well into step 3 for some equipment (mostly with adapted equipment, not all is purpose built to remove physical human interaction elements)... Trucking is looking towards step 2 probably.
It's coming, probably soon. It's already been developed and proven in mining for both remote single- and multi-drill and haul truck control, as well as fully autonomous drills and trucks. That's for humans watching remotely from the mine or a desk thousands of miles away including full intervention and control. I helped develop those systems/sub-system/vehicle integrations literally a decade ago now. There's still a lot of testing for full mine ops integration, but those systems as individual vehicle units have come a hell of a long way in 10 years. Fugro (offshore ROVs) has been doing work in this area for awhile (remote/fully autonomous), multi-vehicle control is next for them I'm sure. Depending on need having both multi-vehicle and fully autonomous is an extremely useful tool.
@Just-Jeepin
Here's another one to add to the list.
https://www.robothusiast.com/kodiaks-autonomous-trucks-reach-highway-milestone/ -
@looseonexit Cool! I always imagined self-driving Ubers with a remote drone operator to handle any issues. One "driver" could control dozens of cars and any issues could just go into a queue and be assigned to an available operator.
I think you are basically describing the same thing.
-
Sometime in 2021, the startup plans to begin doing away with human supervision, letting the trucks drive themselves from pickup to delivery without anybody on board.
By 2024,TuSimple plans to achieve Level 4 autonomy, meaning that its trucks will be able to operate without a human driver under limited conditions that may include time of day, weather, or premapped routes. At that point, TuSimple would start selling the trucks to fleet operators.
I really don't want to be around one of these things when (not if, WHEN) it encounters a situation that it can't handle.
-
@just-jeepin What I still dont' get is why self driving trains aren't a thing. 1 dimensional, weather independent.
-
@gmporschenut-also-a-fan-of-hondas I imagine the mile-long stopping distance plays a role; you want someone to be able to make that call based on incomplete information, since sensors on the train won't be all that effective over such a distance.
-
@just-jeepin I don't mean as a full train. A single railcar can carry 100tons vs max of 16 for a tractor trailer. If its automated just make dozens of little trains.
-
@gmporschenut-also-a-fan-of-hondas Excellent point. Dunno.
-
@just-jeepin Its like an arguement about drone warfare.
https://www.intelligent-aerospace.com/unmanned/article/14177174/us-armys-new-drone-swarm-may-be-a-weapon-of-mass-destructionIt isn't about replacing 1 driver with one robot, its replaceing all the drivers with more robots then they could ever compete against.
-
jminer
-
jminer
-
CarsOfFortLangley
-
jminer