D.Shepardson, July 7, “The National Highway Traffic Safety Administration has opened a special investigation into a recent crash of a Cruise self-driving vehicle in California that resulted in minor injuries, the agency said on Thursday.
The auto safety agency did not identify the specific crash, but a Cruise vehicle operating in driverless autonomous mode was involved in a crash involving minor injuries on June 3 in San Francisco, according to a report filed with the California Department of Motor Vehicles. … ” Read more Hmmmm… The police report indicates that the Cruise vehicle stopped while making a protected left turn, yielding to avoid being T-boned by a speeding Prius that might run its red. Instead the Prius changed to its left turn lane and broadsided the Cruise vehicle. I can’t wait to see the Cruise 360 video of that crash. Hopefully the Prius’ insurance company will reimburse the Federal Government for its expenses incurred in its special investigation of the crash that it caused. Alain
SmartDrivingCars ZoomCast Episode 274/ PodCast 274 w/ Anil Lewis of the National Federation of the Blind & Katherine Freund, founder ITN America
F. Fishkin, Juuly 9, “More MOVES progress in NJ, NHTSA investigates a Cruise crash, and two fatal Tesla crashes, Musk makes more headlines and layoffs at Argo AI. Plus from the 5th annual Smart Driving Cars Summit, Anil Lewis of the National Federation of the Blind and ITN America founder Katherine Freund. Join Princeton’s Alain Kornhauser and co-host Fred Fishkin for episode 274 of Smart Driving Cars.”
Technical support provided by: https://www.cartsmobility.com/
The SmartDrivingCars eLetter, Pod-Casts, Zoom-Casts and Zoom-inars are made possible in part by support from the Smart Transportation and Technology ETF, symbol MOTO. For more information: www.motoetf.com. Most funding is supplied by Princeton University’s Department of Operations Research & Financial Engineering and Princeton Autonomous Vehicle Engineering (PAVE) research laboratory as part of its research dissemination initiative
A. Hawkins, July, 7, “Two fatal Tesla crashes are being examined by investigators at the National Highway Traffic Safety Administration. Reuters reported that NHTSA opened a special investigation into a recent fatal crash in California, in which a Tesla driver killed a pedestrian. And an agency spokesperson confirmed to The Verge that a crash that took place on July 6th in Florida is also under examination.
The crash in Florida took place on Interstate 75, just south of Gainesville, where a Tesla vehicle smashed into the rear of a stationary tractor-trailer that was parked at a truck stop. Two people inside the Tesla, the driver and a passenger, were killed, according to Fox 35. A spokesperson for NHTSA said the agency was aware of the crash and was currently communicating with Tesla about it.
A spokesperson for the Florida Highway Patrol said it is not known yet whether Autopilot was activated at the time of the crash. …” Read more Hmmmm… Another crash into a stationary object in the lane ahead. I know that I keep repeating my self, but AV code is structured to to track objects by placing on each object a position vector and velocity vector; and even acceleration and jerk and … vectors.
The problem may well be in how these position vectors include data with respect to how reliably “can I pass underneath the object if I encounter it”.
At some point in the code, objects are classed into two buckets… 1. Don’t hit these, and 2. Don’t worry about these.
Approach velocity ( the difference between my car’s velocity and the velocity of the object) likely plays an important role in this classification.
For objects “in the lane ahead” one can reliably determine if the object ahead is moving or stationary.
If it is moving, then great.. follow it! If it makes it through, then I’ll also make it through, so just follow it and don’t rear-end it. (Not that simple, but you get the point)).
If it is stationary, oh my goodness, where did that come from? If it was moving, then surely I’ve been watching it come to rest and no problem, I’ve been slowing down too. If not, then surely I must be able to pass underneath it. It’s just an overpass, or a sign, or traffic light, or tree, or … that I encounter all the time above my road ahead. No problem! … I can easily pass under, so just disregard it!
If it is a parked truck… Yipes!!!
If it is moving at me, oh my goodness again; but I see these all the time. Especially on two lane roads that are turning to the right and there is moving traffic in the oncoming lane. Disregard these so you don’t start hitting the brakes, because surely we’ll somehow pass by each other safely. This one is really tough. If it ever happens that a Tesla, or anyone of these “AVs”, is involved in a head on collision then NHTSA better do a real deep dive into what the system did and didn’t do during the whole approach sequence. I suspect that transients of this situation are watched, but quickly disregarded. The question is.. the definition of transient and at what point is it taken seriously. Reliabilities and uncertainties are everything here.
I’m sure the “head-on situation” has been studied in simulation and may well have been studied in an actual demonstration. On a test track, one could send two Teslas towards each other with no one inside either and see what happens. They’ve probably done that. If so, please let us know what happened. If it hasn’t been done, then IIHS or NHTSA should do it. NHTSA and/or IIHS could readily take a Tesla with no one inside and hive it drive down one of its crash testing areas. Have they done that? They should. Should be easier to do than one of their normal crash test. More importantly, the Tesla should come to rest and no crash happened. Costs them nothing! They must have done that. Sorry to be so stupid here.
Also, I’ve not heard of a head-on collision of a Tesla vehicle traveling in its lane. When that one occurs, it will be interesting to see how “AutoPilot” handled an object in the lane ahead that was heading towards the Tesla. I’ve also not heard of a Tesla AutoPilot/FSD successfully avoiding a head-on. It would be great to learn about them, if they exist. Alain
R. Mitchel, July 6, “In the late 1980s, the U.S. Army turned to outside experts to study how pilots of Apache attack helicopters were responding to the torrent of information streaming into the cockpit on digital screens and analog displays. The verdict: not well.
The cognitive overload caused by all that information was degrading performance and raising the risk of crashes, the researchers determined. Pilots were forced to do too many things at once, with too many bells and whistles demanding their attention. Over the next decade, the Army overhauled its Apache fleet, redesigning cockpits to help operators maintain focus…. ” Read more Hmmmm…Just crazy!! Begin by removing the big screens from the front seat. Alain
A. Hawkins, July, 7, “An “erratic” pedestrian attacked a Waymo autonomous vehicle late Tuesday evening in Tempe, Arizona, smashing the windshield and injuring the safety driver, the company said. It was the latest incident of people in Arizona attacking Waymo vehicles — and occasionally their safety drivers — as the company ramps up its commercial service in the state. …” Read more Hmmmm… And we thought that no one misbehaved in Arizona. Alain
M. Harris, July 1, ” he most serious crash to date involving a self-driving truck might have resulted in only moderate injuries, but it exposed how unprepared local government and law enforcement are to deal with the new technology.
On May 5, a Class 8 Waymo Via truck operating in autonomous mode with a human safety operator behind the wheel was hauling a trailer northbound on Interstate 45 toward Dallas, Texas. At 3:11 p.m., just outside Ennis, the modified Peterbilt was traveling in the far right lane when a passing truck and trailer combo entered its lane.
The driver of the Waymo Via truck told police that the other semi truck continued to maneuver into the lane, forcing Waymo’s truck and trailer off the roadway. She was later taken to a hospital for injuries that Waymo described in its report to the National Highway Traffic Safety Administration as “moderate.” The other truck drove off without stopping…. ” Read more Hmmmm… If this is really the “worst automated truck crash” all this stuff is really good! It wasn’t even a crash. Slow news month so far. We’re even reporting good news. Alain
A. Hawkins, July 7, “Tesla plans to open up its Supercharger network to non-Tesla electric vehicles in the US in late 2022, according to a White House memo.
The company has been allowing non-Tesla EVs to use its Supercharger plugs in several cities in Europe as part of a limited pilot program but has been quiet about when US charging stations would be available to non-Tesla EV owners. A “fact sheet” published by the White House on June 28th and noticed by InsideEVs indicates that those EV owners may be able to use Superchargers as soon as the end of this year.
“Later this year, Tesla will begin production of new Supercharger equipment that will enable non-Tesla EV drivers in North America to use Tesla Superchargers,” the White House states. …” Read more Hmmmm… Another revenue opportunity for Elon. Alain
A. Hawkins, July, 7, “The first passenger station in the “Vegas Loop” network of vehicle tunnels that’s being built by Elon Musk’s Boring Company was revealed Thursday. The station is situated underneath Resorts World Las Vegas, the first in what is expected to be 55 stops along 29 miles of tunnels.
The Boring Company already operates a small version of this “Teslas in Tunnels” system underneath the Las Vegas Convention Center, which opened in early 2021 and involves two 0.8-mile tunnels. Afterward, Musk’s startup proposed a massive citywide expansion that was eventually approved by Clark County officials last year. The system uses human-controlled Model X and Y vehicles to transport passengers, despite Musk’s previous statements about using sleds to carry cars through the tunnels. …” Read more Hmmmm… More advances. Alain
Calendar of Upcoming Events